See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/269690577

Language Acquisition and Use in Multilingual Contexts. Theory and Practice

Book · January 2013

CITATIONS READS

2 5,606

2 authors:

Anna Flyman Mattsson Catrin Norrby

Lund University Stockholm University
9 PUBLICATIONS 66 CITATIONS 83 PUBLICATIONS 379 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Meet and Greet: Introducing self and other in intercultural communication View project

ELIAS (Early Language and Intercultural Acquisition Studies) View project

All content following this page was uploaded by Catrin Norrby on 25 February 2017.

The user has requested enhancement of the downloaded file.

Language Acquisition and Use in Multilingual Contexts

Flyman Mattsson, Anna; Norrby, Catrin

Published: 2013-01-01

Link to publication

Citation for published version (APA):

Flyman Mattsson, A., & Norrby, C. (Eds.) (2013). Language Acquisition and Use in Multilingual Contexts.
(Travaux de l'Institut de Linguistique de Lund; Vol. 52). [Publisher information missing].

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors
and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the
legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private LUNDUNI
VERSI
TY
study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal ? PO Box117
22100L und
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove +4646-2220000
access to the work immediately and investigate your claim.
TRAVAUX DE L’INSTITUT DE LINGUISTIQUE DE LUND 52

Language Acquisition and Use in

Multilingual Contexts
Theory and Practice

Anna Flyman Mattsson & Catrin Norrby (eds.)

© 2013 The authors

Travaux de l'Institut de linguistique de Lund 52

ISSN 0347-2558

ISBN 978-91-7473-625-0 (print)

ISBN 978-91-7473-626-7 (pdf)

Printed by Media-Tryck, Lund University

Lund 2013
Gisela Håkansson
65 years
21 July 2013
Tabula Gratulatoria
Malin Ågren Jan Einarsson Merle Horne
Lund Lund Lund
Gilbert Ambrazaitis Lena Ekberg & Alex Housen
Lund Christer Platzack Bryssel
Lund & Stockholm,
Annika Andersson Leena Huss
Lund
Lund Uppsala
Maria Eklund
Kristina Asker Kenneth Hyltenstam
Heinonen
Uppsala Stockholm
Uppsala
Monica Axelsson Enheten för nordiska Institutionen för
Stockholm svenska och
språk
flerspråkighet
Kristof Baten Vasa universitet
Stockholms
Gent Anna Flyman universitet
Camilla Bettoni Mattsson
Lise Iversen
Verona Lund
Kulbrandstad & Lars
Siv Björklund Qarin Franker Anders Kulbrandstad
Vasa Göteborg Elverum
Ulla Börestam, Anne Golden Sofie Johansson
Uppsala Oslo Kokkinakis
Göteborg
Sally Boyd Jonas Granfeldt
Göteborg Lund Victoria Johansson
Lund
Barbro Bruce Ann-Sofi Green
Malmö Lund Gunlög Josefsson
Lund
Lena Bruzaeus Marianne Gullberg
Staffanstorp Lund Marit Julien
Lund
Aafke Buyl Pia Gustafsson
Bryssel Lund Päivi Juvonen
Stockholm
Gunilla Byrman Björn Hammarberg &
Växjö Britta Hammarberg Marie Källkvist
Stockholm Lund
Anna Cabak Rédei
Lund Kristina Hansson Anastasia M Karlsson
Lund Lund
Gerd Carling
Lund Christina Hedman Satomi Kawaguchi
Stockholm Sydney
Ylva Carlsson
Stockholm Staffan Hellberg Kristin Kersten
Solna Hildesheim
Lars Olof Delsing
Lund Anne Holmen Jörg-U. Kessler
Köpenhamn Ludwigsburg
Bruno di Biase
Sydney Arthur Holmer Iwona Kowal
Lund Kraków
Verner Egerland
Lund
Lars Larm Carita Paradis Jan-Olof Svantesson
Lund Lund Lund
Hanna Lehti-Eklund Anders Philipsson Gudrun Svensson
Helsingfors Stockholm Växjö
Sara Lenninger Manfred Pienemann Jan Svensson
Lund Paderborn & Lund
Newcastle
Anke Lenzing Kari Tenfjord
Paderborn Henrik Rahm Bergen
Lund
Inger Lindberg Vi Thanh Son
Stockholm Anne-Charlotte Lund
Rendahl
Jan Lindström Anita Thomas
Stockholm
Helsingfors Lund
Andreas Rhode
Linnéuniversitetet Sofia Tingsell
Köln
Växjö & Kalmar Göteborg
Maria Rydell
Karen Lund Stockholm Karina Vamling
Köpenhamn Malmö
Marie Rydenvald
Kazuyo Lundström Joost van de Weijer
Göteborg
Lund Lund
Eva-Kristina Salameh
Benjamin Lyngfelt Ulla Veres
Malmö
Göteborg Göteborg
Susan Sayehli
Maisa Martin Lund Åke Viberg
Jyväskylä Uppsala
Susanne Schötz
Nisse, Mattis & Kaarlo Voionmaa
Lund
Malva Mattsson, Göteborg
Lund Halldór Ármann
Sigurðsson Birgitta Waters
Tim McNamara Växjö
Lund
Melbourne
Berit Söderman Åsa Wengelin
Jürgen M. Meisel Göteborg
Uppsala
Hamburg & Calgary
SOL-biblioteket Åsa Wikström
Martin Mirko Lunds universitet Lund
Stockholm
Göran Sonesson Elisabeth Zetterholm
Shidrokh Namei
Lund Växjö
Uppsala
Anja Steinlen Jordan Zlatev
Ulrika Nettelbladt
Nürnberg Lund
Lund
Michaël Stenberg
Catrin Norrby Lund
Stockholm
Sven Strömqvist
Anders Ohlsson
Lund
Lund
Gunlög Sundberg
Tore Otterup
Stockholm
Göteborg
Table of Contents
Anna Flyman Mattsson & Catrin Norrby
Introduction.................................................................................................... 9

Aafke Buyl & Alex Housen

Testing the applicability of PT to receptive grammar knowledge in
early immersion education
Theoretical considerations, methodological challenges and some
empirical results............................................................................................ 13

Jonas Granfeldt & Malin Ågren

Stages of processability and levels of proficiency in the Common
European Framework of Reference for Languages
The case of L3 French.................................................................................... 28

Marianne Gullberg
So you think gestures are compensatory?
Reflections based on child and adult learner data ....................................... 39

Arthur Holmer
Parallel grammars
On being separated by a common language ................................................ 50

Merle Horne
Acquisition of prosody
Word accents, phrasing, and morphosyntax in a Swedish-English
bilingual child at 30–32 months of age ......................................................... 60

Victoria Johansson
Writing – it’s like learning a new language! ................................................ 70

Marie Källkvist
Bilingualism in the university classroom and student engagement
in deep learning approaches........................................................................ 80

Satomi Kawaguchi
The relationship between lexical and syntactic development in
English as a second language ....................................................................... 92
Kristin Kersten & Andreas Rohde
Teaching English to young learners ........................................................... 107

Inger Lindberg & Kenneth Hyltenstam

Flerspråkiga elevers språkutbildning ......................................................... 122

Manfred Pienemann, Jörg-U. Keßler & Anke Lenzing

Developmentally Moderated Transfer and the role of the L2 in L3
acquisition .................................................................................................. 142

Eva-Kristina Salameh & Ulrika Nettelbladt

Learning in two languages
Consequences for lexical development in Swedish and Arabic .................. 160

Anja K. Steinlen
“Flera språk – fler möjligheter” – Immigrant children’s acquisition
of English in bilingual preschools............................................................... 170

Jan-Olof Svantesson
Mongolisk skrift, mongoliska staten och mongolisk nationalism ............ 185

Joost van de Weijer

The Obligatory Contour Principle in Swedish............................................ 194

Elisabeth Zetterholm
Svenska med en touch av Aussie ............................................................... 203
Introduction

Anna Flyman Mattsson

Catrin Norrby

This volume brings together current linguistic research in a range of pre-

dominantly multilingual contexts. The authors draw on data from different
languages and speech communities around the world, and together the
chapters offer a broad picture of language acquisition, development and use
– among both children and adults. While this is a valuable undertaking in its
own right, the main reason for this collection is to pay tribute to Gisela
Håkansson on her 65th birthday. Throughout her career Gisela Håkansson
has made, and continues to make, a very significant contribution to the field
of linguistics, in particular through her research in first and second language
acquisition, multilingualism and language impairment. The authors and
editors of this volume have all collaborated with Gisela in one way or other:
as partners in international and national research projects, or as colleagues,
particularly in the Centre for Languages and Literature at Lund University,
where she is professor of linguistics. Through her research, inspirational
teaching and public engagements in the wider community, she is also a
mentor and friend. Many colleagues, former students and friends, in Sweden
and abroad, wish to congratulate Gisela on her birthday, which the tabula
gratulatoria bears witness to. The 16 peer-reviewed chapters, organised
alphabetically, are briefly introduced below.
The volume opens with a contribution by Aafke Buyl and Alex Housen
which explores the applicability of Processability Theory (PT) to L2 recep-
tive grammar acquisition. The analysis of comprehension data from an
immersion school in Brussels, where francophone children learn English as
L2, suggests that L2 receptive grammar acquisition is governed by the same
processing procedures as productive grammar acquisition, with results over-
all in line with the general predictions of PT.
PT also provides the theoretical underpinning for the chapter by Jonas
Granfeldt and Malin Ågren in which they investigate the relationship bet-
ween second language development, as outlined in PT, and second language
proficiency as measured by the CEFR test, based on written L3 French data
produced by Swedish secondary school students. The results show a corre-
lation between CEFR ratings and the PT analysis, particularly at lower

9
Language Acquisition and Use in Multilingual Contexts

levels, where communicative proficiency and morphosyntactic development

largely go hand in hand.
Marianne Gullberg’s chapter examines the role of gestures in child and
adult learner data and poses the question whether gestures are compensatory.
The main findings demonstrate that gestures do not replace, but typically co-
occur with speech to form an integrated system. Nevertheless, there are situ-
ations when gestures are used as a compensatory device to solve interactive
or grammatical difficulties, indicating that gestures can be compensatory, but
not without qualifications.
The contribution by Arthur Holmer discusses the relationship between
input and output in first language acquisition, arguing that while members of
a certain speech community can produce identical structures, their internal
grammars can radically differ. Such parallel grammars could be in stable
balance, as is the case in Swedish, but could also be unstable, opening up for
language change, as illustrated in the chapter by the Austronesian language
Tgdaya Seediq spoken in Taiwan.
Acquisition of prosody in a simultaneously bilingual boy at 30–32
months, exposed to both Swedish and English in the home, is the topic of
Merle Horne’s contribution. The results show that the boy is acquiring in-
flectional morphology and the morpho-phonological rules for associating
word accents in Swedish with different grammatical affixes. While the data
provide insights into to the acquisition of the prosody-morphology mapping,
the author calls for more comprehensive longitudinal studies in order to
better understand this relationship.
In Victoria Johansson’s chapter we turn to the relation between speech
and writing, and in particular how young learners develop their writing skills
over time. Based on both spoken and written tasks, the results demonstrate
that the youngest participants (10-year-olds) rely on linguistic and pragmatic
features typical of spoken language when writing, and that becoming a
competent writer takes time and effort.
The renewed interest in translation tasks in L2 learning contexts, in part-
icular from the perspective of deep approaches to learning, is the topic of
Marie Källkvist’s chapter. Her longitudinal study of the effects of translation
on L2 grammar development in three Swedish EFL university classrooms
indicates that carefully designed translation tasks can be a very useful learn-
ing tool in classrooms where all participants share the same L1.
With Satomi Kawaguchi’s chapter we return to the framework of PT and,
in particular, the Lexical Mapping Hypothesis developed within PT. Twenty-
two Japanese L1 speakers with English as L2 participated in a vocabulary
size test and a translation task into English. The results demonstrate that can-
onical mapping precedes non-canonical mapping, and that successful non-
canonical mapping is characteristic of advanced syntactic development. In
terms of lexical size, the results suggest that only learners with a large voca-
bulary can handle non-canonical mappings without problems.

10
 Language Acquisition and Use in Multilingual Contexts

Kristin Kersten and Andreas Rohde address the question of early foreign
language learning in the context of European pre- and primary schools. They
draw on findings from their research on bilingual kindergartens as part of the
EU-funded ELIAS project (see also the chapter by Anja K. Steinlen). In add-
ition, they also discuss two specific primary school programmes in
Germany: early start and CLIL, where a substantial number of content
subjects are taught in the target language. The chapter concludes that an
early introduction of the L2 is beneficial, as long as the teaching methods
focus on communicative content and meaningful interaction.
With the chapter by Inger Lindberg and Kenneth Hyltenstam the focus
shifts to policy issues, and how the Swedish school system manages to look
after the needs of multilingual students – here referring to students with an
L1 other than Swedish, or an additional L1 other than Swedish. The authors
critically examine the development of the subject Swedish as a second
language, and point to shortcomings in its implementation, leading to its
overall low status. They present a number of suggestions for enhancing the
language education outcomes for multilingual students of varying back-
grounds and proficiency levels.
The chapter by Manfred Pienemann, Jörg-U. Keßler and Anke Lenzing
contributes to the ongoing debate in SLA research about the role of transfer.
The authors examine recent research findings which claim that L2 transfer
accounts for the structural outcome in the L3. Based on a critical review of
this research, and on their own study of the acquisition of Swedish as L3 by
German L1 speakers with varying L2s, the authors conclude that learners
only transfer structures – from L1 or L2 – when they are developmentally
ready, lending support to the Developmentally Moderated Transfer Hypo-
thesis developed within the framework of PT.
Eva-Kristina Salameh and Ulrika Nettelbladt’s chapter examines lexical
development in bilingual children. Forming part of a project on bilingual
education for Swedish-Arabic bilingual children, they investigated the size
and organisation of the children’s lexicon by means of a word association
test. Compared to a control group of Swedish-Arabic bilinguals educated in
Swedish only, the bilingual group’s lexicon was more hierarchically orga-
nised with greater use of paradigmatic associations. This, the authors argue,
underscores the importance of offering bilingual children education in both
their languages to promote their successful linguistic development.
Bilingual education is also the focus in Anja K. Steinlen’s chapter, which
discusses the success of early English immersion programmes in German,
Belgian and Swedish bilingual preschools. Her research, forming part of the
ELIAS project (see also Kersten & Rohde, this volume), showed no signi-
ficant differences in the receptive English grammar and vocabulary know-
ledge between children of immigrant and non-immigrant backgrounds. This
result contrasts sharply with earlier claims that children of other L1s than the

11
Language Acquisition and Use in Multilingual Contexts

majority language are disadvantaged by the early introduction of another

additional language.
Jan-Olof Svantesson charts the history of different writing systems deve-
loped for Mongolian, starting with the introduction of the Uighur Mongolian
script in the 1200s, used until the adoption of the Cyrillic alphabet in the
1940s. New writing systems in Mongolia have been initiated almost exclus-
ively by the state; however, nationalism was the driving force behind the
unsuccessful attempt to replace the Cyrillic script by reinstating the old
Uighur Mongolian script in the 1990s. The author argues that its failure is
explained by increased literacy among the population, where people have
learnt and use the Cyrillic script, while the old script is completely unknown
to a large majority.
Constraints on how consonants and vowels combine – the phonotactic
rules of a language – is the topic of Joost van de Weijer’s contribution. It
focuses on the avoidance of identical segments, the so-called obligatory
contour principle, known to occur in many languages. Through an empirical
analysis of Swedish phonological structure, the author shows that repetition
of identical consonants in the same word is often significantly lower than
their overall frequency would suggest.
In the final chapter, by Elisabeth Zetterholm, we return to language edu-
cation in a university classroom context. The chapter discusses the acqui-
sition of Swedish pronunciation by speakers of Australian English, who were
either enrolled at the University of Melbourne or on exchange to Lund
University. The results indicate that all learners, irrespective of the learning
environment, displayed pronunciation features consonant with earlier find-
ings for English background speakers learning Swedish. However, the
students learning Swedish in Sweden showed greater fluency and were cap-
able of holding a spontaneous conversation with the researcher in Swedish.

12
Testing the applicability of PT to
receptive grammar knowledge
in early immersion education
Theoretical considerations, methodological challenges and some
empirical results

Aafke Buyl
Alex Housen

Introduction
Since the morpheme studies in the 1970s (Dulay & Burt 1974; Larsen-
Freeman 1975), the notion of universal developmental stages in L2 grammar
acquisition has been much investigated in SLA research, and has, thanks to a
large body of empirical evidence, found widespread (though not universal)
acceptance (Ellis 2008; Ortega 2009). An explanatory account of staged and
predictable L2 grammar acquisition is offered by Processability Theory (PT)
(Pienemann 1998, 2005). PT is more comprehensive than many other
accounts of L2 grammar development in that it makes clear and falsifiable
predictions (Jordan 2004), accounts for a wide range of grammatical
phenomena, and is psychologically plausible for a wide range of
typologically diverse L1s. What has remained largely unexplored, however,
is whether PT applies to receptive as well as productive grammar
acquisition. Although Pienemann (2007) has claimed that “at any stage in
the development the learner can produce and comprehend only those L2
linguistic forms which the current state of the language processor can
handle” (p.137, emphasis ours), empirical data for the theory have come
almost exclusively from production data. This study presents one of the first
attempts to investigate the applicability of PT to receptive grammar
acquisition.
The empirical data presented in this paper were partly collected within the
framework of the ELIAS (Early Language and Intercultural Acquisition
Studies) project (Kersten et al. 2010), which investigated L1, L2 and
intercultural learning in nine immersion (pre)schools in Germany, Belgium
and Sweden between 2008 and 2010.

13
Language Acquisition and Use in Multilingual Contexts

The present paper bears a close relationship to Gisela Håkansson’s work

both in its use of PT as a theoretical framework (Håkansson 1997, 2001;
Håkansson et al. 2002) and in its use of data from the ELIAS project, in
which Gisela Håkansson was coordinator of the Swedish study. Previous
exploratory attempts to investigate the applicability of PT to receptive
grammar acquisition have been undertaken by Gisela Håkansson (e.g. 2012),
using the Swedish data she collected for the ELIAS project.

Principles of PT
Processability Theory is a theory of L2 grammar acquisition designed to
explain the staged development in L2 learners’ grammar knowledge. Based
on Levelt’s (1989) Model of Speech Production, PT assumes that in the
course of the production process, grammatical information is stored in
memory and retrieved at later points in the language generation processing to
unify grammatical information between constituents (e.g. number agreement
between the subject and the verb). This exchange of information, or feature
unification, it is further assumed, is executed in the course of the language
production process by five hierarchically-ordered ‘processing procedures’:
1. the lemma procedure, which accesses lexical entries
2. the category procedure, which accesses the categorical information of
the lexical entries
3. the phrasal procedure, which unifies information within phrases (e.g.
between determiner and noun)
4. the S-procedure, which unifies information between phrases (e.g.
subject-verb agreement) and
5. the subclause procedure, which builds subclauses.

PT predicts that L2 learners acquire the processing procedures necessary for

the exchange of grammatical information in the same hierarchical order as
they are executed in production. Because all L2 learners develop processing
procedures in the same order, the grammatical phenomena that require one
of these processing procedures become available to all L2 learners in the
same order.
PT’s processing procedures are claimed to be universal across languages.
However, since each language has its own grammar with its own word order
rules and morphology, developmental schedules (i.e. which grammatical
structures arise at each stage) are language-specific. For the development of
L2 grammar in English, PT, in its most recent version, (Pienemann 2005;
Pienemann & Keßler 2011) distinguishes six developmental stages (Table
1).

14
 Language Acquisition and Use in Multilingual Contexts

Table 1. Developmental Stages in English.

Stage Processing Procedure Example(s)
6 subclause-procedure ‘Cancel Inversion’ (I wonder where he is)
5 S-procedure subject-verb agreement (he sees you)
4 VP–procedure(*) tense agreement (I have seen you)
3 phrasal procedure plural-agreement (two cats)
2 category procedure plural –s (cats)
1 word/lemma single words; formulae
(*)Verb Phrase procedure, a procedure specific to English which deals with the
exchange of grammatical information within verb phrases

Theoretical considerations
Studying receptive L2 grammar acquisition within a PT framework requires
a consideration of the theoretical plausibility of the applicability of PT to
receptive grammar acquisition. As Ellis (2008) points out, “PT is in actuality
a theory of language production” (461, emphasis ours). A crucial theoretical
consideration therefore is that, in order for PT to be applicable to language
comprehension, the principles of Levelt’s model that underlie PT should be
psychologically plausible for L2 grammar comprehension as well.
In more exact terms, applying PT to comprehension assumes (a) that
language comprehension involves the same processing procedures as
language production and (b) that these processing procedures also steer the
development of learners’ receptive L2 grammar processing abilities. Neither
of these issues can be resolved by looking at the available psycholinguistic
(or L2 acquisition) research. Psycholinguistics has not yet reached consensus
on the exact nature of the language comprehension process (Fernández &
Cairns 2011; Garman 1990; Van Gompel & Pickering 2007). A relevant
finding from psycholinguistic research is that agreement processing is a
psychologically real process in language comprehension (Pearlmutter et al.
1999) but whether this means that feature unification also plays as central a
role in (L2) grammar comprehension and L2 grammar development as it is
claimed by Levelt’s model and PT, cannot be resolved at this stage.
The unresolved nature of receptive language processing not only justifies
an investigation of the applicability of PT to receptive grammar acquisition,
it also forms a further rationale for this study: an empirical PT study on
receptive grammar acquisition may inform us on the similarities between
receptive and productive processing, and thus have implications beyond the
field of L2 acquisition research.

15
Language Acquisition and Use in Multilingual Contexts

Methodological challenges
Studying receptive grammar acquisition within a PT framework also poses
major methodological challenges. A first challenge, common to all studies of
receptive L2 grammar knowledge, relates to the definition and assessment of
receptive grammar knowledge. Receptive grammar knowledge can be
defined as the ability to process receptive grammar knowledge for
comprehension. Since the comprehension of utterances, in contrast to their
production, does not always require grammatical parsing, but can sometimes
rely on semantic cues (Bates et al. 1984; Boland 1997; Garman 1990; Van
Gompel & Pickering 2007), such as event probability (Jurafsky 1996) or
noun animacy (McDonald 1987), it must be ensured that semantic
knowledge does not form a confounding factor in the assessment of
receptive grammar processing. Furthermore, because PT is built around the
mechanism of feature unification (or agreement), the assessment of learners’
receptive processing of the different types of feature unification defined by
PT forms an additional methodological prerequisite and challenge.
An additional and specific methodological challenge for comprehension
studies using PT rather than other models is the concept of ‘emergence’.
According to PT, the acquisition of a processing procedure does not (have
to) result in native-like production. Rather, PT claims that when a processing
procedure is acquired, the grammatical phenomena that require this
processing procedure will emerge in the learners’ production, i.e. they are
used productively and systematically (as opposed to merely as memorized
chunks), though not necessarily correctly in all obligatory contexts.
Emergence has been operationalized for spontaneous production only,
namely as first, systematic use. Previous PT studies that did not use
spontaneous production data came up with ad-hoc solutions such as several
acquisition criteria (Glahn et al. 2001) or group accuracy rates (Baten 2011),
but a widely applicable operationalization of emergence for non-spontaneous
production and comprehension is not yet available. This study will therefore
adopt its own ad hoc solution to the problem (see below, Design and
Methodology: Emergence).

Research questions
The research questions that guided our study are:
a) Is there a universal developmental order in the receptive acquisition of L2
grammar?
b) If so, is this pattern in line with the predictions made by PT?

16
 Language Acquisition and Use in Multilingual Contexts

Design and methodology

Participants
The participants are francophone children learning English as an L2 in an
immersion school in Wallonia, Brussels. The school uses English as the
medium of instruction for 50% of the curriculum (±14 hours/week), starting
in the third year of kindergarten (age 5–6) and continuing throughout the six
years of primary school (age 6–12). All participants spoke French as an L1.
About 25% of them used an additional language at home. None of the pupils
had had any direct or sustained contact with English prior to entering the
English immersion programme around the age of five.
For the longitudinal study, thirteen pupils were tested four times over the
course of 3.5 years, i.e. in kindergarten (T1) and in the first (T2), second
(T3) and third (T4) year of primary school. The time spent in the immersion
programme (incl. holidays) was 7 months at T1, 18 months at T2, 30 months
at T3 and 37 months at T4.
The cross-sectional study comprises three age/proficiency cohorts,
consisting of 72 pupils from the first (n=28; including the 13 learners
participating in the longitudinal study), second (n=23) and third (n=21) year
of primary school. All three cohorts were tested once (coinciding with T2 in
the longitudinal study). Time spent in the immersion programme at the time
of testing was 18, 29 and 41 months for primary 1, 2 and 3 respectively.

Target phenomena, instrument and procedure

The present study investigates the development of six grammatical
phenomena. Four of these are predicted to emerge at stage 2 of the
Processability Hierarchy for English: plural marking on lemmas (PLU; e.g.
cats), canonical word order (SVO; e.g. The boy is kissing the girl), genitive
‘s (GEN; e.g. The girl is feeding the boy’s dog) and negation expressed by
the negator not (NEG; e.g. The duck is not eating). The two remaining
phenomena involve subject-verb agreement, either with main verbs (AGRv;
e.g. The sheep eats) or with the copula be (AGRc; e.g. The deer is white) and
hence are predicted to emerge at stage 5. Learners’ knowledge of the six
phenomena was tested with the ELIAS Grammar Test (GT) (Steinlen et al.
2010), a picture selection task developed for testing L2 grammar develop-
ment of the immersion learners in the ELIAS project (Kersten et al. 2010).
There are six prompts for each phenomenon. Each prompt was orally
presented to the learner simultaneously with a set of three response pictures
(one correct, one incorrect, one distractor). Learners responded by pointing
to the picture they felt matched the prompt. All learners were tested
individually by the same researcher, who also provided the prompts.

17
Language Acquisition and Use in Multilingual Contexts

Implicational scaling
The data were analysed by means of implicational scaling, the preferred
method of analysis within PT research (Pienemann 2005). Of interest in the
present study is whether the scaling analysis (see Hatch & Lazaraton 1991
for more information on the procedure) shows that the stage 2 phenomena
(GEN, NEG, SVO, PLU) are acquired before the stage 5 phenomena
(AGRv, AGRc), consistently across individuals.
Within stage 2 and stage 5, individual variation is allowed. A scalability
coefficient, which measures the degree to which learners’ acquisition orders
are consistent with the overall acquisition order determined by the scalability
– i.e. including the order within the PT stages – is therefore not of much
relevance for the present analysis. Nonetheless, to give the reader a notion of
the scalability, we will present the coefficient of reproducibility (CR; see
Hatch & Lazaraton 1991 for more complex measures of scalability).
Following Rickford (2002), scalability is accepted when the coefficient of
reproducibility is .93 or higher.

Emergence
In implicational scaling, an acquisition criterion is set to determine whether
an individual participant’s score for a grammatical phenomenon indicates
that the phenomenon is acquired or not. As explained earlier, PT uses an
emergence criterion that looks at the first, systematic use. A challenge for
language comprehension research, then, is the development of a psycho-
linguistically plausible operationalization of the emergence criterion for
language comprehension, since checking for first systematic and productive
use is not possible here.
In the ELIAS GT, the seven scores that can be obtained for each
grammatical phenomenon (i.e. ranging from 0 out of 6 to 6 out of 6 correct)
yield six possible emergence/acquisition criteria: a ≥1/6, ≥2/6, ≥3/6, ≥4/6,
≥5/6 and 6/6. For example, with a ≥3/6 criterion, all scores of 3 out of 6
correct, or better, are considered emerged or acquired. Because PT, for
production, does not consider a grammatical structure as acquired when only
one or two instances of correct use can be found (Keβler & Liebner 2011),
the ≥1/6 and ≥2/6 criteria will not be used in this study either. Since any
remaining acquisition criterion involves an arbitrary choice, we use all four
remaining criteria: ≥3/6, ≥4/6, ≥5/6 and 6/6.
An important statistical consideration in any multiple choice task is that
certain scores will be below chance performance, meaning that the
probability that a participant could attain the score by guessing is too high
(i.e. >.05) (Howell 2010). In the ELIAS GT we can be certain that a
participant was not guessing only when s/he obtained a score of 5 or 6 out of

18
 Language Acquisition and Use in Multilingual Contexts

6 (Table 2). We will nonetheless include the ≥3/6 and ≥4/6 scale in the
analysis as they may still yield relevant information concerning general
developmental tendencies among the learners.
Table 2. Chance Performance (* = score is above chance performance).
N number of prompts 6
π chance level 0.33
k number of correct responses 3 4 5 6
P≥k probability of obtaining .31 .10 .02* .00*
score k or higher by chance

Results

Cross-sectional study
In the implicational scales in tables 3 to 6 at the end of this section (showing
acquisition criteria ≥3/6, ≥4/6, ≥5/6 and 6/6 respectively), a ‘+’ in a column
indicates that the participant(s) passed the acquisition criterion for the
grammatical phenomenon in question, while a ‘-’ signals that this was not
the case. The scales are contracted, meaning that participants who had the
same pattern of acquired and non-acquired items are combined into one row.
The columns headed by ‘#ID’ give information about the number of
participants that showed a particular pattern of acquired and non-acquired
items. Deviations from the pattern predicted by PT are between brackets.
The total number of acquired grammatical phenomena decreases as the
emergence/acquisition criterion is raised. In the ≥3/6 scale (Table 3), for
example, 25 of the 72 participants have passed the acquisition criterion for
all grammatical phenomena. This drops to zero participants in the 6/6 scale
(Table 6). Thus, the statistical effect of chance performance is not so strong
as to distort the general picture emerging from these scales.
The ≥3/6 and ≥4/6 scales both show an overall rank order (top row) that is
in accordance with PT: AGRc and AGRv are ranked last (based on the
number of participants that have acquired the grammatical phenomena),
meaning that overall more participants can process category information
than inter-phrasal agreement information. Violations of the PT rank order
can be observed at the level of the individual rank order in all three of these
scales, e.g. 18 participants show a pattern of pluses and minuses that is not in
line with PT in the ≥3/6 and ≥4/6 scale. As mentioned earlier, however, we
cannot draw any valid conclusions from these patterns as we cannot rule out
that this individual variability is influenced by participants’ guessing.
In the ≥5/6 and 6/6 scales, too, the observed rank orders are in line with
the rank order predicted by PT, since AGRc and AGRv are ranked last. At a

19
Language Acquisition and Use in Multilingual Contexts

more detailed level, it can be observed that violations of the PT rank order at
the level of the individual participant, though present, are fairly small. In the
≥5/6 scale, only 4 of the 72 participants show a rank order not in accordance
with PT: one learner passed the 5/6 threshold for AGRc without yet having
done so for GEN and PLU, a second participant had acquired AGRv while
not yet having acquired GEN and PLU, a third participant had acquired
AGRv while not yet having acquired SVO, GEN and PLU and finally one
participant had acquired AGRc while not yet having acquired NEG, GEN
and PLU.
Finally, there are no instances of participants having acquired AGRc
and/or AGRv without having acquired any of the stage 2 phenomena,
indicating that the emergence of stage 5 phenomena never precedes the
emergence of stage 2 phenomena altogether. The observed deviations could
furthermore be deemed irrelevant, because of PT’s claim (or disclaimer) that
the theory “does not predict that whatever can be processed will indeed be
acquired; Instead the theory predicts that what cannot be processed will not
be acquired” (Pienemann, 2005: 40). Thus, the emergence of at least some
stage 2 phenomena before the emergence of phenomena from later stages
may be sufficient support for PT. Admittedly, this ad hoc solution com-
promises the falsifiability of PT when taken to extremes. More decisive
perhaps are the CRs. The CR is acceptable for the 6/6 scale (CR .93) and
borderline acceptable for the ≥5/6 scale (CR .92). Given that the (acceptable)
within-stage variation is included in the calculation of this coefficient, both
coefficients are in favour of a universal acquisition order in line with PT.
One reflection that cautions against too readily dismissing the deviations
from the PT pattern in this study is that no stage 3 or 4 phenomena were
included in the analysis. Thus the question arises as to whether the data
would still support PT if this had been the case.

20
 Language Acquisition and Use in Multilingual Contexts

Table 3. Implicational scale - acquisition criterion: ≥3/6 correct (CR .86).

#ID NEG GEN SVO PLU AGRv AGRc
25 + + + + + +
11 + + + + - +
14 + + + + + -
(2) + + + (-) (+) (+)
(2) + (-) + + (+) (+)
(1) + + (-) + (+) (+)
2 + + + + - -
(3) + + + (-) - (+)
(3) + + + (-) (+) -
(1) + + (-) + - (+)
(1) + (-) (-) + (+) (+)
1 + + + + - -
(1) + (-) + + - (+)
1 + + + - - -
(1) + (-) (-) + - (+)
(2) + + (-) (-) (+) -
(1) (-) + (-) (-) - (+)

Table 4. Implicational scale - acquisition criterion: ≥4/6 correct (CR .84).

#ID NEG GEN SVO PLU AGRv AGRc
2 + + + + + +
10 + + + + + -
8 + + + + - +
6 + + + + - -
(7) + + + (-) (+) -
(3) + + + (-) - (+)
(1) + + (-) + - (+)
13 + + + - - -
(2) + (-) (-) + (+) -
3 + + - + - -
(3) + (-) + (-) - (+)
1 + - + + - -
4 + + - - - -
1 - + - + - -
1 + - - + - -
(1) - (-) (-) + (+) -
4 + - - - - -
1 - - - + - -
1 - - - - - -

21
Language Acquisition and Use in Multilingual Contexts

Table 5. Implicational scale - acquisition criterion: ≥5/6 correct (CR .92).

#ID NEG SVO GEN PLU AGRc AGRv
2 + + + + + -
9 + + + + - -
16 + + + - - -
3 + + - + - -
(1) + + (-) (-) (+) -
(1) + + (-) (-) - (+)
1 - + + + - -
14 + + - - - -
2 + - + - - -
1 + - - + - -
(1) + (-) (-) (-) - (+)
2 - + + - - -
1 - + - + - -
(1) (-) + (-) (-) (+) -
10 + - - - - -
1 - + - - - -
2 - - + - - -
4 - - - - - -

Table 6. Implicational scale - acquisition criterion: 6/6 correct (CR .93).

#ID NEG SVO GEN PLU AGRc AGRv
12 + + + - - -
2 + + - + - -
1 - + + + - -
13 + + - - - -
2 - + - + - -
1 - - + + - -
17 + - - - - -
8 - - - - - -
1 - - + - - -
15 - - - - - -

Longitudinal study
For the presentation of the longitudinal data, which in the traditional format
would consist of 13 tables of implicational scales, an alternative, graphic
way of visualizing and summarizing the results was designed, based on
Baten (2011). In the set of six squares below (Figure 1), each square re-
presents one of the six targeted grammatical phenomena, with the top row
representing the stage 2 features and the bottom row representing the stage 5
features.

22
 Language Acquisition and Use in Multilingual Contexts

Figure 1. Graphic presentation of longitudinal implicational scales (adapted from

Baten 2011).

When a grammatical phenomenon is acquired by a learner (according to the

acquisition criterion used), the square representing the grammatical
phenomenon is shaded in black.
The figures at the end of this section represent the longitudinal develop-
ment of the thirteen participants (rows headed by ID1 to ID13) over the four
test times (T1, T2, T3 and T4), using the acquisition criteria ≥3/6 (Figure 2),
≥4/6 (Figure 3), ≥5/6 (Figure 4) and ≥6/6 (Figure 5).
In general, the results show that all learners show gradual progress from
T1 to T4. This progress is not always linear, however. Several participants
show U-shaped behaviour in the sense that they may have passed the
acquisition criterion at a certain test time but fail to do so at a later test time.
For example, in the ≥3/6 scale, learned ID1 had ‘acquired’ AGRc at T2 but
not at T3.
It is noteworthy that in all four scales, the majority of learners have
acquired at least some of the stage 2 phenomena before they reach stage 5. In
the ≥5/6 and 6/6 scale there are no exceptions to this developmental
tendency. In the ≥4/6 scale, learner ID11 has acquired AGRv at T1 while not
yet having acquired the stage 2 phenomena, and in the ≥3/6 scale learner ID8
has acquired AGRc at T1 without having acquired the stage 2 features.
These two exceptions aside, the longitudinal data support PT in the sense
that the acquisition of stage 5 phenomena never precedes the acquisition of
(at least some of the) stage 2 phenomena. Instances in which the learners
have not yet acquired all stage 2 phenomena by the time they have acquired
stage 5 phenomena are abundant, but these cases need not contradict the
applicability of PT to language comprehension.

23
Language Acquisition and Use in Multilingual Contexts

Figure 2. Longitudinal scales, ≥3/6 criterion.

Figure 3. Longitudinal scales, ≥4/6 criterion.

24
 Language Acquisition and Use in Multilingual Contexts

Figure 4. Longitudinal scales, ≥5/6 criterion.

Figure 5. Longitudinal scales, 6/6 criterion.

25
Language Acquisition and Use in Multilingual Contexts

Summary and conclusion

This paper explored the applicability of PT to receptive grammar acquisition.
The data presented on the acquisition of the stage 2 features GEN, NEG,
SVO and PLU and the stage 5 features AGRc and AGRv were in line with
the general predictions of PT. From a theoretical point of view, this would
mean that receptive L2 grammar acquisition is governed by similar pro-
cessing procedures as productive L2 grammar acquisition – a finding which
is relevant not only for PT research, but also for the general domain of
language processing, since it means that there must be crucial similarities
between receptive an productive processing. More research is needed for a
more comprehensive understanding of the processes at work.
Methodologically, this paper identified two issues to be dealt with in PT
research on receptive L2 grammar acquisition: the definition and assessment
of morphosyntactic processing and feature unification, and the operational-
ization of emergence. The present, exploratory study has dealt with these
issues by using results from a picture selection task and using four different
acquisition criteria for the implicational scaling analysis. Further research
using alternative research designs and methods is needed to address these
issues in more detail.

References
Baten, K. 2011. Processability Theory and German case acquisition. Language
Learning, 61 (2). 455–505.
Bates, E., MacWhinney, B., Caselli, C., Devescovi, A., Natale, F. & Venza, V.
1984. A cross-linguistic study of the development of sentence interpretation
strategies. Child Development, 55. 341–354.
Boland, J. 1997. The relationship between syntactic and semantic processes in
sentence comprehension. Language and Cognitive Processes, 12 (4). 423–484.
Dulay, H. & Burt, M. 1974. Natural sequences in child second language acquisition.
Language Learning, 24. 37–53.
Ellis, R. 2008. The Study of Second Language Acquisition. 2nd ed. Oxford: Univer-
sity Press.
Fernández, E. & Cairns, H. 2011. Fundamentals of Psycholinguistics. John Wiley &
Sons.
Garman, M. 1990. Psycholinguistics. Cambridge: University Press.
Glahn, E., Håkansson, G., Hammarberg, B., Holmen, A., Hvenekilde, A. & Lund, K.
2001. Processability in Scandinavian second language acquisition. Studies in
Second Language Acquisition, 23. 389–416.
Gompel van, R. & Pickering, M. 2007. Syntactic parsing. In G. Gatskil (ed.), The
Oxford Handbook of Psycholinguistics. Oxford: University Press. 289–307.
Håkansson, G. 1997. Language impairment from a processing perspective. Lund
Working Papers in Linguistics, 46. 133–151.
Håkansson, G. 2001. Tense morphology and verb-second in Swedish L1 children,

26
 Language Acquisition and Use in Multilingual Contexts

L2 children and children with SLI. Bilingualism: language and cognition, 4 (1).
85–99.
Håkansson, G. 2012. English in the Pre-School – Some Results from the ELIAS
Project. Paper presented at The International Symposium on Processability
Approaches to Language Acquisition (PALA). Gent.
Håkansson, G., Pienemann, M. & Sayehli, S. 2002. Transfer and typological prox-
imity in the context of second language processing. Second Language Research,
18 (3). 250–273.
Hatch, E. & Lazaraton, A. 1991. The Research Manual: Design and Statistics for
Applied Linguistics. New York/London: Newbury House.
Howell, D. 2010. Statistical Methods for Psychology. 8th ed. Belmont, USA: Wads-
worth Publishing.
Jordan, G. 2004. Theory Construction in Second Language Acquisition. Amster-
dam/Philadelphia: John Benjamins Publishing Company.
Jurafsky, D. 1996. A probabilistic model of lexical and syntactic access and disam-
biguation. Cognitive science, 20 (2). 137–194.
Kersten, K., Rohde, A., Schelletter, C. & Steinlen, A.K. (eds.), 2010. Bilingual Pre-
schools Volume 1: Learning and Development. Trier: WVT.
Keβler, J-U. & Liebner, M. 2011. Diagnosing L2 Development: Rapid Profile. In M.
Pienemann & J-U. Keβler (eds.), 2011. 133–148.
Larsen-Freeman, D. 1975. The acquisition of grammatical morphemes by adult ESL
students. TESOL Quarterly, 9 (4). 409–419.
Levelt, W. 1989. Speaking: From Intention to Articulation. Cambridge, MA: The
MIT Press.
McDonald, J. 1987. Sentence interpretation in bilingual speakers of English and
Dutch. Applied Psycholinguistics, 8. 379–413.
Ortega, L. 2009. Sequences and processes in language learning. In M.H. Long & C.
Doughty (eds.), The Handbook of Language Learning. Malden, MA: Wiley-
Blackwell. 81–108.
Pearlmutter, N., Garnsey, S.M. & Bock, K. 1999. Agreement processes in sentence
comprehension. Journal of Memory and Language, 41 (3). 427–456.
Pienemann, M. 1998. Language Processing and Second Language Development:
Processability Theory. Amsterdam/Philadelphia: John Benjamins Publishing
Company.
Pienemann, M. 2005. Cross-linguistic aspects of processability theory. Amsterdam/
Philadelphia: John Benjamins Publishing Company.
Pienemann, M. 2007. Processability Theory. In B. VanPatten & J. Williams (eds.),
Theories in Second Language Acquisition: An Introduction. Mahwah, NJ:
Lawrence Erlbaum Association. 137–154.
Pienemann, M. & Keßler, J-U. (eds.), 2011. Studying Processability Theory: An
Introductory Textbook. Amsterdam: John Benjamins Publishing Company.
Rickford, J. 2002. Implicational scales. In K. Chambers, P. Trudgill, & N. Schilling-
Estes (eds.), The Handbook of Language Variation and Change. Malden, MA:
Blackwell Publishing. 142–167.
Steinlen, A.K., Håkansson, G., Housen, A. & Schelletter, C. 2010. Receptive L2
grammar knowledge development in bilingual preschools. In K. Kersten et al.
(eds.), Bilingual Preschools Volume 1: Language and Development. Trier:
WVT. 69–100.

27
Stages of processability and levels of
proficiency in the Common European
Framework of Reference for Languages
The case of L3 French

Jonas Granfeldt
Malin Ågren

Introduction
The work of Gisela Håkansson has for many years been an important source
of inspiration for both of us. In particular her longtime work on L2 Swedish
within PT has been decisive for us in previous attempts to discuss L3 French
within this framework (Ågren 2008). The present paper is an extension of
this work.
The aim of the present paper is to investigate empirically the relationship
between second language proficiency (L2P) and second language develop-
ment (L2D) in a corpus of written L3 French. Second Language Acquisition
(SLA) has traditionally been concerned with describing and understanding
L2D, most notably through the study of developmental sequences and
stages. Language testers and language testing research are interested in
capturing and measuring the broader concept of L2P at a given time. The
question concerning a developmental relationship between L2P and L2D is
not new and has been answered differently in the past. Within SLA, some
researchers view L2D and L2P as separate theoretical constructs (Pienemann
& Johnston 1987; Pienemann & Mackey 1992). R. Ellis (2008), for example,
calls for attempts to match developmental levels and proficiency levels since
he suspects that these two linguistic dimensions might in fact be a com-
parison of “apples and oranges” (Ellis 2008, note 7). The question we ask in
this study is to what extent L2P and L2D develop in parallel in a group of L3
learners of French.
L2P can be defined as “a person’s overall competence and ability to per-
form in L2” (Thomas 1994:330, footnote 1), to which we would like to add
“at a given point in time” in order to underline the fluctuating and develop-
mental aspects of L2P. Hulstijn (2011, 2012) has recently suggested a

28
 Language Acquisition and Use in Multilingual Contexts

subdivision of L2P into Basic Language Cognition (BLC) and Higher

Language Cognition (HLC). The separation is motivated by the fact that
more advanced aspects of L2P can be related to contextual (learning) factors
which depend on the individuals’ intellectual capacities and degree of formal
schooling, rather than on purely linguistic skills. Higher levels of L2P cannot
be reached, Hulstijn argues, without formal education. The distinction
between BLC and HLC is thus needed in order to separate linguistic and
intellectual skills as components of L2P (cf. Cummins, 1980, on a related
division). BLC concerns the implicit knowledge and automated use of fre-
quent and basic morphosyntactic constructions and lexical items.
L2P has been operationalized and measured in a variety of ways (see
Hulstijn 2011) but there are essentially two broad approaches to measure
L2P. On the one hand, the psychometric approach includes standardized
tests (i.e. TOEFL, DELF), cloze tests, C-tests etc.1 On the other hand, the
holistic approach relies either on learners’ self-rating or behavioral rating
scales where expert raters assess L2 performance. In the holistic approach,
criterion-referenced scales of language proficiency can be used and expert
judges evaluate learner language by applying these scales, independently of
target language. Currently, the most prestigious and well-known example is
the Common European Framework of Reference for Languages (CEFR)
which assesses communicative language proficiency. We return to the CEFR
below.
Understanding and defining L2D is one of the core topics in SLA
research. L2D can tentatively be defined as the progressive growth of one or
more aspects of the interlanguage system (phonology, morphology, syntax,
etc.). L2D could be viewed as a necessary but not sufficient subcomponent
of L2P. It is often described via the definition of developmental sequences,
which crucially are thought to be invariable and impermeable with respect to
external factors such as learning situation, type of input, etc. An increasing
number of different models and theories have been put forward to describe
and account for L2D in different learners (Towell & Hawkins 1994;
Pienemann 1998, 2005; Sharwood-Smith & Truscott 2005). Models of L2D
tend to be language independent and some have also been empirically tested
on cross-linguistic data. One well-known and established model of L2D is
Processability Theory (PT) (Pienemann, 1998; Pienemann & Håkansson
1999) which over the years has been tested on a variety of typologically
different languages. We return to PT below.

1
TOEFL is also known as the Test of English as a Foreign Language. It is designed and
administrated by the Educational Testing Service. The DELF test is the French equivalent.
DELF is the acronym of Diplôme d'études en langue française. The test is designed and
administrated by Centre international d'études pédagogiques. A C-test is a specific type of
cloze-test used in language testing (Grotjahn, 2010).

29
Language Acquisition and Use in Multilingual Contexts

The specific aim of the present study is to investigate empirically the

possible relationship between CEFR, a model of communicative L2P, and
PT, a model of L2D. The rationale is that the relationship between L2P and
L2D is debated and needs further attention. With Hulstijn’s (2011, 2012)
division between BLC and HLC in mind, it seems important to understand at
what point in the learners’ trajectory L2D and L2P might be more or less
associated. A possible hypothesis is that L2P in the BLC range would be
closer associated with L2D since both are defined as constructs reflecting
implicit knowledge of language and automated use. Moreover, both are in-
dependent from contextual and individual factors, such as the degree of
formal schooling. Hulstijn criticises the CEFR for not being clear about the
relationship between L2P and L2D and says that:
Any association between CEFR levels of L2P [L2 Proficiency] and L2
development as studied in the second language acquisition (SLA) literature
would be completely misplaced […], unless empirical studies show evidence
in its support. (Hulstijn 2011:241)

The remainder of this paper will be dedicated to a small-scale empirical

study which aims to address this challenge.

The Common European Framework of

Reference for Languages (CEFR)
The CEFR provides a language-independent description of communicative
proficiency at six levels. The levels (A1, A2, B1, B2, C1 and C2) are orga-
nised according to three broad proficiency bands: Basic User (A), Inde-
pendent User (B) and Advanced User (C). The CEFR is action-oriented;
language learners are viewed as language users and as social agents who
accomplish communicative activities. These activities all involve language
in a broad sense and the CEFR describes what a learner can do with respect
to a specific task at a certain level of communicative proficiency and how
well s/he can do it.
Table 1. From Overall written interaction (Council of Europe, 2001, chapter 4).
CEFR Level Descriptor
B1 Can convey information and ideas on abstract as well as concrete
topics, check information and ask about or explain problems with
reasonable precision.
A2 Can write short, simple formulaic notes relating to matters in areas
of immediate need.

The CEFR encompasses four categories of language activities of this type:

reception, production, interaction and mediation.

30
 Language Acquisition and Use in Multilingual Contexts

The scales of the CEFR define linguistic, pragmatic and socio-linguistic

competences needed to carry out the activities. In particular chapter 5
presents communicative language competences (cf. Canale & Swain 1980).
The way scales are presented in the CEFR could lead one to believe that
functional and competence-based scales should be interpreted together.
Hence, learners at, say, level A2 with respect to Overall written interaction
(Council of Europe, 2001:83) should simultaneously be at the same level A2
with respect to competence-based scales in chapter 5, like for example
Vocabulary control (Council of Europe 2001:112). In his paper, Hulstijn
(2007:664) discusses such a “parallel” reading of CEFR functional and com-
petence-based scales from a SLA perspective. At least three types of L2
language users can be identified: a) learners who can do few language tasks
but with high linguistic quality, b) learners who can do many language tasks
but with low linguistic quality and c) learners whose range of language tasks
is parallel to their linguistic ability. Learner types a and b display “uneven
profiles” but as a result of the way CEFR presents the scales, only the third
learner type is included.2

Processability theory
Processability Theory (Pienemann 1998, 2005) is a psycholinguistic theory
of SLA which explains developmental sequences in L2 acquisition in terms
of language processing. According to this approach learners develop skills
needed to process the target language grammar in a highly systematic way.
Importantly, grammatical structures can only be produced in the L2 if the
necessary processing procedures are available.
The processing hierarchy proposed in Pienemann’s original version of
PT, illustrated in Table 2, identifies five stages of development based on
different levels of information exchanged between constituents (i.e. feature
unification). The main idea is that the activation sequence of processing
procedures used in language production (from 1 to 5 below) is also valid for
language acquisition, which follows the same implicational order. Starting
from stage 1, all subsequent stages of development mirror increasing de-
mands of processing capacity involved in the morphosyntactic operations.

2
On a single occasion the CEFR recognizes the existence of “uneven profiles” (Council of
Europe 2001: 17), i.e. learners who are at different levels of proficiency in different activities,
but nothing is said about the frequency or specificities of such learners.

31
Language Acquisition and Use in Multilingual Contexts

Table 2. Stages of development and processing procedures according to PT (adapted

from Pienemann, 1998).
PT stage Processing Procedure Information Example of morphol.
exchange outcome in French
5 Subordinate clause Main– sub. clause Subjunctive in sub.clause
procedure
4 S(entence)procedure Inter-phrasal Subject-verb agreement
3 Phrasal procedure Phrasal NP agreement
2 Category procedure No exchange Lexical morphemes
1 Word or lemma access Words, chunks -

According to PT, the notion of storage of grammatical information in

memory is crucial to the acquisition process. The further away the source
and the target of feature unification, the longer grammatical information
needs to be stored in memory, and the later the morphosyntactic structure
will be acquired. Therefore, inter-phrasal agreement (stage 4) takes longer
time to acquire than phrasal agreement (stage 3). The former asks for ex-
change of grammatical information over phrasal boundaries (between NP
and VP), an operation that involves higher processing procedures, whereas
the latter calls for more local information exchange within NP. As illustrated
in previous cross-linguistic investigations of PT (see Pienemann 2005) the
processing hierarchy affects morphology and syntax differently in various
languages due to their different morphosyntactic rule systems. In this study,
we focus on how this model applies to the development of morphosyntax in
L3 French (Ågren 2008).

Research questions
We have two research questions for the present study:
 RQ1: To what extent is L2P, as measured by the CEFR, and L2D as
defined by PT related in a corpus of written L3 French?
 RQ2: How frequent is the presence of uneven profiles in the data?

Data and method

The learners
All participants were pupils at different schools in the city of Lund in
southern Sweden recruited through personal contacts with teachers at the
schools. Written data were collected from 38 L3 learners of French. 36 of the
38 learners (95%) reported that Swedish was their mother tongue and the
language used in the home. The learners were at two different levels: 22 in
year 9 (15–16 years old, 59% female) and 16 in their final year of upper-

32
 Language Acquisition and Use in Multilingual Contexts

secondary school (18–19 years old, 69% female). All 38 learners success-
fully completed both tasks and were rated as being at least at the A1 level.

The tasks
A website was designed to be used for data collection. The students wrote
their texts directly on the web page, without any kind of support. A total of
76 L3 French texts were collected (38 x 2) using this procedure. All learners
were asked to complete two written communicative tasks. The tasks were
adapted from two of the five tasks used by Alanen et al. (2010) in a study of
young and adult learners’ L2 English and L2 Finnish linked to the CEFR.
Task 1 instructed students to write an email message to their French teacher
explaining why they had been absent from school and asking for some
information on an upcoming French test. The instructions to the older
learners in upper-secondary school were similar but slightly more elaborated
than those used in year 9 (for further details, see Granfeldt et al. in press). In
task 2, students were asked to write a narrative about something nice, funny
or special that they had experienced. They were instructed to explain what
happened to them and why the event was particularly exciting or memorable.
Participants were allowed 40 minutes to complete both tasks. All participants
were able to complete the tasks within this time frame.

The CEFR raters

Two experienced CEFR raters were asked to read the texts and provide them
with a CEFR score ranging from A1 to C2. As a basis for their assessment,
the raters were given a CEFR scale that had been compiled from several of
the CEFR scales. The compiled scale consisted of “can-do statements”, and
accuracy was never mentioned. The raters assessed the texts independently
of each other and were asked to do the following: a) rate each text using the
CEFR scale (A1, A2, B1, B2, C1 and C2), b) indicate the degree of certainty
of each rating on a 4-point scale ranging from “completely certain” to “com-
pletely uncertain”, and c) to provide an alternative CEFR rating if, and only
if, the indicated degree of certainty was low (i.e. “uncertain” or “completely
uncertain”). No texts were included in the final analysis where both raters
were “completely uncertain” or “uncertain”.
The degree of inter-rater agreement between the two raters was measured
by Kronbach’s alpha. Alpha was measured to .804 which according to
DeVellis (1991:85) corresponds to a “very good reliability”.

PT-analysis
One of the authors with previous experience of PT read the learner texts and
analyzed them according to the PT framework. Since the tasks were adapted
to match CEFR criteria, this meant that data density for some structures was
low. Therefore, each analysis was evaluated on a 4-point scale according to
the degree of certainty of the analysis. Out of the 76 CEFR rated texts, 61

33
Language Acquisition and Use in Multilingual Contexts

texts contained a sufficient amount and varied set of structures in order to

ensure a reasonable PT analysis with a certainty score of 3 or 4. These 61
texts were kept for the final comparison with the CEFR ratings.
Within the PT framework the emergence criterion is applied. Emergence
refers to the first systematic and productive use of a certain structure, which
indicates that the learner, in principle, can carry out a specific grammatical
operation. In our PT analysis, the researcher studied the systematic use of
certain structures based on three sources of available evidence in each
learner text: 1) minimal pairs, 2) creative constructions (i.e. overuse of a
grammatical rule), and 3) a certain amount of lexical variation in the use of a
particular structure (e.g. the same morpheme used with a range of different
lexemes). In the PT analysis, at least two minimal pairs or two creative
constructions or three varied lexical items within a morphological pattern
were required in order to identify a specific developmental stage (cf. Ågren
2008; Pallotti 2007; Pienemann 1998).

Results
In Figure 1 below, CEFR ratings of the 61 texts are plotted against the
analysed PT-stage. Each learner text is represented by a circle in the figure.
Since inter-rater agreement was estimated to be sufficiently high (see above),
a single CEFR score for each text was computed by calculating a mean score
from the two CEFR ratings. We observe that, overall, there is a linear
correlation between the CEFR level and the PT stage.

B2B2

B1B1

A2
A2

A1
A1

Figure 1. Correlation of CEFR rating and PT developmental stage.

Figure 1: Correlation of CEFR rating and PT developmental stage
The CEFR ratings cover four levels and range from A1 to B2. Texts that are
plotted between two CEFR-levels have been rated differently by the two

34
 Language Acquisition and Use in Multilingual Contexts

raters. For example, a text which is plotted between A1 and A2 has been
rated A1 by one rater and A2 by the other rater.
The results with respect to PT also cover four stages and range from PT
stage 1 to stage 4. We observe that there is an increasing amount of variation
in the relationship between the CEFR level and the analysed PT stage.
According to the results in this small-scale pilot study, PT stages 1 and 2 are
associated with CEFR levels A1 and A2. Interestingly, the dispersion
increases at more advanced stages and PT stage 4 contain texts that have
been rated from A2 (1 text) up to above B2 (2 texts). We also observe that,
with a single exception, no text rated below B1 was analysed as PT stage 4.
To answer our first research question, we statistically investigated the
strength of the association between the rated CEFR score (the mean) and the
analysed PT stage using a Spearman rank order correlation analysis. The
results indicate a very strong association between the average rated CEFR
level and the analysed PT stage, (rs[62] = .86, p<0.001).
In the remainder of this section we will address the second research
question, namely the presence and frequency of uneven profiles.
For the purpose of this study we define a balanced profile as a text which
follows the main linear trend expressed in Figure 1. These are texts where
communicative proficiency (L2P), measured by CEFR, and morphosyntactic
development (L2D), measured by PT, go hand in hand (cf. “flat profile”,
Council of Europe, 2001:43). We observe that balanced profiles seem to be
dominant at lower levels and stages, which is not very surprising due to the
limited command of the L2/L3 at these stages. Example (1) below illustrates
a balanced profile at a more advanced level, where utterances are linked into
a clear and coherent narrative (CEFR, level B2) and where the morpho-
syntactic procedures at PT stage 4 have emerged (see phrasal agreement,
stage 3, in solid underlining; interphrasal agreement, stage 4, in dashed
underlining).
(1) JL text 2: CEFR level B2 (both raters), PT stage 4

Il y a quelques semaines, je suis allée visiter une amie qui habite à Göteborg.
En même temps, il y avait aussi ma meilleure amie de Lund qui allait visiter
son amie à Göteborg. Nous, toutes les deux, sont donc allées en train pour
Göteborg. En sortant du train à la gare de Göteborg, j'ai vu mon amie et on
s'embrassées. Ma meilleure amie, elle a aussi rencontré son amie. Soudain,
ma copine regarde la copine de ma meilleure amie et elle recoît un regard
étonné. En fait, la copine de ma meilleure amie et ma copine de Göteborg
étaient camarades de classe. Quand on a toutes compris cette coincidence
sensationelle on s’est mises à rire...

However, it is not always the case that high/low communicative proficiency

correlates with high/low morphosyntactic development. Texts showing un-
even profiles appear as outliers in Figure 1. For instance, as exemplified in

35
Language Acquisition and Use in Multilingual Contexts

(2), the learner ES’s text was judged by both CEFR raters as belonging to
level B1, where the learner is able to write accounts of experiences,
describing feelings and reactions in simple connected text and link series of
discrete elements into a connected, linear sequence of points. Indeed, ES is
telling a rather straightforward narrative, where events from the past and the
present are linked by the use of different tenses (imperfect and present
tense), adverbials (ne…jamais, la première fois, chaque fois…) and pro-
nominal reference (la chanson... je l’ecoutais). In Figure 1, CEFR B1 and
B1+ and PT Stage 4 are closely associated. However, the PT analysis of ES’
texts indicates stage 3, because the use of subject-verb agreement was not
consistent enough to indicate availability of processing mechanisms at PT
stage 4 (phrasal agreement in solid underlining has emerged, interphrasal
agreement in dashed underlining is questionable). Thus, the communicative
skills expressed in (2) seem to be more advanced than what could perhaps be
expected given the stage of development.
(2) ES text 2: CEFR level B1 (both raters), PT stage 3

Je ne oblie jamais le premiere fois que je ecoutais à Edith Piaf. J'avait 12 ans
et la musique en français etait très beau et je commencait de pleurer. C'est la
première fois que je comprennait que je veux lire la français à l'école. La
chanson etais "Paris" et c'est encore ma chanson favorie en français. Chaque
fois j'ecoute la chanson je pense que la première fois que je l'ecoutais. C'est
une memoire très forte...

Other texts, as exemplified in (3), show the opposite pattern where the level
of morphosyntactic development was analysed at PT stage 4 (consistent
subject-verb agreement) whereas the communicative proficiency was rated
at a somewhat lower level (A2/B1) than could be expected (cf. the trend in
Figure 1).
(3) TA text 2, CEFR level B1/A2, PT stage 4

Je vais me souvenir toujours de ce jour lorsque moi et mes soeurs étions avec
vos amis à notre maison au bord de la mer. Là, il y a beaucoup de vaches.
Un jour nous avez oublié le collier de notre chien dans un endroit où il y a
des vaches. Quand nous sommes allées le chercher il y a telle de vaches dans
cet endroit. Nous pensions que cela était très intéressant. Ma petite soeur
voulait monter qu'elle n'avait rien peur et elle était allé proche d'un vache.
Alors la mère de cette vache était furieux et commençait à courir après nous.
Nous avons couru en pleurant. Je n'ai jamais couru si vite... Donc tout allait
bien et maintenant c'est plutôt un bon souvenir.

Even though uneven profiles are present in the text sample examined in this
study, we conclude that the majority of texts show rather homogeneous pro-
files, where L2P and L2D point in the same direction.

36
 Language Acquisition and Use in Multilingual Contexts

Summary
The present small-scale study aimed at investigating a possible relationship
between second language proficiency (L2P) and second language develop-
ment (L2D) in a corpus of L3 written French. The learners were Swedish
secondary students. The learner texts were assessed according to the CEFR
by two experienced CEFR raters, measuring communicative L2P. The same
texts were analysed according to Processability Theory, which is a theory of
L2D. We found a strong overall correlation between the CEFR ratings and
the PT analysis (rs[62] = .86, p<0.001). We want to underline, however, that
the observed correlation cannot in any way be taken as evidence that the
underlying constructs are related.
In addition, we observed that the existence of uneven profiles in the data,
i.e. learners with stronger communicative proficiency than morphosyntactic
development or vice versa, typically becomes more frequent at more ad-
vanced stages. Up to CEFR B1 and PT stage 3, learners’ communicative
proficiency and morphosyntactic development seem to develop more or less
at the same rate. This preliminary finding is potentially interesting in the
view of L2P as divided between BLC and HLC (Hulstijn 2012). If BLC re-
flects implicit knowledge and automated language use, it actually comes a
bit closer to the definition of L2D used in PT, which, in turn, could explain
the better fit between CEFR and PT at lower levels and stages, as indicated
in Figure 1. Future research will have to investigate this association in more
detail.

References
Ågren, M. 2008. À la recherche de la morphologie silencieuse: sur le développement
du pluriel en français L2 écrit. Études Romanes de Lund, 84. Lund University.
Alanen, R., Huhta, A. & Tarnanen, M. 2010. Designing and assessing L2 writing
tasks across CEFR proficiency levels. In I. Bartning, M. Martin & I. Vedder
(eds.), Communicative proficiency and linguistic development: intersections
between SLA and language testing research. Eurosla Monographs: European
Second Language Association. 21–56.
Canale, M. & Swain, M. 1980. Theoretical bases of communicative approaches to
second language teaching and testing. Applied Linguistics, 1. 1–47.
Council of Europe, 2001. Common European Framework of Reference for
Languages: language, teaching, assessment. Cambridge: Cambridge University
Press.
Cummins, J. 1980. The construct of language proficiency in bilingual education. In
J.E. Alatis (ed.), Current issues in bilingual education (GURT 1980).
Washington, DC: Georgetown University Press. 81–103.
DeVellis, R.F. 1991. Scale Development. Newbury Park: Sage Publications.
Ellis, R. 2008. Investigating grammatical difficulty in second language learning:
Implications for second language acquisition research and language testing.
International Journal of Applied Linguistics 18 (1). 4–22.
Granfeldt, J., Gyllstad, H. & Källkvist, M., in press. Linguistic correlates to comm-

37
Language Acquisition and Use in Multilingual Contexts

unicative proficiency levels of the CEFR: The case of syntactic complexity in

L2 English and L3 French. In C. Rosén, P. Simfors & A.K. Sundberg (eds.),
Rapport från ASLA:s vårsymposium, Linköping, 2012.
Grotjahn, R. 2010. Der C-Test: Beiträge aus der aktuellen Forschung. Frankfurt am
Main: Peter Lang.
Hulstijn, J. 2007. The shaky ground beneath the CEFR: Quantitative and qualitative
dimensions of language proficiency. The Modern Language Journal, 91. 662–
667.
Hulstijn, J. 2011. Language proficiency in native and nonnative speakers: An agenda
for research and suggestions for second language assessment. Language
Assessment Quarterly, 8. 229–249.
Hulstijn, J. 2012. The construct of language proficiency in the study of bilingualism
from a cognitive perspective. Bilingualism: Language and Cognition, 15 (2).
422–433.
Pallotti, G. 2007. An operational definition of the emergence criterion. Applied Ling-
uistics, 28 (3). 361–382.
Pienemann, M. 1998. Language processing and second language development:
Processability Theory. Amsterdam: John Benjamins Publishing Company.
Pienemann, M. (ed.), 2005. Cross-linguistic Aspects of Processability Theory.
Amsterdam: John Benjamins Publishing Company.
Pienemann, M. & Johnston, M. 1987. Factors influencing the development of lang-
uage proficiency. In D. Nunan (ed.), Applying Second Language Research,
Adelaide: NCRC. 45–142.
Pienemann, M. & Mackey, A. 1992. An empirical study of children’s ESL develop-
ment and Rapid Profile. In P. McKay (ed.), ESL development. Language and
literacy in schools. Commonwealth of Australia and national languages and
literacy institute of Australia. 115–259.
Pienemann, M. & Håkansson, G. 1999. A unified account toward the development
of Swedish as L2. Studies in Second Language Acquisition, 21 (3). 383–420.
Sharwood-Smith, M. & Truscott, J. 2005. Stages or continua in second language
acquisition: A MOGUL solution. Applied Linguistics, 26 (2). 219–240.
Thomas, M. 1994. Assessment of L2 proficiency in second language acquisition
research. Language Learning, 44 (2). 307–336.
Towell, R. & Hawkins, R. 1994. Approaches to second language acquisition.
Clevedon: Multilingual Matters.

38
So you think gestures are compensatory?
Reflections based on child and adult learner data3

Marianne Gullberg

Introduction
When I told Gisela that I wanted to examine how adult second language (L2)
speakers used gestures as communication strategies in my doctoral
dissertation, I remember her swallowing hard, then smiling and cheerfully
saying How exciting! I thought it fitting to honour her by revisiting the can
of worms I opened then, and take a new look at the question of whether, and
if so how, gestures can be said to be compensatory.
In a seminal paper from 1985 entitled So you think gestures are non-
verbal? David McNeill challenged the then dominant view of gestures as a
communicative frill of no consequence to our understanding of language and
linguistic processing (McNeill 1985). The paper listed arguments for why
gestures are in fact verbal (i.e. linguistic), highlighting their close
relationship with spoken language. Some 30 years later, this position has
become well established. Evidence continues to accumulate for the close
connection between gesture and language in language development, break-
down, in processing, etc. Although the link itself is no longer questioned, the
exact nature of the relationship and the reasons for why we gesture remain
illusive.
It is a common lay assumption that speakers in expressive trouble use
hand and foot solutions to resolve them. Gestures – particularly represent-
ational or referential gestures which convey meaning about a referent (e.g.
size, shape, etc.) – are seen as a compensatory tool to bridge the gap between
communicative intention and available expressive means. This view is also
common in research targeting language users who are “challenged” or “less
competent”. It is explicit in studies of adult second language acquisition and
3
A version of this paper was given at the GESPIN 2011 conference, Bielefeld, 2011. I thank
the audience for their input. I gratefully acknowledge financial support from the Max Planck
Institute for Psycholinguistics; the Nederlandse Organisatie voor Wetenschappelijk
Onderzoek, MPI 56-384; and Vetenskapsrådet A0667401. I also thank my collaborators M.
Graziano, B. Narasimhan, H. Hendriks, and M. Hickmann.

39
Language Acquisition and Use in Multilingual Contexts

bilingualism (e.g. Nicoladis, Pika & Marentette 2009). It is also implicit in

many studies of child language acquisition, where gestures are viewed as
precursors to language in infants (Bates 1979; Liszkowski 2008), or as
markers of transitional knowledge states in older children (Goldin-Meadow
& Butcher 2003). It is also found in studies of atypically developing, or
impaired populations (e.g. children with specific language impairment (SLI),
or Down syndrome), and patients with aphasia (Blake, Myszczyszyn, Jokel
& Bebiroglu 2008; Fex & Månsson 1998; Goodwin 2000; Rose 2006;
Stefanini, Recchia & Caselli 2008).
It is not an unreasonable notion. A linguistic compensatory device needs
to have certain properties. First, it needs to have expressive power and rich
semiotic affordances. Gestures do. Although gestures are not themselves
language, they can carry the full expressive burden as seen in so-called home
sign, the conventionalised gesture systems developed by isolated deaf
children (Goldin-Meadow 2003; Senghas, Kita & Özyürek 2004), and the
ease with which speakers can express propositional content in gesture alone
(Singleton, Goldin-Meadow & McNeill 1995).
Second, a linguistic compensatory device needs to be tightly linked to
language. Gestures are. All current theories of gesture see language and
gestures as linked, forming an integrated whole that is planned and pro-
cessed together in comprehension, production, and development (Clark
1996; Kendon 2004; McNeill 1992, 2005). Even if the precise nature of the
relationship is under debate, the link itself is undisputed (De Ruiter 2007;
Kendon 2004; Kita & Özyürek 2003; Krauss, Chen & Gottesman 2000;
McNeill 1985). Third, a compensatory device should be relevant to
addressees. Gestures are. There is a large body of behavioural and neuro-
cognitive evidence that addressees attend to, and integrate gesture
information as part of their linguistic processing (Gullberg & Kita 2009;
Hostetter 2011; Kelly, Özyürek & Maris 2010; Willems, Özyürek & Hagoort
2007).
Gestures thus seem to fit the bill and to have all the necessary properties
for a compensatory device. The question remains whether compensatory
needs drive their production and use. The assumptions regarding
compensation make a number of predictions. In this chapter I discuss two of
them, confronting them with some empirical facts. The first one is the notion
that gestures replace speech in communicative trouble. The second is that
gestures are used to convey additional meaning relative to speech when
speech is deficient.

Do gestures replace speech?

One of McNeill's arguments for why gestures are verbal (linguistic) was the
tight co-production of the modalities. He claimed that when speech stops,

40
 Language Acquisition and Use in Multilingual Contexts

gesture stops too. Yet, it is a common lay and professional assumption that
gestures replace speech: when speech stops, gesture starts. That is, if you do
not know the word for, say, key, you perform a gesture in silence pretending
to manipulate a key to open a door. Interestingly, this view is also found in
some of the theories on the speech-gesture relationship. The Lexical
Retrieval Hypothesis, for example, considers gesture production to be driven
by lexical retrieval difficulties (Krauss et al. 2000). Gestures are thought to
activate lexical forms (of content words) through cross-modal priming of
motor patterns. Some L2 studies draw on this theory to explain why L2
speakers typically have higher gesture rates than native speakers, suggesting
that L2 speakers gesture to activate lexical forms (e.g. Nicoladis et al. 2009).
Other theories see gestures as facilitating the selection and linearization of
information for expression in speech (information packaging) where gestures
help speakers spatially explore their options (Kita 2000). Both sets of
theories predict that, if gestures are to facilitate lexical retrieval or
formulation, they should predominantly occur with disfluencies and in
silence: when speech stops, gesture starts.
However, there is some evidence that, contrary to these predictions,
gestures and speech stop together. For example, adult stutterers’ gestures
tend to stop with speech disruption and to resume when speech becomes
fluent again (Mayberry & Jaques 2000). In normal speakers, gestures stop
some 200 ms before speech does, suggesting a tight integration in production
planning (Seyfeddinipur 2006). Adult L2 speakers of French and Swedish
produce significantly more strategic gestures that are complementary to
speech (i.e. occur with speech) than substitutive (i.e. occur in silence;
Gullberg 1998).
Preliminary results from an ongoing study of fluent and disfluent
narrative speech in normally developing monolingual children (age 4, 6 and
9), native and non-native adult speakers of different languages suggest that
all groups display the same patterns (Graziano & Gullberg 2010). Examining
intra-clausal disfluencies (filled and unfilled pauses, interruption, repetition,
lengthening, self-correction, and combinations), we find that gestural strokes
(i.e. the most meaningful parts of gestural movements, Kendon 1980) overall
occur with fluent rather than disfluent speech in all groups.
In (1), an adult Dutch L2 speaker of French switches into English when in
trouble. During the disfluency (they ehm eh eh) the ongoing gesture stroke is
suspended in mid-air with immobile hands, despite three long pauses. As she
resumes fluent speech, saying they can sit on both sides, both hands start
moving up and down as she indicates a seesaw in a representational gesture.
During the disfluent phase, her hands do nothing. In (2), a native Dutch
speaker produces a gesture outlining a square shape as he says met z'n
handen ‘with his hands’. He abandons it as he produces a filled pause. In the
long silence that follows, there is no gesture. Only when he resumes speech
saying de dossierkast open doet ‘opens the filing cabinet’ does he produce a

41
Language Acquisition and Use in Multilingual Contexts

gesture as if pulling out a drawer. In example (3), finally, an Italian child

aged 6 starts a gesture preparation as she says ci faceva ‘s/he made’. She
then becomes disfluent and she suspends into a hold when she repeats the
determiner una. She then drops her hands entirely as she says mh ‘hm’. In all
speakers, thus, gestures stop when speech stops. Disfluencies, whether filled
or silent, are typically accompanied by nothing or by suspended gestures.
When speech resumes, gestures resume too.
(1) [they ehm <1.9 s> eh <1.6 s> eh <2.2 s> they can sit on both sides] 4
(subject D17L2)
(2) nou als die als die zeg maar eh met z'n [met z'n handen] eh <3.7 s> de
dossierkast open doet (subject D05L1)
'well as he as he kind of uh with his with his hands uh <3.7 s> opens the
filing cabinet'
(3) mh ci faceva [una una] mh (Graziano & Gullberg 2010)
'mh [s/he] made [a a] mh'

Interestingly, when gestures are completed during disfluencies, they are not
representational gestures related to the content of the word sought. Instead,
such gestures (called pragmatic gestures by Kendon (2004), conduit
gestures by McNeill (1985), and thinking gestures by Gullberg (2011b))
comment on the breakdown itself but do not reflect properties of the sought
referential expression. In (4), a Dutch L2 speaker of French performs a wrist
circling movement during the filled pause and then throughout the silent
pause that follows. The gesture indicates an on-going word search but
nothing about the content sought.
(4) il est eh eh[m <1.21 s>] (subject D21L2)
he is uh uh[m <1.21 s>]

These findings all support McNeill’s original claim that gestures and speech
form an integrated system that is co-produced. The results do not tally with
assumptions of gestures replacing lexical items in silence. The answer to the
question of whether gestures compensate for speech by replacing it cannot
be “yes” – at least not in a simplistic sense.

4
Gestural strokes are indicated in square brackets; gestural holds, that is, gestures that have
stopped in mid-air, are indicated by underlining. Pauses are indicated in pointy brackets with
duration given in seconds.

42
 Language Acquisition and Use in Multilingual Contexts

Do gestures replace lexical content?

Another common set of assumptions concern the semantic connection
between speech and gestures and cross-modal co-expressivity. A first
assumption is that (representational) gestures compensate for lexis. A
second, related assumption is that gestures are (therefore) non-redundant
with speech, and convey additional meaning. Both of these assumptions
raise questions. McNeill claimed that gestures and speech form an integrated
mode of expression, characterised by semantic-pragmatic and temporal
coordination or co-expressivity. Note that co-expressive cannot mean
identical (cf. De Ruiter 2007; McNeill 1985). Since gestures convey
information in a different format from speech, they clearly provide
information that may not be expressed in speech such as size, speed,
direction, and so on. Nevertheless, analyses of co-expressivity that focus on
specific semantic components often find remarkable overlap across the
modalities.
The domain of caused and voluntary (translocational) motion has been
well examined in this regard. For example, the semantics of placement verbs
(e.g. to put a cup on a table) differ dramatically cross-linguistically (e.g.
papers in Kopecka & Narasimhan 2012). Native speakers of different
languages gesture differently when talking about placement depending on
the semantics of their placement verbs (Gullberg 2009, 2011a). For example,
English speakers saying put generally gesture only about the path of the
movement reflecting verb semantics ‘cause to move to another location’
(Gullberg, 2009). In contrast, Dutch speakers, who must use one of two
verbs (zetten ‘set’ vs. leggen ‘lay’) depending on the properties of the object
being placed, produce gestures with hand shapes incorporating the object
with the path of the movement, reflecting the more complex semantics of
these verbs, ‘cause a particular object to move to a location such that it ends
up in position X’ (Gullberg, 2011a). Adult native speakers thus focus on the
same information both in speech and gesture. Crucially, English speakers do
not gesture about objects when the verb does not refer to them. Speech and
gestures are co-expressive.
Although the two Dutch verbs are obligatory and frequent in the input,
Dutch children have difficulties with them even at age five (Narasimhan &
Gullberg 2011) using only one of them, leggen, for all situations.
Interestingly, children using only one verb gesture only about path in non-
adultlike ways. In contrast, children who use both verbs, zetten and leggen,
also produce gestures that incorporate the object with the path, as do Dutch
adults (Gullberg & Narasimhan 2010). In other words, when children
distinguish verbs based on objects, they gesture about objects. When they do
not, they gesture only about paths. Crucially, there is no evidence that
children who use only one verb express object information in gesture instead
of in speech. They do not seem to compensate for linguistic problems in

43
Language Acquisition and Use in Multilingual Contexts

gesture. The object simply is not relevant to them. Speech and gestures are
co-expressive even if not adult-like.
Similarly, adult English learners of L2 Dutch have difficulties acquiring
the two Dutch verbs. Like Dutch children they use only one of the verbs for
all placement scenes, or dummy verbs like doen ‘do’ or intransitive
constructions. Their gestures typically look English in that they gesture
about path but not about objects. That is, their path gestures indicate that
their Dutch forms mean something like ‘cause to move to another location’,
just like put. Arguably, speech and gestures are co-expressive. Gestures very
rarely add the object information absent from speech (Gullberg 2009).
Again, there is little evidence that adult L2 speakers compensate for
semantic shortcomings in gesture. The object simply is not relevant to them.
Studies of voluntary motion where the semantic components examined
are path vs. manner of motion show much the same patterns both in children
(e.g., Hickmann, Hendriks & Gullberg 2011) and in adult L2 speakers (e.g.
Stam 2006).
The overall findings suggest that speech and gestures in child (age 3;6
onwards) and adult learners are mainly co-expressive. These results support
the claims in the literature about speech and gesture as an integrated system.
Crucially, the data strongly suggest that to “compensate” and express
additional semantic information in gesture, you need to know that it is
relevant (cf. Gullberg 2011b). Again, the answer to the question of whether
gestures compensate for absent meaning in speech cannot be “yes” – at least
not in a simplistic sense.

Do gestures never compensate?

Does all this mean that gestures are never used as a compensatory device?
Of course not. Gestures can be recruited to solve linguistic problems. When
we examine cases of expressive problems in adult L2 speakers, identified by
clusters of behavioural cues (hesitations, repetitions, slow articulation, or
overt appeal for help), following Faerch & Kasper (1983), we find that
different problems come with different gestural solutions (Gullberg 1998,
2011b). Lexical difficulties are often resolved jointly between learners and
native speakers (NSs).
In (5), a Dutch L2 speaker of French tries to express peindre ‘to paint’ by
saying couloir son maison, an approximation of ‘colour his house’. She also
performs a painting gesture that then goes into a hold. It is held while the NS
uses an echo question, and is then repeated when the learner repeats couloir.
It is then held again while the NS suggests peindre ‘to paint’ while also
performing a painting gesture. It is only at the NS’s second attempt that the
learner accepts peindre, at which point she finally drops her hands.

44
 Language Acquisition and Use in Multilingual Contexts

(5) L2: il veut [couloir son maison 'he wants [colour his house'
NS: il veut 'he wants'
L2: eh couloir son eh 'uh colour his uh'
NS: [peindre] [paint]
L2: une autre cou <…> another col <…>
NS: [peindre] [paint]
L2: oui] peindre yes] paint

Learners thus typically attempt a spoken solution accompanied by a repre-

sentational gesture which is held while the NS provides lexical suggestions.
NSs often repeat the learners’ gestures as they offer solutions. When the
learner accepts and integrates an offered solution, the gestural hold is re-
leased and the initial gesture is repeated, often in unison with the NS. These
types of solutions are also attested in classroom interaction (e.g. Lazaraton
2004), and between aphasic patients and care givers (e.g. Goodwin 2000).
Grammatical difficulties are more complex than solutions to lexical
problems. They often involve the use of space for the creation of abstract
representations of discourse which serve as scaffolding for potential
disambiguation (Gullberg 2006). They typically draw little overt attention
from NSs during the moments of difficulty but are exploited in clarification
sequences. Problems with the interaction itself arising from multiple
disfluencies also need to be managed. We have already seen how pragmatic
gestures with wrist circling movement (the thinking gesture) can be
deployed to stall during a word search (example (4)). These are often
accompanied by withdrawn gaze. Gestures in silence serve as effective floor
holders, managing turn-taking.
In general then, gestures can compensate, but with important quali-
fications. First, gestures do not compensate for lexis in silence – they accom-
pany spoken approximations and are involved in interactive solutions. Gest-
ures also compensate for grammatical and interactive difficulties. Gestural
compensatory behaviour is also subject to individual variation in individual
communicative styles (Gullberg 1998).
Despite the widespread assumption that children also use gestures to
solve (lexical) problems, there is virtually no evidence of such child be-
haviour. Children with SLI and Down syndrome display increased numbers
of representational gestures in naming tasks (Fex & Månsson 1998; Stefanini
et al. 2008), but we know little about the exact details of the temporal
relationship between speech and lexical representations in these studies.
Children also use deictic expressions such as like that which clearly refer
directly to an accompanying gesture. However, they do not seem to serve as
gestural compensation. Examples (6) and (7) from French and English
indicate that children use deictic expressions and gesture with the

45
Language Acquisition and Use in Multilingual Contexts

appropriate lexical material, grimper ‘climb up’ in French and hop in

English.
(6) Elle a [grimpé comme ça]
‘She [climb.ed.up like that]’. [both hands performing repeated grasping
movements while moving upwards]
(7) He was [hopping like that] [moving whole body up and down on chair]
(both examples from Hickmann, et al. 2011: 149)

Again, this does not look like compensation. Such observations raise the
possibility that only mature speakers engage in gestural compensatory be-
haviour. At the very least, it is not straightforwardly evident in children.

Are gestures compensatory? Yes and no

Returning to the initial question, are gestures compensatory, the annoying
answer seems to be yes and no. Speakers can and do recruit gestures to
compensate for linguistic shortcomings, but not all gestures are compen-
satory, not even in language learners. Predictions about gesture production in
silence are not borne out and analyses of co-expressivity point to general
overlap in meaning. Although McNeill was essentially right when he said
that when speech stops, so does gesture, the speech-gesture link is not auto-
matic and compulsory. As seen in adult compensatory behaviour, it can be
decoupled and achieved (De Ruiter 2007; Kendon 2004; Schegloff 1984).
Individuals may choose not to gesture at all. Adult L2 speakers may choose
to tackle a lexical problem by focusing on content or on pragmatic aspects of
interaction. It requires pragmatic interactive skills to do the latter which
children may not have. Fundamentally, you need to know that information is
missing in order to compensate for it.
Compensation as a notion remains surprisingly ill- or undefined (cf.
Gullberg, de Bot, & Volterra 2008). The nature of the relationship between
speech and gesture is multifaceted and depends on the speaker, the
developmental stage, and the type of problem. A more nuanced view of the
notion of compensation is therefore (still) required, and underlying assump-
tions need to be made explicit. We need to consider how and when gestures
are compensatory, and whether there are independent criteria to define them
as such. And it is important to do so. The notion of compensation raises
theoretical challenges for everyone, acquisitionists and gesture scholars
alike. Only if we improve on definitions and criteria can we reshape models
currently built on adult, native, mature speakers to include learners. And
only then can we gain a better understanding of the role of gesture in
language development, and of the intricate relationship between speech and
gesture in general.

46
 Language Acquisition and Use in Multilingual Contexts

References
Bates, E. 1979. The emergence of symbols. Cognition and communication in in-
fancy. New York: Academic Press.
Blake, J., Myszczyszyn, D., Jokel, A. & Bebiroglu, N. 2008. Gestures accom-
panying speech in specifically language-impaired children and their timing with
speech. First Language, 28 (2). 237–253.
Clark, H. 1996. Using language. Cambridge: Cambridge University Press.
De Ruiter, J-P. 2007. Postcards from the mind: The relationship between speech,
gesture and thought. Gesture, 7 (1). 21–38.
Faerch, C. & Kasper, G. 1983. On identifying communication strategies in inter-
language production. In C. Faerch & G. Kasper (eds.), Strategies in inter-
language communication. London: Longman. 210–238.
Fex, B. & Månsson, A-C. 1998. The use of gestures as a compensatory strategy in
adults with acquired aphasia compared to children with specific language
impairment (SLI). Journal of Neurolinguistics, 11 (1–2). 191–206.
Goldin-Meadow, S. 2003. The resilience of language: What gesture creation in deaf
children can tell us about how all children learn language. Hove: Psychology
Press.
Goldin-Meadow, S. & Butcher, C. 2003. Pointing toward two-word speech in young
children. In S. Kita (ed.), Pointing: Where language, culture, and cognition
meet. Mahwah, NJ: Erlbaum. 85–107.
Goodwin, C. 2000. Gesture, aphasia, and interaction. In D. McNeill (ed.), Language
and gesture. Cambridge: Cambridge University Press. 84–98.
Graziano, M. & Gullberg, M. 2010. When words fail: Child and adult language
learners and their gestures during expressive difficulties. Paper, 4th Conference
of the International Society for Gesture Studies.
Gullberg, M. 1998. Gesture as a communication strategy in second language dis-
course. Lund: Lund University Press.
Gullberg, M. 2006. Handling discourse: Gestures, reference tracking, and comm-
unication strategies in early L2. Language Learning, 56 (1). 155–196.
Gullberg, M. 2009. Reconstructing verb meaning in a second language: How
English speakers of L2 Dutch talk and gesture about placement. Annual Review
of Cognitive Linguistics, 7. 222–245.
Gullberg, M. 2011a. Language-specific encoding of placement events in gestures. In
J. Bohnemeyer & E. Pederson (eds.), Event representations in language and
cognition. Cambridge: Cambridge University Press. 166–188.
Gullberg, M. 2011b. Multilingual multimodality: Communicative difficulties and
their solutions in second language use. In J. Streeck, C. Goodwin & C. LeBaron
(eds.), Embodied interaction: Language and Body in the material world.
Cambridge: Cambridge University Press. 137–151.
Gullberg, M., de Bot, K. & Volterra, V. 2008. Gestures and some key issues in the
study of language development. Gesture, 8 (2). 149–179.
Gullberg, M. & Kita, S. 2009. Attention to speech-accompanying gestures: Eye
movements and information uptake. Journal of Nonverbal Behavior, 33 (4).
251–277.
Gullberg, M. & Narasimhan, B. 2010. What gestures reveal about the development
of semantic distinctions in Dutch children's placement verbs. Cognitive
Linguistics, 21 (2). 239–262.
Hickmann, M., Hendriks, H. & Gullberg, M. 2011. Developmental perspectives on

47
Language Acquisition and Use in Multilingual Contexts

the expression of motion in speech and gesture. A comparison of French and

English. Language, Interaction and Acquisition, 2 (1). 129–156.
Hostetter, A. 2011. When do gestures communicate? A meta-analysis. Psychological
Bulletin, 137 (2). 297–315.
Kelly, S., Özyürek, A. & Maris, E. 2010. Two sides of the same coin: speech and
gesture mutually interact to enhance comprehension. Psychological Science, 21
(2). 260–267.
Kendon, A. 1980. Gesticulation and speech: Two aspects of the process of utterance.
In M. Key (ed.), The relationship of verbal and nonverbal communication. The
Hague: Mouton. 207–227.
Kendon, A. 2004. Gesture. Visible action as utterance. Cambridge: Cambridge Uni-
versity Press.
Kita, S. 2000. How representational gestures help speaking. In D. McNeill (ed.),
Language and gesture. Cambridge: Cambridge University Press. 162–185.
Kita, S. & Özyürek, A. 2003. What does cross-linguistic variation in semantic co-
ordination of speech and gesture reveal?: Evidence for an interface represen-
tation of spatial thinking and speaking. Journal of Memory and Language, 48
(1). 16–32.
Kopecka, A. & Narasimhan, B. (eds.), 2012. Events of putting and taking: A cross-
linguistic perspective. Amsterdam: Benjamins.
Krauss, R., Chen, Y. & Gottesman, R. 2000. Lexical gestures and lexical access: a
process model. In D. McNeill (ed.), Language and gesture. Cambridge:
Cambridge University Press. 261–283.
Lazaraton, A. 2004. Gesture and speech in the vocabulary explanations of one ESL
teacher: A microanalytic inquiry. Language Learning, 54 (1). 79–117.
Liszkowski, U. 2008. Before L1: A differentiated perspective on infant gestures.
Gesture, 8 (2). 180–196.
Mayberry, R. & Jaques, J. 2000. Gesture production during stuttered speech:
Insights into the nature of gesture-speech integration. In D. McNeill (ed.),
Language and gesture. Cambridge: Cambridge University Press. 199–214.
McNeill, D. 1985. So you think gestures are nonverbal? Psychological Review, 92
(3). 271–295.
McNeill, D. 1992. Hand and mind. What the hands reveal about thought. Chicago:
University of Chicago Press.
McNeill, D. 2005. Gesture and thought. Chicago: University of Chicago Press.
Narasimhan, B. & Gullberg, M. 2011. The role of input frequency and semantic
transparency in the acquisition of verb meaning: Evidence from placement verbs
in Tamil and Dutch. Journal of Child Language, 38 (3). 504–532.
Nicoladis, E., Pika, S. & Marentette, P. 2009. Do French-English bilingual children
gesture more than monolingual children? Journal of Psycholinguistic Research,
38 (6). 573–585.
Rose, M. 2006. The utility of arm and hand gesture in the treatment of aphasia.
Advances in Speech-Language Pathology, 8 (2). 92–109.
Schegloff, E. 1984. On some gestures' relation to talk. In J. Atkinson & J. Heritage
(eds.), Structures of social action. Cambridge: Cambridge University Press.
266–296.
Senghas, A., Kita, S. & Özyürek, A. 2004. Children creating core properties of lang-
uage: Evidence from an emerging sign language in Nicaragua. Science, 305
(5691). 1779–1782.

48
 Language Acquisition and Use in Multilingual Contexts

Seyfeddinipur, M. 2006. Disfluency: Interrupting speech and gesture. Radboud

University, Nijmegen.
Singleton, J., Goldin-Meadow, S. & McNeill, D. 1995. The cataclysmic break bet-
ween gesticulation and sign. In K. Emmorey & J.S. Reilly (eds.), Language,
gesture, and space. Hillsdale, N. J.: Erlbaum. 287–311.
Stam, G. 2006. Thinking for Speaking about motion: L1 and L2 speech and gesture.
IRAL, 44 (2). 143–169.
Stefanini, S., Recchia, M. & Caselli, M.C. 2008. The relationship between spontan-
eous gesture production and spoken lexical ability in children with Down
syndrome in a naming task. Gesture, 8 (2). 197–218.
Willems, R., Özyürek, A. & Hagoort, P. 2007. When language meets action: The
neural integration of gesture and speech. Cerebral Cortex, 17 (10). 2322–2333.

49
Parallel grammars
On being separated by a common language

Arthur Holmer

Introduction
First language acquisition is generally understood as a process whereby a
learner is exposed to language input and on the basis of this data formulates
a set of hypotheses which eventually develop into the native grammar of the
speaker. One important consequence of this concerns the nature of the
shared grammar of two speakers within a given speech community: while it
is clear that their grammars must be capable of producing output which is
similar enough, we have no guarantee that the grammars are built up in the
same way. Strictly speaking, the internal grammar of a given speaker of a
language is (or could be) unique: we have no way of assessing in which
ways the grammar of one speaker is wired differently from that of another
speaker, as long as the output they produce is identical.
In this sense, each instance of language acquisition is a potential oppor-
tunity for mutation to occur. What, then, prevents the grammar of a given
language from spontaneously mutating into another grammar? One obvious
candidate is the sheer body of output/input. The fidelity of the reproduction
is a direct consequence of the amount and quality of input data available.
Thus, while output->input may be a fairly efficient way of transferring a core
grammar (i.e. structures highly frequent in the input), once we get outside
this domain, the limits of what is marginal and ungrammatical may mutate
spontaneously.
When this occurs on a large scale, the process is referred to as reanalysis,
a major factor in language change. One dramatic example is the word order
pattern of Tongan and Niuean (Otsuka 2005, see also Massam 2005), where
almost identical bodies of data (the two languages being more or less mutu-
ally comprehensible) are analysed by dramatically different syntactic
mechanisms on the basis of certain properties of adjectives within the object
NP.
Another, possibly more striking, such example is the reanalysis of the
common Kartvelian split-ergative case alignment pattern (e.g. that found in
Georgian) into a fully accusative pattern in Megrelian and a fully ergative

50
 Language Acquisition and Use in Multilingual Contexts

pattern in Laz, despite the fact that these two variants are mutually
intelligible dialects often classified under a single language. This is dis-
cussed in detail in Harris (1985), cf. Holmer & Vamling (in preparation) for
an account of the actual process of reanalysis.
Further, given the importance of input data, it also follows that minority
languages which are under great pressure from outside, and which are not
used as the everyday means of communication in most domains of their
society, would also be more likely to undergo spontaneous mutation and
thereby diachronic change. We will see evidence of this in the next section.
First, however, let us examine a case where parallel grammars are
affecting the margins of grammaticality: word order acceptability judgments
in Swedish subordinate clauses without complementizers.

Parallel grammars

Subordinator omission in Swedish

In Swedish, as in the other mainland Scandinavian languages, subordinate
clauses differ from main clauses in having a recognizable word order
pattern. Thus, while main clauses display verb-second ordering, and
(crucially here) with the verb preceding any sentence adverbs (1a), sub-
ordinate clauses (prescriptively at least) do not allow V2 and display an
ordering with the verb preceding any sentence adverbs (1b, c).
(1) a. Josefin vill inte köpa boken.
Josefin wants NEG buy book.DET5
‘Josefin doesn't want to buy the book.’
b. Jag tror att Josefin inte vill köpa boken.
1S believe COMP Josefin NEG wants buy book.DET
‘I believe Josefin doesn't want to buy the book.‘
c. *Jag tror att Josefin vill inte köpa boken.
1S believe COMP Josefin wants NEG buy book.DET

5
Abbreviations used in this paper are the following. ACC=accusative; AUX=auxiliary;
COMP=complementizer; DEM=demonstrative; DET=determiner; E=ergative; GEN=genitive;
LF=Locative Focus (or Locative voice); LOC=locative; M=masculine; F=feminine;
NEG=negation; NOM=nominative; PF=Patient Focus (or Patient voice); PRF=perfective;
PST=past tense; REL=relativizer; S=singular. Further, common conventions used include: * =
ungrammatical; and % = acceptable for some speakers.

51
Language Acquisition and Use in Multilingual Contexts

The next salient fact about Swedish word order is that the complementizer
att can readily be omitted, with no grammaticality consequences (2). All
speakers of Swedish agree that example (2) is fully acceptable.
(2) Jag tror Josefin vill köpa böcker.
1S think Josefin wants buy books
‘I think that Josefin wants to buy books.’

It is when these two facts are combined that an interesting phenomenon

occurs. In (2), the bridge verb tror ‘believes’ is followed by a string which
linearly speaking could be either a main clause or a subordinate clause. Since
there is no sentence adverb, we cannot decide which structure the string
actually has. However, if we insert a sentence adverb, such as the negation
inte ‘not’, the situation changes dramatically. Neither order is fully accep-
table to all speakers (3a, b). When asked to negate an example like (2), many
speakers spontaneously offer (3c) as an alternative, where the negation is
realized in the matrix clause instead of in the embedded clause.
(3) a. %Jag tror [Josefin inte vill köpa böcker].
1S think Josefin NEG wants buy books
‘I think that Josefin doesn’t want to buy books.’
b. %Jag tror [Josefin vill inte köpa böcker].
1S think Josefin wants NEG buy books
‘I think that Josefin doesn’t want to buy books.’
c. Jag tror inte [Josefin vill köpa böcker].
1S think NEG Josefin wants buy books
‘I think that Josefin doesn’t want to buy books.’

It should be noted that neither (3a) nor (3b) are sharply ungrammatical.
Rather, both can be characterized as awkward. In a study conducted in 2003
(results published in Holmer 2006) the following acceptability percentages
were recorded (4). Notice that while both the negation and the adverb ofta
‘often’ have similar effects, the percentages are not the same. In particular
the acceptability of embedded main clause word order with ofta is signi-
ficantly higher than with the negation (presumably because the alternative,
raising the adverbial to the matrix clause, is not available for ‘often’ as it is
for the negation).
(4) Subordinate clause cues: NEG-V 83%
ofta-V 86%
Main clause cues: V-NEG 46%
V-ofta 69%

52
 Language Acquisition and Use in Multilingual Contexts

This varying degree of acceptability can be accounted for by positing that

each speaker has a different grammar. Given that the input on which acqui-
sition of negated subordinate clauses is based most commonly consists of
clauses with an overt subordinator, and with matrix negation (3c) as a viable
alternative, it follows that the learner can derive two conflicting grammars
from this data. These are exemplified in (5).
(5) a. postverbal COMP att can be freely realized as Ø
b. omission of att is a signal of main clause status

Both of these alternatives are equally compatible with the input data. For
speakers applying (5a), example (2) above is a subordinate clause which
happens to have a Ø-complementizer which is a variant of att. In contrast,
for speakers applying (5b), example (2) is structurally speaking a main
clause which happens to be used in an embedded context. Naturally,
speakers of each variant will have a certain degree of tolerance towards the
other variant, explaining why the total sum of both orders exceeds 100%.
Furthermore, both of these alternatives are reasonable guesses, given the
input: alternative (5a) is the simplest, but alternative (5b) could easily be an
overgeneralization from omission of om ‘if’ in asyndetic conditionals (cf. 6a,
b), which force the embedded verb to raise to pre-subject position (or, struc-
turally speaking, to C°6, as would be the case in a main clause).
(6) a. Om du sjunger så går jag.
if 2S sing then go 1S
‘If you sing, then I'll leave.’
b. Sjunger du så går jag.
sing 2S then leave 1S
‘If you sing, then I'll leave.’

Further, the grammar of (5b) is exactly that found in German. If the com-
plementizer dass ‘that’ is omitted, the result is obligatory main clause word
order (7a – b).
(7) a. Ich glaube *(dass) er krank ist.
1S believe COMP 3S.M sick is
‘I believe that he is sick.’
b. Ich glaube er ist krank.
1S believe 3S.M is sick
‘I believe that he is sick.’

6
i.e. the structural position of complementizer head, which is reserved for subordinators in
subordinate clauses, but is generallly seen (in cartographic gererative analyses) as being the
landing site of the finite verb in verb-second constructions.

53
Language Acquisition and Use in Multilingual Contexts

Although both (5a) and (5b) are equally natural guesses from the input data,
and are compatible with the same data in most common situations, they
make conflicting predictions when it comes to negated embedded contexts
with att-omission. And it is exactly in these cases where intuitions diverge.
These divergent intuitions are clear evidence that the internal grammars of
Swedish speakers can differ with respect to the omission of att. At the same
time, we are clearly dealing with a single speech community (there is no
dialectal or generational patterning in the variation). Crucially, in most cases,
except the easily avoidable constructions in (3a, b), the output of each
grammar is fully compatible with the output of the other, and therefore the
unity of the interpersonal grammar of Swedish is preserved despite the
underlying differences.
It cannot be excluded that the variation represents a temporary instability
which may be resolved over time, with one grammar becoming entirely
dominant over the other. However, we have no evidence that this is the case.
What we do see is a synchronic state of affairs with a functioning shared
grammar based on (at least) two different internal grammars. As far as we
can know, this may well be the norm. It will be shown in what follows that
such variation can also, given the right conditions, migrate from the margins
of grammaticality to the core of the grammar. This is the case in Seediq.

Prenominal relativization in Seediq

Tgdaya Seediq is an Austronesian language spoken in the mountains of
north-central Taiwan. The data discussed here comes from recordings and
interviews conducted in the village of Gluban (清流/ Chingliu in Chinese) in
the Kuohsing valley north of Puli. Like most other aboriginal languages of
Taiwan (henceforth referred to as Formosan languages), Seediq is verb-
initial, displaying VOS order with full NP arguments. As is to be expected
from a verb-initial language, the unmarked relativization pattern is N-Rel
(8a). Nevertheless, this is not the only pattern. Seediq also has an alternative
word order pattern where the relative clause is split across the N, with the
verb preceding the N, while the remainder of the relative clause follows the
noun (8b). This construction is mentioned by Zhang (2000) and discussed by
Aldridge (2004), both referring to it as a head-internal relativization
construction. Finally, a third construction (Rel-N) is also to be found, but
this construction is not accepted by all speakers (8c).
(8) a. Wada=mu sbet-un ka huling.
AUX=1S.E beat-PF NOM dog
[t-n-bug-an=na Watan].
<PST>rear-LF=3S.E Watan
‘I beat the dog which Watan bred.’

54
 Language Acquisition and Use in Multilingual Contexts

b. Wada=mu sbet-un ka.

AUX=1S.E beat-PF NOM
[t-n-bug-an=na huling na Watan].
<PST>rear-LF=3S.E dog GEN Watan
‘I beat the dog which Watan bred.’
c. %Wada=mu sbet-un ka.
AUX=1S.E beat-PF NOM
[t-n-bug-an=na Watan] huling.
<PST>rear-LF=3S.E Watan dog
‘I beat the dog which Watan bred.’

There are no dialectal differences between speakers who accept (8c) and
those who do not: all the data presented here is from a single dialect spoken
in a single village by a close-knit speech community (in fact, in some cases
even members of the same family may pattern differently). Further, there
does not seem to be a generational difference either (both the youngest and
the oldest speaker consulted belong to one group). Finally, all speakers
interviewed are bilingual, some with Seediq and Mandarin, others with
Seediq and Japanese, and some with all three. There is no correlation
between the bilingualism pattern and syntactic preference in this respect.
For this reason, we refer to the two groups of speakers as Tgdaya A and
Tgdaya B respectively. Tgdaya A (which does not allow Rel-N) is the
variant spoken by four of the speakers I have consulted with respect to this
issue, while Tgdaya B (which does allow Rel-N) is the variant described by
Zhang (2000) and Aldridge (2004), and which was spoken by three of the
speakers I have consulted (including the oldest and the youngest of my
consultants).
The Rel-N pattern exemplified in (8c) is typologically unexpected in a
verb-initial language. Nevertheless, it is common across the Formosan
languages. Thus, most Formosan languages have Rel-N as an option (9a)
beside Rel-N (9b), and at least Tsou and Bunun (9c) have Rel-N as the
unmarked norm.
(9) a. ulaya isuwa idru na ku=tinepuk-an na suwan?
LOC where DEM DET 1S=beaten-LF DET dog
‘Where’s the dog I beat?’ Puyuma
b. ulaya isuwa idru na suwan na ku=tinepuk?
LOC where DEM DET dog DET 1S=beaten
‘Where’s the dog I beat?’ Puyuma
c. mundaan ca [mina'u-s hutan a] uva'az.
left NOM ate-ACC sweet.potato REL child
‘The child that ate sweet potatoes left.’ Takituduh Bunun

55
Language Acquisition and Use in Multilingual Contexts

It is hypothetically possible that Tgdaya B may have evolved in analogy

with other Formosan languages (in particular perhaps Takituduh Bunun,
which is spoken in an area adjacent to Seediq). However, both varieties of
Tgdaya Seediq are spoken by close relatives, and we have no other evidence
of any Bunun-like properties in Seediq (such as Bunun VSO word order,
which would be sharply ungrammatical in Seediq). These facts seem to
suggest spontaneous mutation as a more likely explanation than borrowing.
In the following I will present a possible hypothetical scenario for such a
mutation, consistent with the data and with other known facts about the
structure of Seediq. It should however be noted that we have no way of
knowing if this account is historically true.
Crucially, both groups of Tgdaya speakers have access to the typo-
logically expected N-Rel construction (10a). However, as befits a language
with clitic pronouns, most spontaneous discourse tends to avoid large-scale
use of full NPs. Therefore, what is usually found in discourse is the reduced
relative clause in (10b). Further, as in several other Formosan languages,
word order can be used to represent information structure, in that new
information precedes old information. Thus, parallel to (10b), there exists an
alternative with a restrictive reading (10c).7
(10) a. Wada=mu gguy-un ka nasi.
PST=1S.E steal-PF NOM pear
[l<n>amu=na baki].
<PF.PRF>pick=3S.E old.man.
‘I stole the pears which the old man picked.’
b. Wada=mu gguy-un ka nasi [l<n>amu=na].
PST=1S.E steal-PF NOM pear <PF.PRF>pick=3S.E.
‘I stole the pears which he picked.’
c. Wada=mu gguy-un ka [l<n>amu=na] nasi.
PST=1S.E steal-PF NOM <PF.PRF>pick=3S.E pear
‘I stole the pears which he picked (not those from the tree).’

That (10b) is a reduced variant of (10a) is hardly a challenge for the learner
to work out. She may, on the other hand, have greater problems with (10c).
Here several competing hypotheses are possible. If she has already become
acquainted with the split (or internally headed) construction (11a), cf. also
(8b), the null hypothesis would be to analyse (10c) as a reduced variant of
(11a), analogous to the reduction which derives (10b) from (10a).
If, however, she has not encountered (11a), which is not surprising given
its relatively low frequency, the null hypothesis is quite different. Again in
analogy with the reduction deriving (10b) from (10a), where the position of
7
Presumably derived by leftward (head-) movement of the verb past the noun.

56
 Language Acquisition and Use in Multilingual Contexts

the verb indicates the underlying position of the entire relative clause, the
null hypothesis would now be that a preverbal reduced relative clause is
simply a reduced version of a preverbal relative clause (11b). In this way, the
learner hypothesizes that (11b) is a possible structure in Tgdaya Seediq.
(11) a. Wada=mu gguy-un ka.
PST=1S.E steal-PF NOM
[l<n>amu=na nasi na baki].
<PF.PRF>pick=3S.E pear GEN old.man
‘I stole the pears which the old man picked.’
b. %Wada=mu gguy-un ka.
PST=1S.E steal-PF NOM
[l<n>amu=na na baki] nasi.
<PF.PRF>pick=3S.E GEN old.man pear
‘I stole the pears which the old man picked.’

From the above, we can see how prenominal relative clauses can be acquired
spontaneously without necessarily being part of the input. Once such a struc-
ture is acquired, moreover, the learner never sees any evidence against the
existence of such a structure, since negative evidence is not part of the
equation. Instead, there is more circumstantial evidence which actually
supports the reanalysis in (11b).
The most common structure for relativizations is that where the overt
arguments of the relative clause are clitics, while full NPs, if used in
discourse, are usually introduced first, as in (12a). This construction, which
is universally accepted, is not a relativization at all, but simply two apposed
main clauses, where one is serving as the topic of the other, and where the
quantifier kana ‘all’ in clause-final subject position is resumptive for the
patient. However, given the optionality of the NOM marker ka, this structure
is linearly ambiguous, and can easily be reanalysed as a topicalized pre-
nominal relative clause. When linking the two clauses, some Tgdaya A
speakers take the entire fronted clause and realize it as the predicate of a
cleft construction (12b). This construction is presumably what most accura-
tely merits the term “internally headed relative clause”.
(12) a. L<n>amu=na baki (ka) nasi.
<PF.PRF>pick=3S.E old.man NOM pear
wada=mu gguy-un kana.
PST=1S.E steal-PF all
‘The old man picked the pears, I stole them all.’

57
Language Acquisition and Use in Multilingual Contexts

b. [L<n>amu=na baki ka nasi].

<PF.PRF>pick=3S.E old.man NOM pear
ka wada=mu gguy-un.
NOM PST=1S.E steal-PF
‘I stole the pears that the old man picked.’
(lit. ‘The old man picked the pears is what I stole’)

Thus, even for speakers of Tgdaya A, apparent prenominal relative clauses

do occur, but only in fronted position. However, for speakers of Tgdaya B,
this is simply a clear instance of a prenominal relative clause, which corro-
borates their original hypothesis.
In this way, several factors conspire to enable prenominal relative clauses
to emerge as the null hypothesis based on data which does not directly
include them. All of these factors are available to all speakers, but it is sim-
ply a coincidence which hypothesis a given speaker chooses. Both grammars
can capture the same input, and usually produce the same output, but directly
affect acceptability judgments. This is exactly what we find.
There is one important difference between the structures in Tgdaya A and
those in Tgdaya B, however. The output of Tgdaya A is perfectly compatible
with the structure of Tgdaya B (from the point of view of a speaker of
Tgdaya B, a Tgdaya A speaker is simply not making use of the entire spect-
rum offered by the grammar). In contrast, the converse is not true. The
grammar of Tgdaya B allows prenominal relativization and can be expected
to produce such constructions as its output as well. Once such constructions
become part of the input for the next generation, the only possible hypothesis
will be that prenominal relative clauses are part of the grammar. And of
course, through this process, they become part of the grammar.
It follows, therefore, that the grammar of Tgdaya B is dominant and can
be expected to expand in the long term at the expense of Tgdaya A. Re-
analysis can only go from Tgdaya A to Tgdaya B, not vice versa. At some
point in time, if the language were to develop with no external interference
in an ideal situation, all speakers of Tgdaya Seediq would have grammars
where both N-Rel and Rel-N coexist.
In this respect, the situation in Seediq relativization differs substantially
from that in Swedish complementation: the Swedish parallel grammars
presumably represent a stable system where the output of neither grammar
challenges the well-formedness conditions of the other. In contrast, the
Tgdaya Seediq parallel grammars presumably represent a temporary
instability which can only be resolved in one direction: by the univers-
alization of the Tgdaya B grammar. This is, given time and the survival of
the Seediq language, a falsifiable prediction of the proposed hypothesis.

58
 Language Acquisition and Use in Multilingual Contexts

Conclusion
In this paper, I have argued that the homogeneous shared language of a
speech community need not reflect a set of identical internal grammars on
the part of the speakers, but rather that it is quite likely that the internal
grammars of the speakers may be radically different, although the output
they produce is largely identical. One prediction of this claim is that differ-
ences between the parallel grammars of different speakers should be
noticeable along the margins of grammaticality, even when the core gram-
mar is identical on the surface. This prediction seems to be borne out by the
data presented here, which illustrates two types of parallel grammars: 1)
coexisting parallel grammars which may be in a stable balance, i.e. when the
two grammars produce mutually marginal output; 2) coexisting parallel
grammars which are inherently unstable, i.e. when one grammar produces
output which is ungrammatical according to the other grammar, but not vice
versa (in which case we expect the more liberal grammar to produce output
which would contradict the hypotheses underlying the more restrictive
grammar, undermining these in the next generation).
While we cannot quantify this, it is to be assumed that the amount of
input a learner is exposed to may have some effect, and we therefore expect
a greater amount of instability in endangered minority languages than in
languages which are the medium of instruction in schools.

References
Aldridge, E. 2004. Internally headed relative clauses in Austronesian languages.
Language and Linguistics 5 (1): 99–130.
Harris, A. 1985. Diachronic syntax: The Kartvelian case. Syntax & Semantics, 18.
Orlando: Academic Press.
Holmer, A. 2006. The place of Swedish in word order typology. I Å. Viberg (ed.).
The typological profile of Swedish. Special issue of Sprachtypologie und
Universalienforschung, 59. 76–102.
Holmer, A. & Vamling, K. In preparation. The parameters of Kartvelian ergativity.
Massam, D. 2005. Lexical categories, lack of inflectrion and predicate fronting in
Niuean. In A. Carnie, H. Harley & S. Dooley, (eds.). Verb First. On the syntax of
verb-initial languages. Linguistik Aktuell 73. Amsterdam: Benjamins. 227–242.
Otsuka, Y. 2005. Two derivations of VSO: A comparative study of Niuean and
Tongan. In A. Carnie, H. Harley & S. Dooley, (eds.). Verb First. On the syntax
of verb-initial languages. Linguistik Aktuell 73. Amsterdam: Benjamins. 65–90.
Zhang Y-L. 2000. Saideke-yu cankao-yufa. [Reference grammar of Seediq]. Taipei:
Yuanliou.

59
Acquisition of prosody
Word accents, phrasing, and morphosyntax in a Swedish-English
bilingual child at 30–32 months of age

Merle Horne

Introduction
Gisela Håkansson has made many important contributions to the study of
language acquisition in both monolinguals and bilinguals (e.g. Håkansson
1998, 2003). The acquisition of grammatical morphology has been an area of
particular interest for her, in particular verb morphology and the relation
between tense and verb-second word order. The present contribution
attempts to illustrate the intimate relation between the acquisition of morph-
ology, syntax and prosody using material from a bilingual child.
Since inflectional and derivational morphology is intimately associated
with the distribution of word accents in Swedish (Bruce 1977; Riad 2012),
the acquisition of prosodic patterns can be assumed to go hand in hand with
the development of the lexicon. Moreover, since prosodic phrasing and syn-
tactic phrasing are closely related (e.g. Selkirk 2000), the acquisition of pro-
sodic phrasing can be expected to develop as postlexical and syntactic struc-
tures are learned.
Peters and Strömqvist (1996) have shown how the acquisition of morph-
ology in a monolingual child is tied to the development of Accent 1 and
Accent 2. The alternation between Accent 1 and Accent 2 in words with the
same stem does not occur until the definite and plural suffixes in nouns (e.g.
bilen1 ‘the car’ – bilar2 ‘cars’) and verb inflections (kommer1 ‘comes’–
komma2 ‘to come’) are learned (Accent 1 and Accent 2 will be represented
with the subscripts 1 and 2). The present contribution aims at presenting data
from a bilingual child that further illustrates the dynamic period in acq-
uisition around 2.5 years of age where alternation between Accent 1 and
Accent 2 in nominal and verbal morphology is well on the way to being
mastered by the child. The bilingual data provide clues to the child’s on-line
processing of words into stem+affix by e.g. the use of Swedish affixes on
English stems, as well as in the generalization of regular affixes to irregular
stems. The understanding by the child of the affix/word accent relation at

60
 Language Acquisition and Use in Multilingual Contexts

30–32 months of age will be seen to be manifested not only by the use of
correct alternations between Accent 1 and Accent 2, but also by meta-
linguistic reasoning which shows hypothesis-testing regarding grammatical
affix/word accent mapping. An interesting phenomenon in the material is the
prosodic status of the infinitive form of the verb, which appears to have an
Accent 1-like form at this stage. Although the acquisition of word-accent
distribution has come a long way at 32 months, the material seems to
indicate that prosodic phrasing, i.e. the grouping of word-accents and bound-
ary tones into intonational patterns that are found in the adult language (see
e.g. Horne 1994; Roll 2009; Myrberg 2010) is something that is still not at
the target level at this early stage of development.

The bilingual child

The material in the present study comes from recordings of a boy (whom we
will refer to as “Jesper”) between the ages of 30 and 32 months. He is a
simultaneous bilingual speaker (Viberg 1987), i.e. he was exposed to both
English and Swedish (central Swedish variety) in the home. Swedish was,
however, the dominant language (Arnberg 1981; Håkansson 1998), i.e. the
language most often heard and used with playmates, Swedish relatives,
nanny, friends of the family and was the language most often used by the
boy in conversations. The data used here (approximately 250 Swedish
utterances) were recorded at home in just such bilingual interactions.
Although Jesper grew up in southern Sweden, he acquired a central variety
of Swedish prosody, most likely modelled after the prosody of his father. We
will assume therefore that the prosodic patterns that Jesper heard in his
father’s speech in the home served as the main input for the prosody model
that he developed.

Swedish word accents

In Swedish, words are associated with one of two word accents (Bruce 1977,
1986). In Central Swedish, which can be assumed to be Jesper’s prosodic
model, Accent 1 is characterized by a low tone (L*) at the beginning of the
stressed vowel, whereas Accent 2 is realized with a high tone (H*) at the
beginning of the stressed vowel. This contrast is illustrated in Figure 1 with
examples produced by Jesper’s father.
According to Bruce (1977), the difference between Accent 1 and Accent 2
involves a difference in the timing of the same basic tonal contour with the
segmental string (see Figure 2). Given this interpretation, it could be thought
that one accent could be derived from the other on the basis of different
timing of the tones with the segmental sting. As has been shown by Peters &
Strömqvist (1996), Accent 2 is acquired first by monolingual children.

61
Language Acquisition and Use in Multilingual Contexts

Figure 1. Left: F0-contour and waveform for the Accent 1 word anden ’the duck’.
Right: F0-contour and waveform for the Accent 2 word anden ’the spirit’.

This being the case, it could then be thought that Accent 1 could possibly be
acquired when the child perceives the crucial timing difference with respect
to the stressed vowel and adjusts the tonal contour so as to associate it earlier
with respect to the segmental string. The present data provide some indi-
cation that this could possibly be the case (see in particular Figure 5), but
more extensive investigations are necessary in order to determine the mech-
anisms involved in learning the timing distinction between the two word
accents. Perception and production of the prominent focal accent (“sentence
accent rise” in Bruce 1977 (see Figure 2)) is mastered very early, but the
timing difference of the word accent associated with the stressed syllable no
doubt becomes finer tuned when the acquisition of grammatical affixes
begins, since children then are focused on the form of the affixes and per-
ceive that different suffixes are correlated with different word accents on the
same stem, e.g. boll+en1 ‘the ball’, boll+ar2 ‘balls’; klipp+er1 ‘cuts’,
klipp+te1 ‘cut (past tense)’.

Figure 2. Timing difference of same basic tonal pattern in the realization of Accent 1
and Accent 2 in relation to the CV-tier. Note the low tone at the beginning of the
stressed vowel (‫׀‬V) in Accent 1 and the high tone at the beginning of the stressed
vowel in Accent 2 (adapted from Bruce 1977 and Bruce 1986).

62
 Language Acquisition and Use in Multilingual Contexts

When are word accents acquired?

According to Plunkett & Strömqvist (1992), the word accent distinction is
not mastered completely until around the age of 4 years. Accent 2 is ac-
quired first (often before 2 years of age) in Central Swedish, and this has
been assumed to be due to its perceptual salience (Peters & Strömqvist
1996). Accent 2 is realized on words with at least two syllables and the two
high peaks in prominent (focused) Accent 2 words (most often utterance-
final) make them very salient for the child. Monosyllables always have
Accent 1. Due to tonal crowding, Accent 1 monosyllables in utterance-final
position are sometimes realized with a tonal contour that resembles a non-
prominent Accent 2 (i.e. H+L a prominent “sentence accent” or focal high
tone followed by a low boundary tone) as is illustrated in Figure 3.

Figure 3. Left: F0-contour and spectrogram for an Accent 1 word (mus ’mouse’)
realized in utterance final position by Jesper’s father. Notice that the second half of
the vowel (-u-) is associated with a ”sentence accent”/focal H-tone and a final L
boundary tone. At an early age, this H+L pattern could perhaps be interpreted by the
child as the same pattern as an Accent 2 (non-prominent) word accent. Right: Same
F0 pattern produced on mus ’mouse’ realized in utterance-final position by Jesper.

In the language acquisition process, Accent 2 is early generalized to Accent

1 words. There are many reasons for this. One reason is no doubt its salience
(Peters & Strömqvist 1996). Another is the frequency of a phonetic H+L
contour in utterance-final (“sentence accent”) position both on prominent
(focused) Accent 2 words and monosyllabic Accent 1 words (see Figure 2)
boundary tones). A further contributing factor is that frequent Accent 1
words are often monosyllabic modal or base verbs (är ‘is’, har ‘have’, vill
‘want’, kan ‘can’, ska ‘must’) that occur in phrase-internal position where
they are non-prominent or deaccented (Riad 2012). Thus the perceptual
salience of Accent 1 words in utterance-internal position is considerably
lower than that of Accent 2 words, thus making it no doubt more difficult for
the child to determine their target form. At 30–32 months of age, however,
when the acquisition of nominal and verbal morphology is well underway, as

63
Language Acquisition and Use in Multilingual Contexts

is the acquisition of different syntactic structures, the recordings used in the

present study show evidence that Jesper is well underway to mastering the
mapping of word accents to different morphosyntactic structures.

Morphosyntax at 30–32 months

The recordings reveal that Jesper at 30–32 months had reached a stage in his
acquisition of Swedish which seems to be comparable to that of other
children at the same age (see e.g. Plunkett & Strömqvist 1992). As regards
noun morphology, Jesper used the singular definite suffixes -en and -et, as
well as the plural suffixes -ar, -er, and -or. The processing of the plural
suffix -ar as an independent morpheme is observed for example in the
extension of the ending to form the plural of irregular forms such as hjul
‘wheel(s)’ > hjul+ar. The singular definite ending -en is also used on English
words, e.g. pencil+en ‘the pencil’, piano-n+en ‘the piano’ (with insertion of
an intervocalic -n-).
As regards verb morphology, present tense forms from different declen-
sions are common at 30–32 months: komm+er ’come(s)’, raml+ar ’fall(s)’,
ork+ar ’manage(s)’, blöd+er ’bleed(s)’, rid+er ’ride(s)’, prat+ar ’speaks’,
läs+er ’read(s)’, bo+r ’live(s)’, hö+r ’hear(s)’. Imperative forms also occur,
e.g. hämt+a ’fetch’, titt+a ‘look’, in addition to infinitive forms. Future
tense forms using kommer att ’going to’ also occur, e.g. Du kommer inte att
lyssna på Jesper ’You are not going to listen to Jesper’.
Above the lexical level, Jesper is observed to produce many different
constructions with modal verbs + infinitive, e.g. vill + infinitive: Jag vill rita
en segelbåt ’I want to draw a sailboat’; måste + infinitive: Jag måste betala
’I have to pay’; kan + infinitive: Du kan ringa ’You can call’; få + infinitive:
Du får låna den boken ’You can borrow that book’. Even constructions with
preplaced adverbs and non-canonical (Adv V S (O)) word order are
observed, e.g. Så/Nu kommer jultomten ’So/Now Santa Claus is coming’; Då
har han presenter ’Then he has presents’; Där bor farmor och farfar
’Grandmother and grandfather live there’. Syntactic constructions involving
topicalization are also produced at 30 months: Nej, det gör pappa ’No,
Daddy does that’; En sån ska jag ha. ’I want one of those’. Even question
word order is produced: Kan vi släppa ut honom? ’Can we let him (=Santa
Claus) out?’; Finns det flera här? ’Are there several here?’
The recordings from 32 months contain relative clauses, e.g. Vad är det
som fastnar där? ’What is it that gets stuck there?’ as well as subordinate
adverbial clauses, e.g. När jag fyller år då ska jag gå i skolan ’When I have
my birthday, then I will go to school’. Productive compounding, thought to
be correlated to the acquisition of relative clauses (Plunkett & Strömqvist
1992) is also observed at 32 months, e.g. elefantpresent ’elephant-present’,
i.e. a present that is an elephant.

64
 Language Acquisition and Use in Multilingual Contexts

Prosody at 30–32 months

The use of inflectional morphology leads to morphophonological alternation
between Accent 1 and Accent 2: bil+en1 ’car +def.sg.’, bil+ar2 ’car+s (pl)’,
komm+er1 ’come+s’ – komm+a2 ’come+inf’. Therefore, at 30–32 months,
one expects, and indeed finds evidence for the child’s association of the
different word accents with different grammatically inflected forms. Figure 4
shows present (gör) and preterit (gjorde) forms of the verb göra ’to do’
associated with Accent 1 and Accent 2, respectively.
In addition to the spontaneous production of morphophonological alter-
nation between the two word accents, additional evidence for the processing
of inflectional morphemes as independent units comes from utterances
where a verb stem is mapped onto different possible suffixes and word
accent patterns. For example, on one occasion, Jesper seems to be involved
in metalinguistic reasoning, testing different possible verb forms to express
the present tense of the verb hålla ‘to hold’ and produces the utterance: Inte
hålla… håller… *hållar ‘Not to hold… holds …. holds’ with different word
accent patterns. The infinitive form hålla ‘to hold’ has an Accent 1-like
pattern with a L at the beginning of the stressed vowel. This is a common
accentual pattern for infinitives in the speech of Jesper at this age (see also
below for a further discussion of infinitives; see also Figure 5). In the present
tense form, håller ‘holds’, there is also a L in the stressed vowel, although
not as early as in the infinitive. In *hållar ‘holds’ (non-target suffix -ar) ,
there is a low tone at the beginning of the stressed syllable with a steep rise
that then falls in the stressed vowel. The suffix -ar is associated with a
prominence (focal) H tone. Thus the Accent 2-like contour on *hållar
provides evidence for the child’s association of the suffix -ar with Accent 2.

Figure 4. Left: F0-contour and waveform for the present tense form gör ’does’
(Accent 1). Right: F0-contour for past tense gjorde ’did’ (Accent 2) produced by
Jesper (30 mo.).

65
Language Acquisition and Use in Multilingual Contexts

Figure 5. Three different forms of the verb hålla ’to hold’ produced by Jesper (30
mo.) Left: hålla (infinitive) with a L tone at the beginning of the stressed vowel, and
with the preceding high in the onset of the syllable (note the voiced h); Center:
håller (present tense) with an Accent 1 contour; preceding high in the syllable onset
(voiced h); Right: *hållar (non-target suffix -ar) with Accent 2-like contour.

The earlier timing of the accentual contour on the other forms perhaps
indicates that the child associates the present tense suffix and infinitive form
with a different timing of the word accent pattern (see also Figure 2).
Evidence for the understanding of the present and preterit tense affixes as
independent morphemes is seen in their productive use to build new verbs
from e.g. interjections. For example, on the basis of the interjection Pang!
’Bang!’ (sound of a pistol), Jesper creates the forms pang+er1 ’bangs’ and
pang+de2 ’banged’ with alternation between Accent 1 in the present tense
form and Accent 2 in the preterite tense form (see Figure 6). Further evi-
dence for Jesper’s processing of verbal tense morphology is seen in the
attachment of the regular -de preterite suffix to irregular verb-stems such as
kom ’come’ > kom+de2 ’came’ with concomitant association of Accent 2 to
the stem.
Indication of the independent processing of the infinite marker -a (as well
as the singular definite suffix -en) is seen in its use together with English
lexical items, e.g. read+a in utterances containing modal verb + infinitive
constructions like Kan du reada monkeybooken? ’Can you read the monkey-
book?’ What is interesting at this stage, however, is that it is not obvious that
the infinitive marker is associated with the target Accent 2. Rather, the in-
finitive, which occurs most often in utterance internal position is most often
associated with a tonal pattern that resembles Accent 1 (see also above,
Figure 5). Perhaps this is due to the fact that infinitive forms which are often
utterance internal are often realized with Accent 1 or deaccented in adult
speech in e.g. verb-particle constructions like följa med ‘come along’, släppa
ut ‘let out’, lyssna på ‘listen to’ (see Riad 2012). Indeed, following
Christensen (2003), infinitive forms, as opposed to imperative forms, can be
regarded as accentually ”neutralized” forms of present tense forms (e.g.
följer1 ‘follows’ and ritar2 ‘draws’). In one instance in the present data, the

66
 Language Acquisition and Use in Multilingual Contexts

Figure 6. Left: F0 contour, spectogram and waveform for present tense form
pang+er1 ’bangs’ of verb panga ’to bang’ formed by Jesper from the interjection
Pang! ’Bang!’ (sound of a pistol) Right: F0 contour, spectogram and waveform for
past tense form pang+de2 ’banged’ of same verb.

present tense (följer ‘follow’) is used instead of the infinitive form (följa ‘to
follow’) in the utterance Du får följer med mig ‘You can follow me’.
Whether this has any connection with the child’s association of present and
infinitive forms is impossible to say on the basis of the limited data, but the
association of infinitive forms with Accent 1 in the present data is rather
striking.
At 32 months, prosodic phrasing is something which has not reached the
adult norm. Speech rate is also slower than the adult target (ca. 3–4 syllables/
second) in relation to the adult target (ca. 6–7 syllables/second) in spon-
taneous speech. Prosodic phrases thus contain fewer words than in adult
speech. While right-edge prosody is very adult-like, utterance-internal pro-
sody is still under development. This can be seen quite clearly, e.g. in the
production of compounds. In the adult target, compounds have an accentual
pattern that corresponds to Accent 2, i.e. with a H*L on the first stressed
syllable and a prominence rise on the last stressed syllable. In the speech of
the two-year-old, the different components of compounds retain their indi-
vidual tonal patterns. In Figure 7, the difference between Jesper’s F0 contour
for the compound segelbåt ‘sailboat’ and that of his father are presented. As
can be seen, the component morphemes in Jesper’s production are both
produced with a prominent tone, i.e. both segel ‘sail’ and båt ‘boat’ are re-
alized with a focused word accent. In the adult target, on the other hand, only
the final morpheme in the compound, båt ‘boat’, is realized with a focal
accent. In the speech of the 2.5 year-old, the general impression is that the
tones are rather firmly anchored to the syllables to which they are associated.
The kinds of tonal spreading and tonal concatenation patterns that
characterize the adult norm are not something that is characteristic of speech
at this early age. Although right-edge prosody is very adult-like, utterance
initial and internal concatenation patterns (see e.g. Horne 1994; Roll 2009;
Myrberg 2010) are still not as in the target prosodic structure.

67
Language Acquisition and Use in Multilingual Contexts

Figure 7. Left: F0 and waveform for the compound segelbåt ’sailboat’ produced by
Jesper (32 mo.). Notice that both component morphemes of the compound (segel
’sail’ and båt ’boat’) are realized with a prominent tone. Right: F0 and waveform for
segelbåt produced by Jesper’s father showing a prominent (focal) Accent 2 pattern
on the whole compound.

Conclusion and questions for further study

Although the material investigated in the present study is quite limited, it
contains a good number of clues to the on-going process of acquisition of
inflectional morphology, syntax and the mapping of word-accents onto
lexical structure. The material shows that the bilingual child at 30–32
months is in the process of acquiring inflectional morphology and the
morphophonological rules for associating the word accents with different
grammatical affixes. Although the right-edge prosody is very adult-like,
utterance internal prosody has still not reached the adult target.
In order to be able to arrive at a better understanding of the course of ac-
quisition of the prosody-morphology mapping, it is necessary to conduct
more comprehensive longitudinal studies of language acquisition. Controlled
data where the same words can be elicited in different morphosyntactic
contexts is necessary in order be able to better track the acquisition of word
accents and prosodic phrasing. The present data points to the development of
the infinitive form in relation to the present tense and imperative forms as an
interesting area of research which could give us a better understanding of the
relationship between finite and non-finite forms of the verb. Another area
which would lead to more insights into the acquisition of intonation would
be a longitudinal study of the development of compound word prosody.
A further area that could be explored using neurolinguistic methods is the
question as to when word accents start being used in speech parsing. In ERP
(Event-Related Potential) studies using adults (Roll 2009; Roll et al. 2010;
Roll et al. 2013), it has been shown that word accents are used predictively
in speech processing, i.e. they are used to cue up-coming suffixes. Evidence
from the present material and other child data indicate that already at 30–32
months, children seem to be aware of the word-accent/affix relationship.
Thus neurolinguistic studies could provide evidence as to whether they are
actually used in language parsing at this age.

68
 Language Acquisition and Use in Multilingual Contexts

References
Arnberg, L. 1981. Early childhood bilingualism in the mixed-lingual family. Lin-
köping University. Linköping Studies in Education.
Bruce, G. 1977. Swedish word accents in sentence perspective. Lund: CWK
Gleerup.
Bruce, G. 1986. How floating is focal accent? In K. Gregersen & H. Bosbøll (eds.),
Nordic prosody IV. Odense University Press. 41–49.
Christensen, L. 2003. Swedish verbal morphology from the child’s point of view. In
L-O. Delsing, C. Falk, G. Josefsson & H. Sigur∂sson (eds.), Grammar in focus.
Festschrift for Christer Platzack, Vol. 2. Lund: Wallin & Dalholm. 57–63.
Håkansson, G. 1998. Språkinlärning hos barn. Lund: Studentlitteratur.
Håkansson, G. 2003. Tvåspråkighet hos barn i Sverige. Lund: Studentlitteratur.
Horne, M. 1994. Generating prosodic structure for synthesis of Swedish intonation.
Working papers. Lund University, Dept. of Linguistics and Phonetics 43. 72–75.
Myrberg, S. 2010. The intonational phonology of Stockholm Swedish. Stockholm
University, Dept. of Scandinavian Studies.
Peters, A. & Strömqvist, S. 1996. The role of prosody in the acquisition of gramm-
atical morphemes. In J.L. Morgan & K. Demuth (eds.), Signal to syntax. Boot-
strapping from speech to grammar in early acquisition. Mahwah, NJ: Lawrence
Erlbaum. 215–232.
Plunkett, K. & Strömqvist, S. 1992. The acquisition of Scandinavian languages. In
D. Slobin (ed.), The Cross-linguistic study of language acquisition, 3. 457–556.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Riad, T. 2012. Culminativity, stress and tone accent in Central Swedish. Lingua 122.
1352–1379.
Roll, M. 2009. The neurophysiology of grammatical constraints. Lund University,
Dept. of Linguistics and Phonetics.
Roll, M., Horne, M. & Lindgren, M. 2010. Word accents and morphology–ERPs of
Swedish word processing. Brain research 1330. 114–123.
Roll, M., Söderström, P. & Horne, M. 2013. Word-stem tones cue suffixes in the
brain. Brain research (in press). [Epub ahead of print: http://dx.doi.org/10.1016/
j.brainres.2013.05.013]
Selkirk, E. 2000. The interaction of constraints on prosodic phrasing. In M. Horne
(ed.), Prosody: theory and experiment. Studies presented to Gösta Bruce.
Dordrecht: Kluwer. 231–261.
Viberg, Å. 1987. Vägen till ett nytt språk. Stockholm: Natur och Kultur.

69
Writing – it’s like learning a new
language!

Victoria Johansson

Introduction
When young children learn to write, they discover that it is impossible to
”translate” language as they know it – i.e. speech – directly into writing.
This paper discusses how characteristics of spoken language influence the
early stages of writing, and proposes that children must learn a new language
when they start to write. The study is qualitative, based on examples from a
Swedish developmental corpus of cross-sectional data in speech and writing.
Although Swedish data are used, the examples serve to illustrate a more
general discussion about what happens when a child acquire the skills of
writing. This paper mainly addresses the learning of writing that takes place
in preschool and school, and concerns a phonographic alphabet.
The findings should primarily be seen as suggestions of areas that may be
of interest for a more systematic investigation in the future.

Speech and writing: some differences

The differences between speech and writing will only be briefly outlined
here. A good description can be found e.g. in Chafe & Tannen (1987).
It obviously takes much longer to express a written message than a
spoken one. The more time-consuming process of writing thus involves
more effort, which might explain why the (young) writer does not manage
everything in writing that she masters in speech. A second difference is that
language is visible in writing. This makes it possible to edit the message
before it reaches the reader. This is also something that the child who learns
to write has not encountered before, since in speech, the text is finished once
and for all – no one requires you to go through your spoken message again.
The visibility also makes it possible to study relations in language that the
child has not previously noticed: for instance spelling conventions can reveal
morphological patterns. The third main difference is that the reader is not
present during the writing session. The prototypical speaking situation is the

70
 Language Acquisition and Use in Multilingual Contexts

conversation (Chafe 1992), and speech is produced in collaboration between

speaker and writer (Clark 1996). The speaker gets support by the listener;
her reactions tells the speaker if the message is understood. The listener can
also ask questions, thereby helping a young and inexperienced speaker in
structuring the message. The writer has to make assumptions about what the
reader already knows and needs to know in order to understand the written
text.
These three fundamental differences between writing and speaking
explain many of the features characteristic of the inexperienced writer, as we
will see below.

Data
The examples in this paper come from a corpus of spoken and written data,
collected with the purpose to investigate the development of both speech and
writing during the school ages (Johansson 2009). Data consist of 316 texts,
produced by 79 participants from four different age groups: 10-, 13- and 17-
year-olds, and adult university students. Every participant produced four
texts: one spoken and one written narrative, as well as one spoken and one
written expository text. All participants had Swedish as mother tongue, and
none of the participants had any known reading or writing difficulties. The
data were recorded individually in a lab environment. The spoken texts were
videotaped, and the written ones were recorded with a keystroke logging
program that saves information of writing activites such as pausing and
editing that takes place during the text production. The pausing and editing
data are not analyzed in this paper. The data collection was part of the inter-
national Spencer project (described in Berman & Verhoeven 2002). The data
are distributed across age groups, modalities and genres as shown in Table 1.
Table 1. Distribution of the 316 texts in the study. Number of texts in each text type.
Text type 10-year-olds 13-year-olds 17-year-olds Adults
Narrative spoken 20 20 20 19
Narrative written 20 20 20 19
Expository spoken 20 20 20 19
Expository written 20 20 20 19

Elicitation
All texts were elicited by a short wordless video, picturing problems in a
school setting (e.g. children cheating during a test; a girl stealing money that
a woman dropped; a boy who destroys a public telephone where his money
got stuck; two girls who leave when another girl approaches and wants to
talk).

71
Language Acquisition and Use in Multilingual Contexts

The narrative texts were elicited with the following instruction: Jag skulle
vilja att du berättar om ett tillfälle antingen där du räddade någon ur en
knipa, eller där någon räddade dig ur en knipa. (‘I would like you to tell
about one time either when you rescued somebody from a predicament, or
when somebody rescued you from a predicament’). The expository texts
were elicited with the following: Du såg nyss en video som visade människor
som hade hamnat i olika slags svårigheter, människor som hade problem av
olika slag. Jag skulle vilja att du skriver en uppsats/håller ett föredrag där
du diskuterar sådana situationer som människorna i filmen hade hamnat i.
(’You just watched a video showing people in different kinds of troubles.
Now I would like you to write an essay or give a speech where you discuss
the kind of situations that the people in the film got into.’).
Half of the participants produced the narrative texts first, the other half
started with the expository texts. Also, half of the participants performed the
spoken tasks first, half the written tasks. In the following, the order
differences will not be discussed, although we can expect that this has
certain effects, particularly on the spoken texts; both Strömqvist (1996) and
Johansson (2009) show order effects, where the spoken texts are influenced
in structure and lexical variation if they are preceded by a written text on the
same content. Genre differences will only be briefly touched upon here,
although the younger age groups’ competence in the expository genre differ
from the adults (cf. Berman & Verhoeven 2002).
Both written and spoken texts were transcribed according to the CHAT
format (McWhinney 2000). However, to enhance comprehension, the tran-
scriptions of the spoken examples are adapted to the orthographic standard,
except when an unorthodox written form is used as an example in itself.

Transition from speech to writing

The errors that the child makes when learning to write can be explained by
the fact that the writing is not merely a translation of the spoken language,
but consists of different rules on all levels: phonological/graphemic, morph-
ological, lexical, syntactical, and pragmatic. I illustrate this with Swedish
examples from the data and provide an English translation, as well as in-
formation on what type of text they were extracted from. For the purpose of
possible later identification, the unique text code (e.g. [wg18mDNW]) is also
indicated.

Phoneme – grapheme relation

In many languages the phoneme – grapheme relation between speech and
writing is not a one-to-one relation; not all sounds in speech are represented
in writing, and not all letters are represented in speech. This has many

72
 Language Acquisition and Use in Multilingual Contexts

reasons, but one has to do with conservative spelling conventions, which can
be the result of historical development, where a specific spelling reflects a
previous pronunciation (such as the spelling of the Swedish word skjorta,
’shirt’, where the initial sounds historically were pronounced /skj/ but later
palatized to /ʃ/), or have morphological reasons (such as snyggt (’good-
looking’), pronounced with a /kt/, which is the neuter form of the adjective
snygg). The double gg in snyggt is not necessary for the pronunciation, but
makes the connection between the two forms of the adjective visible. This
connection is kept in writing only, since the spoken forms differ.
They boy in example (1) seems not to have observed the preposition i
(’in’) in the expression i alla fall (’in any case’). Also, the boy does not seem
to acknowledge that this expression is constituted of three words/morphemes
(which according to spelling recommendations can be written as three sep-
arate words, or as one word, iallafall). It is easy to believe that the expres-
sion consist of only one word, if one has only heard it. Finally, the boy needs
to know that fall should have a double consonant at the end of the word, due
to the fact that a short vowel (often) is followed by two consonants in
Swedish. The same problem occurs in example (2), where visa (vissa, ’some
people’), gilade (gillade, ’liked’) and varan (varann, ’each other’) follow the
principle that each phoneme is matched by a grapheme, but the writer fails to
meet the criteria of double consonants after a short vowel. Sometimes the
young writer is aware of discrepancies between writing and speech, which
can lead to hypercorrection, as in example (3): *skrivigt (skrivit, ’writtend’).
(1) allafal (’(in) any case’). Written narrative, boy, 10 [wg18mDNW].
(2) visa i filmen gilade inte varan (’some people in the film did not like each
other’). Written expository, girl, 10 [wg06fBEW].
(3) och tänk om den som man skrev av hade skrivigt en jättebra dikt (’and
what if the person you copied had writtend a really good poem’). Written
expository, girl, 10 [wg06fBEW].

Different morphology
Speech and writing do not always have the same morphology. An example is
Swedish verbs belonging to the first conjugation. In writing their preterit
form ends in -de. However, in speech, this ending is often omitted, and the
word becomes homonymous with the infinitive form.
(4) så här börja det (’this is how it start’). Written narrative, boy, 10
[wg18mDNW].

In example (4), the writer has omitted the preterit ending -de of the word
börja (’start’). This is very common in speech, independent of age, as
example (5), from an adult woman, illustrates:

73
Language Acquisition and Use in Multilingual Contexts

(5) och han kom och börja prata (’and he came and started to talk’). Spoken
narrative, adult woman [wu18fDNS].

Word boundaries and formal level of lexicon

In example (1) part of the expression i alla fall was omitted in speech. Diffi-
culties in identifying the word boundaries is typical for texts from early
writing development. In example (6) the writer is ignorant of the fact that i
och för sig (here ’actually’) consists of several words, and treats it as one
word: ifösej.
(6) ifösej så hugger inte snokar i vatten (’actually grass snakes won’t bite in
water’). Written narrative, boy, 10 [wg18mDNW].

One case that illustrates a general difference between speech and writing is
the use of some common Swedish personal pronouns. In most Swedish
dialects the object forms mig (‘me’), dig (‘you’) and sig (3rd pers. reflexive
pers.pron.) are pronounced mej, dej and sej. They can also be spelled in this
way in colloquial written texts. The pronouns de (’they’) and dem (’them’)
are both pronounced dom. Table 2 shows the number of instances, and their
relative frequency (e.g. the use of mej represents 0.11 % of all written words
by the 10-year-olds) in the written data.
Table 2. Distribution of some Swedish pronouns in writing.
Word 10-year-olds 13-year-olds 17-year-olds Adults
mej 5 (0.11 %) 4 (0.05 %) 0 2 (0.01 %)
dej 0 0 0 1 (0.01 %)
sej 0 0 0 7 (0.04 %)
dom 67 (1.43 %) 103 (1.27 %) 0 8 (0.05 %)

Mej is used by 10- and 13-year-olds to a small extent. The adults occ-
asionally use the colloquial spelling, but it should be noted that, all instances
of sej and of dom are produced by the same adult. The great difference
across the age groups lies in the use of dom, which is often used by the 10-
and 13-year-olds, but hardly at all by the other groups. One reason why the
children write dom might be that this is a solution to the problem of choosing
between de and dem – something that is required only in formal writing.
Another example is shown in Table 3, which displays the distribution of
the two words, ju and liksom. Ju has a meaning of ’as we both know’, and
can be described as a particle that the speaker uses to create common ground
with the listener. Liksom can be translated as ’kind of’ and is sometimes used
to modify a word when the speaker does not quite find the expression she is
looking for, as in example (7), where the speaker modifies the verb insåg
(’realized’) with liksom.

74
 Language Acquisition and Use in Multilingual Contexts

(7) det var då jag väl liksom insåg att jag måste faktiskt börja fundera på vem
fan jag är (’that was when I kind of realized that I actually must start to
think about who the hell I am’). Written narrative, adult man
[ws17mDNS].

Liksom also has another use, meaning ’as is’, or ’like’. In this sense, it is
used and accepted in writing. Such use occurs once (out of two written
instances) in the adult data. Ju also has an alternative meaning, namely the
first part of the comparative expression: ju mer … desto mer (’the more…
the more’). This occurs, but sparsely, in the data.
Table 3 shows the distribution of liksom and ju in speech and writing in
the data. One general finding is that liksom is used very seldom in writing,
independent of age. It is used in speech by all groups, with the 13- and the
17-year-olds using it the most. Ju is used in both speech and writing by all
groups, and in speech the use remains at the same proportion of all words
(1.3 %) in all age groups (with a dip in the 13-year-olds, with 1.1 %).
However, in writing the use is more common in the groups of 10- and 13-
year-olds than with the 17-year-olds and adults. One interpretation is that the
two youngest age groups use the same vocabulary in writing as in speech to
a larger extent. This can be explained by several factors. They may observe
and imitate their environment – these data suggest that the 17-year-olds and
the adults use ju to the same extent as the younger groups in speech. Another
explanation is that it requires less effort to use well-known words than new,
or less frequent words. McCutchen (2000) proposes that writers make many
choices in order to reduce the cognitive load during writing. Nevertheless we
observe a developmental pattern, where one expression (ju) is used much
more in writing (as an influence of speech) by the young age groups.
In Table 3 relative frequencies (e.g. ju represents 1.3 % of all spoken
words by the 17-year-olds) are used, since the texts vary substantially in
length between the age groups, and raw numbers would say very little of
how common the words are.
Table 3. Distribution of ju and liksom in speech and writing.
Word 10-year-olds 13-year-olds
Speaking Writing Speaking Writing
ju 89 (1.3 %) 20 (0.4 %) 66 (1.1 %) 46 (0.6 %)
liksom 34 (0.5 %) 0 54 (0.9 %) 1 (0.01 %)

Word 17-year-olds Adults

Speaking Writing Speaking Writing
ju 296 (1.3 %) 38 (0.2 %) 320 (1.3 %) 30 (0.2 %)
liksom 176 (0.8 %) 2 (0.01 %) 147 (0.6 %) 2 (0.01 %)

75
Language Acquisition and Use in Multilingual Contexts

Syntactic variation
The syntax may vary between speech and writing. For instance, Strömqvist
(1996) reports instances of so-called left dislocation, where 15-year-olds in
his study introduce a person or phenomenon first and then make a comment.
In Strömqvist’s study, this was found only in speech, but not in writing. In
my data, this is true for the older age groups, but there are examples of left
dislocation in writing from the youngest age group. In example (8) the boy
introduces one of the problems he discusses by mentioning one of the
characters in the film: han som satt bredvid den där tjejen som visade sitt
papper för honom (’he who sat next to that girl who showed him her paper’).
The character in the film is thus introduced with han (‘he’) followed by a
relative clause som satt bredvid den där tjejen som visade sitt papper för
honom. This relative clause would according to (written) grammar be
followed directly by the verb kollade, but it is not. Instead the pronoun han
(’he’) is repeated: han kollade ju inte ens (‘he didn’t even look you know’)
followed by in total two sentences that constitute a comment about what was
introduced. Adults also use similar constructions, but only in speech, as
shown in example (9), where the speaker first refers to the film scenes with
bullying: om man nu tar mobbningsscenerna då där de liksom inte hälsar på
någon och så där va (‘if we take the bullying scenes then where they kind of
do not greet anyone and so on’), and then makes a comment about these
scenes, by stating that men de finns ju precis lika så här eh man gör ju det
även i vuxenvärlden (‘but there is you know exactly the same like this eh
you do that you know also in the adult world’). The construction might not
be an exact parallel to the previously described left dislocation, but it still
demonstrates a way of introducing topics, and commenting on them, using
syntactic structures typical in speech, but uncommon in writing.
(8) ett annat ganska allvarligt problem är att han som satt brevid den där
tjejen som visade sitt papper för honom han kollade ju inte ens han bara
struntade i läxförhöret och tänkte på något annat (’another pretty serious
problem is that he who sat next to that girl who showed him her paper he
didn’t even look you know he just skipped the test and thought about
something else’). Written expository, boy, 10 [wg04mAEW].
(9) om man nu tar mobbningscenerna då där de liksom inte hälsar på någon
och så där va men det finns ju precis lika så här eh det man gör ju det
även i vuxenvärlden (’if we take the bullying scenes then where they
kind of do not greet anyone and so on but there is exactly the same like
this eh you do that you know also in the adult world’). Spoken
expository, adult woman [ws16fDES].

76
 Language Acquisition and Use in Multilingual Contexts

Pragmatics
On the pragmatic level, speech and writing also differ. The writer must pre-
suppose the previous knowledge of a possible reader. The ability to picture
one’s reader, and to adapt the message so that it can be understood outside
the context where it was produced is developed over the ages. Bereiter &
Scardamalia (1987) for instance, propose a transition from knowledge tellers
to knowledge transformers. The former type of writer generates (fairly co-
herent) texts by translating knowledge in the order it is retrieved. The latter
type supplements this strategy with global planning, where, for instance, the
text is organized according to the needs of the reader.
The participants were asked to discuss the film they had seen. The 10-
year-olds often use the film as the common ground between reader and
writer, using the fact that the reader (i.e. the researcher) had seen the film. In
example (10) we find the expression den fröken eller vad det nu var (‘that
teacher or what it was’), as well as the use of ju, which implies that the
writer assumes that the reader is familiar with the situation.
To make use of the reader’s previous contextual knowledge is important
for any writer, but the young writers in this group produce expository texts
that are often almost impossible to understand for anyone who has not seen
the film. To understand that the reader needs to know more background is
easier in the narrative texts, where the topic is unknown to the researcher
(i.e. a personal story), and where there is little shared knowledge about the
situation and topic. In this context the children often give information that
will help the reader understand the context. This is shown in example (11),
where a 10-year-old boy describes what a fender (‘boat fender’) is, since he
expects that the reader will not know this.
(10) ett annat problem var när den fröken eller vad det nu var tappade sina
pengar då kom det ju en tjej som kunde ha sprungit efter (’another
problem was when that teacher or what it was dropped her money then
came, you know, a girl who could have run after her’). Written
expository, boy, 10 [wg04mAEW].
(11) när vi gick på båten skulle jag ta upp fendrarna de som sitter på båten på
sidan de som gör att man inte repar båten som ligger bredvid (’when we
entered the boat I was to pull up the boat fenders those that hang on the
boat on the side they are the ones that make you avoid scratching the
boat beside you’). Written narrative, boy, 10 [wg10mBNW].

The 13-year-olds use the same strategy, but only in speech, as shown in
example (12). This is the very first sentence from the spoken expository text,
where the speaker does not even define who de (’they’) are, but assumes that
the listener knows that it refers to some children in the film. In writing, it
looks completely different, as is shown in example (13); here the same
person does not presuppose the elicitation film as common ground.

77
Language Acquisition and Use in Multilingual Contexts

(12) dels var det ju fusk de fuskade på prov (’partly it was you know cheating
they cheated on a test’). Spoken expository, girl, 13 [wj02fAES].
(13) fusket hade gått att upptäcka om man hade varit mer vaksam (’the
cheating could have been detected if they had been more vigilant’).
Written expository, girl, 13 [wh02fAWS].

To sum up, the 17-year-olds and the adults show ability to create written
texts that do not presuppose that the reader has seen the film, and spoken
texts that interact with, and use the fact that the listener and speaker have
some common knowledge. The results show that only the youngest part-
icipants, the 10-year-olds, have problems adapting their written expository
texts to the needs of their audience.

Discussion and some conclusions

The examples in this paper illustrate that children who are learning to write
do not treat written language as adults do. I have argued that the differences
can often be explained by the fact that the child relies on spoken language
also when writing. Gradually, written competence develops, and the writer
produces more adult-like texts.
The examples demonstrate that the adults and the group of 17-year-olds
use different language in their spoken and written texts. This is salient on all
linguistic levels. The youngest age groups, and primarily the 10-year-olds,
rely on linguistic features typical of speech also in writing. This could be
linked to several factors, such as: ignorance of spelling conventions and
written morphology, but also an inability to define word boundaries, choose
the right level of formality for certain word forms, or follow written syntax,
as well as lacking ability to understand the need for contextual information.
In this paper, only participants with no known reading and writing
difficulties were included. The relation between speech and writing, and the
writing development may of course look different for other kinds of writers.
The transition from being “only” a speaker into becoming both speaker
and writer is long and laborious, and for most people, the development in
writing will go hand in hand with the general linguistic development which
takes place during the school ages. Accordingly, it is difficult to distinguish
between what can be attributed to the general motoric and cognitive develop-
ment, and what is a product of learning the written language per se. Some
theories of written development, for instance McCutchen (2000), claim that
a more fluent text production will allow the writer to free up more cognitive
capacity for other writing processes. This would mean that a writer who has
automatized spelling conventions (e.g. the phoneme–grapheme-relation, and
written morphology) will have more cognitive space for other processes, for
instance the syntactic structure or pragmatic considerations. This would

78
 Language Acquisition and Use in Multilingual Contexts

explain why the 10-year-olds face problems on all linguistic levels in

writing. They may simply be so concerned about low-level processes like
spelling and punctuation, that they must rely heavily on strategies from
speech to be able to produce text. Two such strategies, illustrated here, are
the syntactic construction with left dislocation, and the failure to create a
written expository text that can be understood by someone who has not
watched the same film. However, the 10-year-olds seem able to create inde-
pendent narrative texts, which suggests that they take the opportunity to be
less explicit when the possibility appears (in this case in the expository
written texts). By doing so they reduce the cognitive pressure of producing a
coherent and independent expository.
In parallel with the process of learning to write, children also develop
their spoken language further. The development of both modalities might
occur simultaneoulsy but are not identical. When the child’s development
reaches adult-like performance, the language user has developed different
sets of phonology, morphology, syntax, lexicon and pragmatics for speech
and writing.

References
Bereiter, C. & Scardamalia, M. 1987. The psychology of written composition. Hills-
dale, NJ: Lawrence Erlbaum Associates.
Berman, R. & Verhoeven, L. 2002. Cross-linguistic perspectives on the develop-
ment of text-production abilities: Speech and writing. Written Language and
Literacy, 5. 1–44.
Chafe, W. 1992. Information flow in speaking and writing. In P. Downing, S.D.
Lima & M. Noonan (eds.), The linguistics of literacy, Vol. 21 of Typological
Studies in Language. Amsterdam: John Benjamins Publishing Company. 17–29.
Chafe, W. & Tannen, D. 1987. The relation between written and spoken language.
Annual Review of Anthropology, 16. 383–407.
Clark, H. 1996. Using Language. Cambridge: Cambridge University Press.
Johansson, V. 2009. Developmental Aspects of Text Production in Writing and
Speech. Traveaux de l’Institut de Linguistique de Lund 48. Lund University.
MacWhinney, B. 2000. The CHILDES Project: Tools for Analyzing Talk. 3rd ed.
Mahwah, NJ: Lawrence Erlbaum Associates.
McCutchen, D. 2000. Knowledge, processing and working memory. Educational
Psychologist, 35 (1). 13–23.
Strömqvist, S. 1996. Discourse Flow and Linguistic Information Structuring: Ex-
plorations in Speech and Writing. Gothenburg Papers in Theoretical Linguistics
78. Dept. of Linguistics, Göteborg University.

79
Bilingualism in the university classroom
and student engagement in deep learning
approaches

Marie Källkvist

Introduction
In today’s globalised world, most teaching situations involve multilingual
individuals, and we are witnessing the monolingual approach (i.e. strict use
of the target language only) gradually giving way to flexible multilingualism
in the teaching of foreign/second (L2) languages (Creese & Blackledge
2010; Hélot & O’ Laoire 2011). Interestingly, this coincides with the publi-
cation of a number of books and articles where applied linguists and
language educators express renewed interest in the judicious use of
translation tasks for facilitating L2 learning (Butzkamm & Caldwell 2009;
Cook 2007, 2010; Duff 1989; Malmkjær 1998; Witte et al. 2009). The
support for the occasional use of translation (or L2-to-L1 comparison) in L2
teaching situations extends to work in Second Language Acquisition (SLA)
where empirical work suggests that providing L2 learners with contrastive
information in the L1 may facilitate the learning of certain areas of L2
grammar (e.g. Kaneko 1992; Kupferberg & Olshtain 1996; Rolin-Ianziti &
Brownlie 2002; Spada & Lightbown 1999).
What comes to the fore in these publications is that the suggested role of
translation emerges as markedly different from what it was in the days of the
grammar-translation method. Translation is now promoted either as a tool
for learning specific areas of grammar only, or as a real-life task, in the
frame of task-based language education, used only judiciously and alongside
a range of other tasks to foster interaction and learning among students.
Benefits attributed to translation are that it involves cognitive processes that
learners naturally engage in in real life (Gonzàlez Davies 2001; Malmkjær
1998); that L1 and L2 become linked in ways that may enhance memory
traces (Hummel 1995, 2010); that translation tasks may hold particularly
strong potential for student-initiated activity and interaction in L2 class-
rooms (Källkvist 2013); that it makes learners ‘notice the gap’ (Gonzàlez

80
 Language Acquisition and Use in Multilingual Contexts

Davies 2001); that it directs attention to accuracy and perhaps caters par-
ticularly to learners who like to use analytical skills (Malmkjær 1998) and
who are introverted rather than extroverted (Sewell 2004). In a recent book-
length treatment of the topic, applied linguist Guy Cook expresses his full
support for the use of translation in L2 teaching and learning (2010:155):
…translation has an important role to play in language learning‒that it de-
velops both language awareness and use, that it is pedagogically effective and
educationally desirable, and that it answers student needs in the contemporary
globalized and multicultural world […] Translation is just such a bridge
between the familiar and the unfamiliar, the known and the unknown. To
burn that bridge or to pretend that it does not exist, hinders rather than helps
the difficult transition which is the aim of language teaching and learning.

It is easy to agree with Cook’s recommendations, but the challenge resides in

conducting further research on the theory and practice of translation for L2
learning purposes, in search for more evidence-based L2 teaching and learn-
ing. In previous publications (Källkvist 2008, 2013), I have argued for the
need to begin building a theoretically informed empirical basis from which
we can gain insights into what specific learning conditions and contexts may
warrant the judicious use of translation.
Although a great deal has been written about the use of translation in L2
education over the past 10-15 years, empirical research addressing the effect
of translation on L2 learning is surprisingly scant (cf. Cook 2010:90). Num-
erous questions worthy of attention therefore come to mind. What value
might translation have for early L2 learners when it comes to certain areas of
the L2 grammar, L2 vocabulary and L2 pragmatics and culture, for instance?
What is the optimal use of translation in the case of advanced-level learners
whose aim is to reach professional-level command?
This chapter aims to synthesize research on the effect of translation on L2
learning and then discuss the value of translation tasks from the point of
view of deep approaches to learning, which is a central concept in higher-
education pedagogy. As there is virtually no research on the effect of trans-
lation on the learning of L2 grammar apart from my own (cf. Cook 2010),
the focus is on my own longitudinal study of L1-to-L2 translation tasks
carried out at a university in Sweden (Källkvist 2008, 2013).
The purpose is not to promote translation as a teaching method; rather, it
is to encourage a continued call for identifying teaching and learning con-
texts in which carefully designed translation tasks may have value. I begin
by surveying research on the effect of translation on L2 learning; the focus
then turns to the effect of translation on L2 classroom interaction, and finally
I discuss translation in relation to the concept of deep approaches to learning.

81
Language Acquisition and Use in Multilingual Contexts

Research on the effect of translation on L2 learning

A number of studies have investigated the use of L1 translation equivalents
for the purpose of enhancing learners’ knowledge of vocabulary in a L2,
leading to a consensus that the provision of L1 cues in learning materials
facilitate rather than hamper L2 vocabulary learning (cf. Danan 1992, 2010;
Lambert 1986; Grace 2000; Hummel 2010; Laufer and Girsai 2008; Prince
1996). There is also a substantial body of research that suggests that L1 use
emerges naturally in L2 classrooms as a learning and communication
strategy used by students during their completion of different types of tasks
(for a review see Moore 2013).
As examples of this type of research, two separate studies found bene-
ficial effects on the retention or comprehension of L2 vocabulary when
learners watched subtitled films where subtitling was reversed (i.e. sound in
L1 and the script in L2) (Danan 1992; Lambert 1986). Two studies (Laufer
& Girsai 2008; Prince 1996) investigated classroom-based learning of L2
vocabulary, both of which found significantly superior effects on learning
when new L2 vocabulary was introduced in the company of L1 translation
equivalents and contrastive information than when the same L2 vocabulary
was presented in L2 co-text. Laufer and Girsai’s study involved teenage
Hebrew-speaking L2 learners of English learning English vocabulary both in
isolation and as collocations. Prince’s study involved French-speaking inter-
mediate-level learners of English at university level in France. He found that
the use of translation equivalents led to significantly more learning for all his
participants, but particularly those in the lower proficiency range. Finally, a
recent study by Hummel (2010) shows that passive translation (copying a
sentence containing an L1 translation equivalent of the target L2 word) led
to significantly more learning in the short term than conditions where learn-
ers were actively translating sentences either from L1 to L2 or vice versa.
My own study of translation and the learning of L2 grammar (Källkvist
2008) is a longitudinal and experimental study carried out in the authentic
module English grammar and written proficiency in three university EFL
classrooms at a Swedish university. Participants were first-semester students
of English who had grown up in Sweden, having been exposed to Swedish
from birth, and who attended all classes over a 13-week period. They had
had 9–10 years’ classroom instruction in English prior to entering university.
On registering for first-semester English, they were divided into two
groups using matched-pair random assignment on the basis of an in-house
placement test of practical English grammar. The two groups were then
taught English grammar by the same teacher (myself), who was also a
participant observer (in the ethnographic sense of the term).
One group was provided with task sheets targeting practical English
grammar, involving translation from Swedish (L1) into English (L2). This
group will be referred to as group T (for ‘translation’). The other group was

82
 Language Acquisition and Use in Multilingual Contexts

provided with task sheets targeting exactly the same morphosyntax the same
number of times, but through gap tasks, noticing tasks, and text-editing
tasks, all of which involved no comparison with their L1, and never trans-
lation. This group is therefore referred to as the NoT group (for ‘no trans-
lation’).
Both groups were pre-tested and post-tested using an identical battery of
tests, which consisted of three parts, all focusing on difficult L2 English
grammar, of which they had no or very limited knowledge at the beginning
of the course: a) a multiple-choice test, b) a short text to be translated from
Swedish into English, and c) a written retelling of a story in English. All
three tests contained the target grammar structures focused on in the course
and covered extensively in the task sheets. The results showed significant
gains on the multiple-choice and translation tests from pre- to post-test for
both groups. There were gains also on the retelling test, in which they
operated only in L2, but due to it being a retelling task, it was not possible to
carry out inferential statistics.
The T group showed greater gain on the multiple-choice test and on the
translation test, but the gain was not significantly greater than that of the
NoT group. On the retelling test, the NoT group had greater gain from pre-
test to post-test than the T group. In this sample, there was thus an exercise
effect (though non-significant): the T group improved more from pre- to
post-test when post-tested through translation. The NoT group showed
greater gain from pre-test to post-test than the T group on the retelling test,
which required no translation.
The study concluded that both the translation and the no-translation treat-
ments led to statistically significant amounts of learning of difficult L2 mor-
phosyntax, at least in the short term. The most important finding to report
here is that L1-to-L2 translation tasks led to learning of difficult L2 gram-
mar, and that this was traceable even in tests that did not involve translation.
All of the above described research into the effect of translation on L2
learning jointly suggests positive effects. This should provide us with fuel
and incentive to further explore the role translation may play in different
learning contexts and with different kinds of learners. One important factor
in L2 learning is interaction in L2, to which I now turn.

A study of translation tasks and interaction

Whereas there are a number of studies focusing on L2 learner interaction
engendered by different kinds of tasks (e.g. Garcia Mayo 2002a, 2002b;
Storch 1997, 1998, 2008; Swain et al. 2009; Toth 2008), there seems to be
only one focusing on interaction and translation tasks (Källkvist 2013). This
study involved the two experimental groups T and NoT described above and
one additional intact group of undergraduate students of English who took

83
Language Acquisition and Use in Multilingual Contexts

the same module as the T and NoT groups. The intact group will be referred
to as the TI group (for ‘translation but intact’ group) and students in this
group were given task sheets with a mix of translation, gap, transformation
and text-editing tasks as part of the same module. The translation tasks de-
signed for use in the T and TI groups were modelled on translation exercises
used in published exercise materials accompanying the assigned reading that
would normally be used. The tasks involving no translation (gap, noticing,
composition and text-editing) mirrored the translation tasks in that exactly
the same grammar features were targeted the same number of times.
The data consist of audio-recordings of 19 (out of 54) 90-minute classes.
The 19 recordings are spread across all three groups, focusing on translation,
and identical translation tasks were completed by groups T and TI. Identical
gap tasks were completed by groups TI and NoT, and an identical text-
editing task was completed by all three groups.
Close reading of the transcribed data showed that interaction engendered
by the translation tasks was different from the other tasks particularly with
regard to (i) student-initiated turns, (ii) degree of focus on the targeted L2
morphosyntax, and (iii) the nature of teacher scaffolding. When the trans-
lation tasks were used, there were significantly more turns initiated by
students, where they either asked questions to the teacher regarding different
language features, typically concerning whether their translation was an
acceptable rendering of the Swedish source text, or to initiate discussion of
different possible translations of the same Swedish word, phrase or sentence.
This result was consistent across both the T and TI groups. It was also found
that when translation was used, there were significantly fewer turns relating
to the targeted L2 grammar feature (for example the use of the past tense
versus the perfect aspect, or the use of the definite versus zero articles) than
when gap tasks were used.
Finally, the teacher used different scaffolding techniques depending on
task type, although the data were too limited for inferential statistical
analysis. When translation tasks were used, the teacher allowed ample time
for questions/discussion of any language feature in the text, using
scaffolding prompts to encourage student discussion. When gap tasks were
used, on the other hand, there were far fewer student-initiated questions, and
the teacher instead used the time available to scaffold students into
verbalising their explicit knowledge of English grammar to explain why a
certain text segment was accurate or inaccurate.
These findings suggest that the translation and gap tasks served different
purposes in these classrooms, targeting different learning outcomes. Gap
tasks were used to achieve a sharp focus on the targeted L2 grammar and to
encourage students to build and verbalise explicit knowledge of the kind
needed when teaching English or when translating. The translation tasks –
on the contrary – were used to encourage student-initiated interaction and to
actively use English to discuss features of the English language.

84
 Language Acquisition and Use in Multilingual Contexts

Further, when translation tasks were used, the student-teacher interaction

was characterised by true information-gap questions from several different
students to the teacher. At times a couple of students became involved,
expressing their opinions regarding the translatability of certain Swedish
words or expressions into English. An observation on the part of the teacher
was that students felt they had expert knowledge of and about Swedish,
giving them confidence to initiate and enter into discussions regarding trans-
lation equivalence. A further anecdotal observation is that the heightened
level of confidence was noticeable in their body language as students
involved in discussion that concerned the meaning and/or use of Swedish
assumed a more assertive posture. Yet another observation was that the
presence of native-English-speaking students (there were two in the TI
group) led to more student-initiated turns in the form of true information-gap
turns, since they too on several occasions brought up issues of translation
equivalence between their L1 – English – and their L2, Swedish.
The study concludes by suggesting that translation tasks have particular
potential for student-initiated activity, mainly for two reasons: firstly,
translating involves a stage of comparison between the two different
languages involved, which entails making choices between different possible
acceptable translations. Secondly, the translation tasks used involved com-
parison as students completed their translations in pairs, small groups or
individually. As a result, there were a number of different target texts and
thus alternative solutions present simultaneously in the classroom, which led
to questions relating to acceptability and idiomaticity.
Finally, the study suggests that translation tasks can easily be adapted to
suit different teaching/learning situations by including or excluding certain
vocabulary, such as words rich in near-synonyms and culture-specific voca-
bulary. If such vocabulary is omitted and – in the case of nouns – common
nouns are replaced by proper nouns, then it was possible to achieve a focus
on grammar features even when translation tasks were used.
The next section brings these research results beyond SLA into the realm
of higher education pedagogy.

Translation and approaches to learning

The concept of approaches to learning has developed over the past 30 years
or so, emerging from research carried out in higher education at Gothenburg
University, Sweden, originally published by Marton and Säljö (1976). In the
early days, the concept of approach related to students’ reading of texts,
whereas the contemporary view of approaches to learning is that they apply
to any learning task and to learners of all ages as they engage with different
kinds of tasks (cf. Ramsden 2003). This concept of approaches to learning is

85
Language Acquisition and Use in Multilingual Contexts

now considered central to higher education pedagogy and two different

approaches are distinguished: surface and deep (Ramsden 2003).
An approach is not tied to specific individuals as everyone is capable of
both deep and surface approaches to learning; instead, an individual adopts
surface or deep approaches when engaging with the material in a task. Typi-
cally, a surface approach is adopted when an individual’s focus is getting a
task over with, often with limited effort. There is then normally a focus on
memorising details or parts (surface learning) instead of searching for
meaning and understanding by relating facts to concepts (deep learning)
(Ramsden 2003). The surface approach is not necessarily negative, however.
It may be both an economical and appropriate approach to adopt when com-
pleting a specific task or it may be helpful to pave the way for adopting a
deep approach to gain conceptual understanding. For instance, it may be eco-
nomical and advantageous to adopt a surface approach when someone has to
memorise facts in a short period of time in order to pass a test (Ramsden
2003). However, recent publications in higher education pedagogy unani-
mously advocate deep approaches as being superior for enhancing under-
standing of course content and in the life-long building of knowledge in
general (Biggs & Tang 2007; Ramsden 2003; Elmgren & Henriksson 2010).
The issue under exploration here is whether translation may have the
potential to engage students in such a way that they adopt the deep approach
to the translation task at hand, and – more broadly – to learning in general as
they progress through L2 education. To my knowledge, there are no pub-
lished studies that relate L2 learning to deep and surface approaches, and
what follows here is therefore explorative and aims to generate hypotheses
and further discussion.
Below is a synthesis of characteristics of surface versus deep approaches
to learning proposed in five recent publications on learning and teaching in
higher education. Following this, in Table 1, I discuss translation tasks
against each of the characteristics of deep approaches to learning.
The following characterise deep learning approaches according to Biggs
and Tang (2007), Elmgren and Henriksson (2010), Entwistle and Peterson
(2004), Kember (2007) and Ramsden (2003):
 Learning tasks capture students’ interest and have a clearly defined purpose,
which is returned to regularly in the course.
 The course focuses on students’ understanding of the course content.
 The course content is related to students’ general knowledge, to students’
knowledge in the subject area prior to the course, and – if possible – to students’
personal experiences.
 Aspects of the course content that may be particularly challenging are identified,
and carefully selected concrete examples are provided to facilitate students’
understanding.
 Students’ are given opportunities for reflection and group discussion.

86
 Language Acquisition and Use in Multilingual Contexts

Entwistle and Peterson (2004) provide a more extensive list, including also
the following:
 Constructive friction is created, i.e. course materials involve learning and
thinking activities that students are unlikely to engage in and solve on their own.
 Students are provided with criteria for assessment and grading.
 Formative assessment is used in order to stimulate students’ understanding of
concepts and to provide feedback.

In Table 1 below, translation tasks are discussed against each characteristic

of deep learning listed above.
Table 1. Translation tasks in relation to aspects of a deep learning approach.
Characteristics of deep Are these characteristics typically present when
learning approaches students engage in translation tasks?
Learning tasks capture The amount of student-initiated interaction engendered
students’ interest and by the translation tasks used in Källkvist (2013) is a
have a clearly stated sign that the tasks were interesting and relevant.
purpose, which is re- Translation is a suitable task to use for the purposes of
turned to regularly in the focusing on L2 form and has the advantage that
course. everyone present in the classroom pays attention to the
same source text.
The course focuses on Translation tasks can be tailored to targeting students’
students’ understanding understanding of course content, for example of Eng-
of the course content. lish grammar, since translation tasks often require prac-
tical application of students’ knowledge of grammar.
For instance, isolated sentences or even parts of senten-
ces can be used if the object of learning is particular L2
grammar (cf. Spada & Lightbown 1999) or L2 voca-
bulary. The translation tasks in Källkvist’s interaction
study (2013) were used to discuss a broad range of
language issues: English grammar, culture-specific
vocabulary, synonymy, collocational restrictions and
the usage of prepositions, the usefulness of bilingual vs.
monolingual dictionaries, in order to target students’
understanding of the complexity of language and
language use.
The course content is Translation links students’ knowledge of their L1 with
related to students’ their developing command of their L2 (cf. discussion in
general knowledge, to Hummel 1995; Danan 2010), and Källkvist’s interact-
their knowledge in the ion study (2013) showed heightened levels of student
subject area, and their engagement when there were discussions of L1 and L2
personal experiences (if equivalence. Also, most people probably have personal
possible). experiences of a real-life situation where they needed to
translate.
Aspects of the course L1 source texts can easily be adapted to different learn-
content that may be par- er groups by the instructor through the inclusion of
ticularly challenging are language features that are particularly challenging. For

87
Language Acquisition and Use in Multilingual Contexts

identified, and carefully instance, the lexical content can be adjusted, and trans-
selected concrete exam- lation is by its very nature a concrete activity (cf.
ples are provided to fac- discussion in Sewell 2004).
ilitate students’ under-
standing of concepts.
Students are given opp- Since translation is a problem-solving activity, it is nat-
ortunities for reflection urally suitable to reflection in groups or individually.
and group discussion. The interaction data in Källkvist (2013) attest to it
having good potential to engender student-initiated
discussion.
Constructive friction is Translation is highly suitable for making students
created. ‘notice the gap’ (between one’s own L2 capacities and
those of native speakers of the L2, cf. González Davies
2001) since the teacher can choose to control the nature
of the source texts. Källkvist’s interaction data (2013)
suggest that students raised many queries regarding
language use that they would not be able to fully solve
on their own.
Students are provided In a course focusing on form in the L2, it is relatively
with criteria for assess- easy to list criteria for both assessment and grading.
ment and grading. This is due to the fact that all students translate the
same source text, whose linguistic content can be con-
trolled by the instructor.
Formative assessment is Given enough resources and time, students can hand in
used in order to stimul- translations for feedback from their instructor during
ate students’ understan- the course as well as at the end.
ding of concepts and for
the purpose of giving
them feedback.

Finally – and very importantly – tasks need to be meaningfully embedded

within the course content and be clearly related to the learning outcomes, the
assessment as well as to students’ education and educational goals in gen-
eral. Mediation through teacher and student agency is likely to facilitate
students’ engagement in the task through adopting deep approaches to
learning. It is also likely that other L2 writing tasks have similar potential to
engender deep learning approaches through teacher and student agency.

Discussion and conclusion

I have argued that findings from the empirical research reviewed above
provide good reason to continue exploring the value of judiciously used
translation tasks in L2 learning contexts. There is also reason to believe that
carefully designed translation tasks in L2 can serve to engage students in
deep learning approaches.

88
 Language Acquisition and Use in Multilingual Contexts

I propose two reasons for this. One is the fact that the instructor can con-
trol the content of the source language text, adjusting it to suit the needs as
well as interests of different student groups. Translation therefore also has
particular potential to create constructive friction by adjusting the level of
difficulty of the source language text. Moreover, by being bilingual rather
than monolingual, it involves a phase of comparison (between the L1 and the
L2 and between different possible alternative translations), which in
Källkvist’s (2013) study led to significantly more student-initiated dis-
cussions and reflection than the equivalent monolingual tasks did. Finally,
owing to teacher control of the source text, translation tasks are relatively
easy to mark and grade.
The second reason pertains to the nature of language competence: gen-
erating the almost infinitely possible number of utterances in a language is
based on knowledge of morphosyntax, lexis, collocations, stylistic and prag-
matic resources etc. This type of knowledge cannot be acquired by the
human mind solely through the memorisation of isolated facts. Particularly
in production tasks that require the composition of a full, coherent, cohesive
and meaningful text through using a range of L2 resources, students need to
draw on the complex knowledge base that language is.
However, translation is only useful up to a point and is a viable choice
only in contexts where the L1 is shared among everybody present, such as in
the context of Swedish higher education in English as a foreign language
where completed upper-secondary-school-level courses in both Swedish and
English are required for admission. Since the ultimate aim of modern
language courses is mastery of the L2 without L1 mediation, there comes a
point when translation needs to replaced by other writing tasks.
At present we know far too little about what specific learning situations
and individuals are likely to be assisted by translation, and we have a long
way ahead before we can deliver evidence-based modern language courses.
For instance, curiously, there are no studies of the role of L1 glosses in L2
vocabulary learning involving pupils and students in Swedish schools and
universities.
Finally, the discussion in this chapter of L2 learning tasks in relation to
deep approaches to learning is explorative. This domain appears wide open
to research, but there seems to be reason to believe that the judicious use of
short, carefully designed translation tasks (and other tailor-made L2 writing
tasks) have the potential of engaging students in deep approaches to learn-
ing. For further study, it would be particularly interesting to focus on stu-
dents’ interaction patterns and negotiations while completing tasks involving
translation. Having students rate the extent to which they feel they are en-
gaging in deep approaches while completing different kinds of writing tasks
in their L2 would also provide interesting further data that are more directly
focused on approaches to learning.

89
Language Acquisition and Use in Multilingual Contexts

References
Biggs, J. & Tang, C. 2007. Teaching for Quality Learning at University. Maiden-
head: McGraw Hill.
Butzkamm, W. & Caldwell, J.A.W. 2009. The bilingual reform: A paradigm shift in
foreign language teaching. Tübingen, Germany: Gunter Narr Verlag.
Cook, G. 2007. A thing of the future: Translation in language learning. International
Journal of Applied Linguistics, 17 (3). 396–401.
Cook, G. 2010. Translation in language teaching. Oxford: Oxford University Press.
Creese, A. & Blackledge, A. 2010. Translanguaging in the bilingual classroom: A
pedagogy for learning and teaching? The Modern Language Journal, 94 (1).
103–115.
Danan, M. 1992. Reversed subtitling and dual coding theory: New directions for
foreign language instruction. Language Learning, 42 (4). 497–527.
Danan, M. 2010. Dubbing projects for the language learner: a framework for inte-
grating audiovisual translation into task-based instruction. Computer Assisted
Language Learning 23 (5). 441–456.
Duff, A. 1989. Translation. Oxford: Oxford University Press.
Elmgren, M. & Henriksson, A-S. 2010. Universitetspedagogik. Stockholm: Nor-
stedts.
Entwistle, N.J. & Peterson, E.R. 2004. Conceptions of learning and knowledge in
higher education: Relationships with study behavior and influences of learning
environments. International Journal of Educational Research, 41 (6). 407–428.
García Mayo, M.P. 2002a. Interaction in advanced EFL pedagogy: a comparison of
form-focused activities. International Journal of Educational Research, 37 (3–
4). 323–341.
García Mayo, M.P. 2002b. The effectiveness of two form-focused tasks in advanced
EFL pedagogy. International Journal of Applied Linguistics, 12 (2). 156–175.
González Davies, M. 2001. Translation in foreign language learning: Bridging the
gap. Paper presented at the APAC–ELT Convention, Barcelona, Spain.
Grace, C. 2000. Gender Differences: Vocabulary Retention and Access to Transla-
tions for Beginning Language Learners in CALL. The Modern Language
Journal, 84 (2). 214–224.
Hélot, C. & O’ Laoire, M. (eds.), 2011. Language policy for the multilingual class-
room: Pedagogy of the possible. Bristol, England: Multilingual Matters.
Hummel, K.M. 1995. Translation and second language learning. Canadian Modern
Language Review, 51 (3). 444–455.
Hummel, K.M. 2010. Translation and short-term L2 vocabulary retention: Hind-
rance or help?. Language Teaching Research, 14 (1). 61–74.
Källkvist, M. 2008. L1-L2 translation versus no translation: A longitudinal study of
focus-on-formS within a meaning-focused curriculum. In L. Ortega & H.
Byrnes (eds.), The Longitudinal Study of Advanced L2 Capacities. New York:
Routledge. 182–202.
Källkvist, M. 2013. Languaging in Translation Tasks Used in a University Setting:
Particular Potential for Student Agency?. The Modern Language Journal, 97
(1). 217–238.
Kaneko, T. 1992. The role of the first language in foreign language classrooms.
Temple University, Japan.
Kember, D. 2007. Enhancing University Teaching: Lessons from research into
award-winning teachers. Abingdon: Routledge.

90
 Language Acquisition and Use in Multilingual Contexts

Kupferberg, I. & Olshtain, E. 1996. Explicit contrastive instruction facilitates the

acquisition of difficult L2 forms. Language Awareness, 5 (3-4). 149–165.
Lambert, W.E 1986. Pairing first- and second-language speech and writing in ways
that aid language acquisition. In J. Vaid (ed.), Language Processing in biling-
uals: Psycholinguistic and neuropsychological perspectives. Mahwah, N.J.:
Lawrence Erlbaum. 65–96.
Laufer, B. & Girsai, N. 2008. Form-focused instruction in second language vocabul-
ary learning: a case for contrastive analysis and translation. Applied Linguistics,
29 (4). 694–716.
Malmkjær, K. 1998. Introduction: Translation and language teaching. In K. Malm-
kjær (ed.), Translation and language teaching: Language teaching and trans-
lation. Manchester, England: St. Jerome. 1–11.
Marton, F. & Säljö, R. 1976. On qualitative differences in learning. II – Outcome as
a function of the learner’s conception of the task. British Journal of Educational
Psychology, 46 (1). 115–127.
Moore, P.J. 2013. An Emergent Perspective on the Use of the First Language in the
English-as-a-Foreign-Language Classroom. The Modern Language Journal 97
(1). 239–253.
Prince, P. 1996. Second Language Vocabulary Learning: the role of context versus
translation as a function of proficiency. The Modern Language Journal, 80, (3).
478–493.
Ramsden, P. 2003. Learning to Teach in Higher Education. Abingdon: Routledge.
Rolin-Ianziti, J. & Brownlie, S. 2002. Teacher use of learners’ native language in the
foreign language classroom. Canadian Modern Language Review, 58. 402–426.
Sewell, P. 2004. Students buzz around the translation class like bees round the
honey pot—why? In K. Malmkjær (ed.), Translation in undergraduate degree
programmes. Philadelphia: John Benjamins. 151–162.
Spada, N. & Lightbown, P.M. 1999. Instruction, first language influence, and deve-
lopmental readiness in second language acquisition. The Modern Language
Journal, 83 (1). 1–22.
Storch, N. 1997. The editing talk of adult ESL learners. Language Awareness, 6 (4).
221–232.
Storch, N. 1998. Comparing second language learners’ attention to form across
tasks. Language Awareness, 7 (4). 176–191.
Storch, N. 2008. Metatalk in a pair work activity: Level of engagement and impli-
cations for language development. Language Awareness, 17 (2). 95–114.
Swain, M., Lapkin, S., Knouzi, I., Suzuki, W., & Brooks, L. 2009. Languaging: Uni-
versity students learn the grammatical concept of voice in French. The Modern
Language Journal, 93 (4). 5–29.
Toth, P.D. 2008. Teacher- and learner-led discourse in task-based grammar instruc-
tion: Providing procedural assistance for L2 morphosyntactic development.
Language Learning, 58 (2). 237–283.
Witte, A., Harden, T. & Ramos de Oliveira Harden, A. (eds.), 2009. Translation in
second language learning and teaching. Oxford, England: Peter Lang.

91
The relationship between lexical and
syntactic development in English as a
second language

Satomi Kawaguchi

Introduction
This cross-sectional study8 investigates the development of argument mapp-
ing in learners of English as a second language within the framework of
Processability Theory (PT), see Pienemann et al. (2005). It explores em-
pirically the mapping hypothesis in English L2 which has not been
previously treated in any detail in PT. My interest in this exploration stems
from the fact that the mapping of thematic roles (such as agent and patient)
onto grammatical functions (such as subject and object) presents different
degrees of difficulty for L2 learners. The study also explores the relationship
between lexical and syntactic development. Following Lexical Functional
Grammar (LFG), e.g. Bresnan (2001) grammatical constructions are lexi-
cally restricted in language learning (cf. Pinker 1984, Tomasello 1992 in L1
acquisition). Therefore, lexical ability may be assumed to be strongly related
to syntactic ability. Especially lexical learning of verbs is important because
it leads to the development of sentences, where more complex verbs are
necessary to construct complex sentences.
The chapter is organized as follows. The next section briefly explains the
lexical mapping hypothesis and two sources of difficulties for argument
mapping. This is followed by the presentation of a study investigating the
lexicon-syntax relationship in Japanese learners of English L2, its results,
and conclusion.

8
This study was funded by the University of Western Sydney Seeding Grants scheme.

92
 Language Acquisition and Use in Multilingual Contexts

Lexical Mapping Hypothesis in PT

Many linguists (e.g. Bresnan 2001; Foley & Van Valin 1984; Givón 1984;
Jackendoff 1972) have suggested a universal hierarchy of thematic roles, as
in (1). This hierarchy orders the relative prominence of the arguments of a
predicator: the higher the level in the hierarchy, the more cognitively
prominent is the argument. Grammatical functions also have a hierarchical
relationship according to their prominence, as in (2). All core functions are
more prominent than non-core functions. So the most prominent role is
universally the agent. This means that the agent is more likely to be encoded
as a core-argument, such as subject, rather than non-argument (Bresnan
2001). On the other hand, the locative role, located at the bottom of the
hierarchy, is less prominent and likely to be encoded as a non-core-argument
rather than core argument.
(1) Thematic hierarchy (Bresnan 2001: 307)
Agent > Beneficiary > Experiencer/Goal > Instrument >
Patient/Theme > Locative
(2) Relational hierarchy (Keenan and Comrie 1977, referred in Bresnan
2001: 96)
core non-core

SUBJ > OBJ > OBJ > OBL > COMPL > ADJUNCT

Based on these universal thematic and relational hierarchies, Pienemann et

al. (2005) first proposed the Lexical Mapping Hypothesis. This is currently
formulated as follows:
In second language acquisition learners will initially map the highest
available role in the semantic hierarchy onto a minimally specified
SUBJ/TOP. We call this: canonical mapping. Then they learn to map further
arguments onto grammatical functions (GF) other than SUBJ or OBJ. Finally,
they learn to attribute prominence to a particular thematic role lower in the
semantic hierarchy by promoting it to SUBJ. At this stage they also learn to
defocus the highest role by suppressing it or mapping it onto a GF other than
SUBJ. We call these non-canonical mapping. (Bettoni & Di Biase, in
preparation.)

The Lexical Mapping Hypothesis, then, predicts that the initial syntactic
structure that learners construct as soon as they are able to produce
utterances of more than one word will utilize canonical mapping. This
contributes to the realisation of canonical order structures in the L2 which
rely on the association Agent-Subject and Patient-Object appearing in a fixed

93
Language Acquisition and Use in Multilingual Contexts

position.9 Such association, in line with other acquisition theories (e.g.

Pinker 1984; Slobin 1982), is assumed to require least processing effort (e.g.
Pienemann et al. 2005). From a psycholinguistic point of view, Agent (rather
than Patient or Theme) is the most prominent participant role in an event
(Jackendoff, 1972), and from a grammatical point of view, Subject is the
most prominent grammatical function (Keenan and Comrie 1977). Thus the
Agent-Subject association is the most harmonious because the most pro-
minent participant role is mapped onto the most prominent grammatical
function taking the most prominent (first) position. This is schematically
represented in Figure 1 with the English example the cat ate the fish where
the Agent <cat> is mapped onto the grammatical function (GF) Subject
while the patient <fish> is mapped on the Object GF. By contrast, in passive
sentences, exemplified in Figure 2, the highest thematic role (i.e. Agent) is
suppressed. However, the suppressed “agent” may appear as Adjunct. Hence
passive constructions are clear cases of non-canonical mapping. L2 learners,
however, find this difficult at first. The lexical mapping predicts that they
will learn this alternative mapping only after canonical mapping is in place.
eat <x, y>
agent patient thematic roles

Subject Object grammatical functions

the cat the fish constituent structure

Figure 1. Active mapping: the cat ate the fish.

be eaten <x>
agent patient thematic roles

ø Subject Adjunct grammatical functions

the fish the cat constituent structure

Figure 2. Passive mapping: the fish was eaten by the cat.

9
Of the six possible ways of ordering Subject, Object and Verb in languages, SVO, SOV and
VSO “are overwhelmingly more frequent, reflecting the universal tendency for the subject to
precede the Object” (Comrie et al. 2003). See also Greenberg (1966), Tomlin (1986).

94
 Language Acquisition and Use in Multilingual Contexts

Recently, it has become possible to measure brain activities involved in

language processing. For example, Yokoyama, Miyamoto et al. (2006)
measured brain activities in the left middle temporal gyrus for processing
three different lexical categories in Japanese native speakers. They found
that activity levels increased in the following order: Noun < Unmarked
Active Verb < Inflected Passive Verb. They explain that “…verbs have
richer lexical information than nouns, including information relating to
subcategorization, argument structure, thematic structure, and so on, all of
which are critical to sentence processing” (p. 1309). Further, Yokoyama,
Okamoto et al. (2006) in a different experiment involving Japanese L1 –
English L2 late bilinguals showed evidence of significant interaction be-
tween sentence type (active vs. passive) and language (L1 vs. L2). They
found greater processing load with passive sentences than with their active
counterpart. These results were consistently found in both the L1 and the L2
of these bilinguals. Speakers use these more costly non-canonical structures
because they are linguistic devices to attribute prominence to thematic roles
other than the Agent. For example, the passive is a linguistic alternative way
to construct a verbal message to place prominence on Patient rather than
Agent (Levelt 1989) and allows the speaker to impart different perspectives
on discourse world situations (Payne, 2011). Next I exemplify canonical and
non-canonical structures used in my experiment.
Canonical mapping: The sentence in (3) represents a typical canonical
mapping construction with a transitive verb break which requires two
arguments where the more prominent role, the Agent, is mapped on the
Subject and the less prominent role, the Patient, is mapped on the Object
grammatical function. There are also some intransitive verbs10 (the
unergative ones) whose sole argument, typically an Agent or Experiencer – a
role high in the thematic hierarchy – maps on the Subject. This is also a
canonical mapping operation, represented in (4).
(3) canonical transitive
Break <Agent, Patient> I broke the stick

Agent Patient
SUBJ OBJ

10
Intransitive verbs, which require only one argument, are divided into unergatives and
unaccusatives (Burzio 1986). These classifications are based on the thematic role that the sole
argument carries in the sentence. The argument of unergative verbs typically bears an agent or
experiencer role as in (a) while that of unaccusative verbs typically bears a theme or patient
role as in (b).
a. Tom cried (Unergative); b. The window broke (Unaccusative)

95
Language Acquisition and Use in Multilingual Contexts

(4) canonical intransitive

cry <Experiencer> the baby cries

Experiencer

SUBJ

Non-canonical mapping: A typical case of non-canonical mapping is the

passive construction, explained above. This type of non-canonical mapping
is usually called ‘structural’ because the alternative lexical entry (be eaten,
versus active eat) creates a structural frame which is regular and predictable.
Causative constructions are similarly non-canonical mapping structures,
which are also regular and predictable alternative constructions (see
Pienemann et al. 2005, and Kawaguchi 2009 for Japanese causatives). Other
non-canonical mappings are created ‘lexically’ in the sense that they are
intrinsically required by the lexical verb, hence they are neither regular nor
predictable so they need to be learned case by case. Characteristically, these
verbs map hierarchically lower thematic roles, e.g., Theme, on the Subject.
For instance, with the unaccusative alternative of the verb close in (5), the
hierarchically lower role, Theme, is mapped on the Subject while the Agent
role of the eventuality of ‘closing’ is actually excluded from the scene al-
together.
(5) Unaccusative verb11
close <Theme> The door closed suddenly

Theme
SUBJ

(6) Psych Verb: OBJ Experiencer (OE)

frighten <Theme, Experiencer> Her screams frightened John

Theme Experiencer
SUBJ OBJ

Another group of verbs which build non-canonical mappings in English are

the so-called Psych verbs (c.f. White et. al. 1999). For example, the verb
frighten, in (6), requires the Theme her screams (i.e. a lower role in the
thematic hierarchy) to be mapped on the Subject while John, the Ex-
periencer, (i.e. a higher role in the hierarchy) is mapped on the less
prominent grammatical function Object. Hence non-canonical mapping can

11
There are alternating (e.g., close, break) and unalternating accusative verbs (e.g., arrive,
appear) in English (see Hirakawa 2003). The former involves non-canonical mapping while
the latter build canonical mapping.

96
 Language Acquisition and Use in Multilingual Contexts

be generated either lexically, as exemplified in (5) and (6) or structurally as

in passives and causatives discussed above.

The study
Given the above context the research questions are as follows:
Q1. Do Japanese-speaking learners of English L2 invariably acquire
canonical mapping before non-canonical mapping?
Q2. Is there a relationship between lexical size in the L2 (Nation and
Beglar 2007) and the acquisition of the different types of non-
canonical mapping?

The informants in this study were 22 Japanese L1 speakers of English L2

(five male and 17 female) aged between 20 and 56 years (mean 31, SD 9.9)
with length of stay in Australia ranging from 9 days to 27 years. They
include: working holiday participants, university students (all undergraduate,
MA and PhD), business people and one professional translator. Adult
informants of varying lengths of stay may provide a wide range of attain-
ment in English L2. An 18-year-old simultaneous bilingual first language
speaker of English and Japanese, born in Australia by native Japanese-
speaking parents, participated as a control since one of the tasks involved
Japanese to English translation. In order to ensure the informants’ ano-
nymity, codes such as JA1, JA2 were assigned.
Two tasks were used for this study. First, the Nation & Beglar (2007)
vocabulary size test which measures vocabulary knowledge up to the 14,000
word families level was used. It is well attested (e.g. Nation 2006) that there
is a significant correlation between vocabulary size and receptive language
abilities (i.e. reading and listening). It is interesting to test whether pro-
ductive language ability also shows a relationship with vocabulary size.
The second task used was a written production translation task. There are
not many studies of productive abilities in the field of second language
acquisition; with the exception of Hirakawa (2003) and White et. al. (1999)
most tasks involve either comprehension tests or grammatical judgment
tests. However, Håkansson and Norrby’s (2007) demonstration that both oral
and written development follow the PT hierarchy paved the way to the use of
written production data to measure L2 acquisition in PT. In the second task,
the informants were asked to translate 25 Japanese sentences into English
and were instructed to use a particular English verb in their translation for
each sentence (see Appendix A). Six of these 25 sentences, which involve
ditransitive verbs, raising and subordination, were not used in the present
study. The 19 verbs tested in the translation production task (summarised in
Table 1) contain five canonical and fourteen non-canonical structures (six
lexical non-canonical and eight structural non-canonical structures).

97
Language Acquisition and Use in Multilingual Contexts

Table 1. The 19 English verbs targeted in the translation production task.

Non-canonical
Canonical
Lexically non-canonical Structurally non-canonical
transitive
(n=5) Intransitive Transitive Passive (including: Causative &
(Unaccusative) (Psych Verb) adjectival & stative Causative-
(n=3) (n=3) passive (n=6) Passive (n=2)
Break Freeze Please Kill (be killed) Wash (make X
Wash Fall Confuse Break (be broken) wash Y)
Kill Fall from Shock Close (be closed) Work (be
Close Confuse (be confused) made to work)
Stop Interest (be interested)
Surprise (be
surprised)

All verbs were selected from the first (most frequently used) vocabulary
band12 i.e. 1 to 1,000 for English except for the verbs shock and confuse
which are in the second band. In some cases the informant’s ability to use
the same verb in canonical and non-canonical ways was tested. For example,
the verb kill was included twice in the translation, in an active and a passive
context respectively.

Results and discussion

This section shows results of the vocabulary size test followed by the
translation task results, after which the relationship between the informants’
lexical size and translation ability relating to canonical and non-canonical
mapping is discussed.
Vocabulary size test. Figure 3 lists the distribution of the 22 informants’
vocabulary size. Minimum and maximum sizes are 3,000 and 12,700 words
(mean 7,141 words, SD=2,466). Each informant’s vocabulary size is listed in
the second column of Table 2 and Table 3. Nation & Beglar (2007) note that
“undergraduate non-native speakers successfully coping with study at an
English speaking university have a vocabulary around 5,000–6,000 word
families. Non-native speaking PhD students have around a 9,000 word
vocabulary” (p. 9). The present study covers informants ranging from well
below undergraduate university level to beyond PhD level. The informant
who attained the highest vocabulary size (12,700) – higher than 18-year old-
native control (11,300) – is a professional English–Japanese translator.

12
English frequency list based on Vp-BNC list
<http://www.lextutor.ca/freq/lists_download/1000_families.txt>

98
 Language Acquisition and Use in Multilingual Contexts

6 5

No. of informants (n=22)

5 4
4 3 3
3 2
2 1 1 1 1 1
1 0 0 0 0
0
1k-
2k-
3k-
4k-
5k-
6k-
7k-
8k-
9k-
0-

10k-
11k-
12k-
13k-
Vocabulary size

Figure 3. Informants’ vocabulary size.

Translation task. The lexical mapping hypothesis predicts that non-

canonical mapping is acquired after canonical mapping. However, it does
not predict relative acquisitional order of lexical versus structural non-
canonical mapping. Table 2 shows the distribution of lexically non-canonical
mapping in the translation task compared with the distribution of canonical
mapping. Table 3 does the same with structurally non-canonical mapping. In
each cell of the two tables, frequency counts of correct argument mapping is
entered before slash (i.e. ‘/’) while the total number of contexts is listed after
the slash. The question mark, ‘?’, indicates that the structure is well-formed
but the informant did not use the verb they were instructed to use in the
translation (e.g. shock): instead they used it as a noun as in The Airplane
accident gave shock to people in the world. Accuracy rates, indicated in
brackets, are calculated on the basis of the correctness of argument mapping
and, therefore, such errors as grammatical agreement (e.g. subj-verb and
plural agreements) are not counted as errors. The shaded area indicates that
the PT usual emergence criterion is satisfied, i.e. that the learner was able to
produce the target structure more than once in different contexts. Table 2 and
3 show the implicational nature of the acquisition of canonical and non-
canonical argument mapping. Further, within lexically non-canonical
mapping (Table 2), unaccusatives are acquired before psych verbs. Thus the
acquisition order is: canonical > unaccusative > psych verb. Also, Table 3
indicates that the acquisitional implication is as follows: canonical > passive
> causative. Thus, all non-canonical mappings are acquired after canonical.

99
Language Acquisition and Use in Multilingual Contexts

Table 2. Canonical versus lexically non-canonical mapping compared to vocabulary

size.
Non-canonical
Informant Vocab. size (x1,000) Canonical Unaccusative Psych Verb
JA15 5.0 2/5 (.4) 1/3 (.33) 1/3 (.33)
JA19 4.6 3/5 (.6) 1/3 (.33) 0/3
JA11 3.0 4/5 (.8) 1/3 (.33) 0/3
JA21 6.8 5/5 (1.0) 1/3 (.33) 1/?1/3 (.33)
JA16 4.7 5/5 (1.0) 2/3 (.67) 0/3 (0)
JA18 5.1 5/5 (1.0) 2/3 (.67) 0/3 (0)
JA17 5.8 5/5 (1.0) 2/3 (.67) 0/3 (0)
JA20 4.1 5/5 (1.0) 3/3 (1.0) 0/3 (0)
JA09 5.4 5/5 (1.0) 2/3 (.67) 0/3 (0)
JA14 6.4 5/5 (1.0) 3/3 (1.0) 0/3 (0)
JA08 6.8 4/5 (.8) 2/3 (.67) 0/3 (0)
JA06 6.9 4/5 (.8) 3/3 (1.0) 0/3 (0)
JA04 9.0 5/5 (1.0) 3/3 (1.0) 0/1(?)/3 (0)
JA07 8.1 5/5 (1.0) 2/3 (.67) 1/3 (.33)
JA22 6.2 5/5 (1.0) 3/3 (1.0) 1/3 (.33)
JA05 7.7 5/5 (1.0) 3/3 (1.0) 1/3 (.33)
JA12 8.8 5/5 (1.0) 3/3 (1.0) 3/3 (1.0)
JA10 9.0 5/5 (1.0) 3/3 (1.0) 3/3 (1.0)
JA01 9.7 5/5 (1.0) 3/3 (1.0) 3/3 (1.0)
JA02 10.1 5/5 (1.0) 3/3 (1.0) 3/3 (1.0)
JA13 11.2 5/5 (1.0) 3/3 (1.0) 3/3 (1.0)
JA03 12.7 5/5 (1.0) 3/3 (1.0) 3/3 (1.0)
NS control 11.3 5/5 (1.0) 3/3 (1.0) 3/3 (1.0)

100
 Language Acquisition and Use in Multilingual Contexts

Table 3. Canonical versus structurally non-canonical mapping compared to

Vocabulary size.
Non-canonical
Informant Vocab size (x1,000) Canonical Passive Causative
JA16 4.7 2/5 (.4) 0/6 (0) 0/2 (0)
JA19 4.6 3/5 (.6) 1/6 (.17) 0/2 (0)
JA11 3.0 4/5 (.8) 1/6 (.17) 0/2 (0)

JA20 4.1 5/5 (1.0) 2/6 (.33) 0/2 (0)

JA18 5.1 5/5 (1.0) 3/5 (.5) 0/2 (0)
JA09 5.4 5/5 (1.0) 3/5 (.5) 0/2 (0)
JA15 5.0 5/5 (1.0) 4/6 (.67) 0/2 (0)
JA04 9.0 5/5 (1.0) 4/6 (.67) 0/2 (0)
JA22 6.2 5/5 (1.0) 6/6 (1.0) 0/2 (0)
JA08 6.8 5/5 (1.0) 2/6 (.33) 1/2 (.5)
JA17 5.8 4/5 (.8) 4/6 (.67) 1/2 (.5)
JA06 6.9 5/5 (1.0) 4/6 (.67) 1/2 (.5)
JA05 7.7 4/5 (.8) 3/5 (.5) 2/2 (1.0)
JA14 6.4 5/5 (1.0) 2/6 (.33) 2/2 (1.0)
JA21 6.8 5/5 (1.0) 5/6 (.83) 2/2 (1.0)
JA07 8.1 5/5 (1.0) 6/6 (1.0) 2/2 (1.0)
JA12 8.8 5/5 (1.0) 6/6 (1.0) 2/2 (1.0)
JA10 9.0 5/5 (1.0) 6/6 (1.0) 2/2 (1.0)
JA01 9.7 5/5 (1.0) 6/6 (1.0) 2/2 (1.0)
JA02 10.1 5/5 (1.0) 6/6 (1.0) 2/2 (1.0)
JA13 11.2 5/5 (1.0) 5/6 (.83) 2/2 (1.0)
JA03 12.7 5/5 (1.0) 6/6 (1.0) 2/2 (1.0)
NS control 11.3 5/5 (1.0) 5/5 (1.0) 2/2 (1.0)

Relationship between vocabulary size and acquisition of argument

mapping. Table 2 and 3 above show that all learners have acquired canonical
mapping and use it fairly accurately even in the lower vocabulary size range.
On the other hand, the relationship between vocabulary size and non-
canonical mapping (Table 2) is only clear with psych verbs, that is, you need
a high vocabulary size (greater than 8,800) to accurately use them. On the
other hand passives (Table 3) appear to satisfy PT’s emergence criterion
when vocabulary size is 4,100 or greater, but are used more accurately when
vocabulary size is 5,000 or greater. However, causatives seem to require a
higher vocabulary size, which, by itself, is no guarantee that this mapping is

101
Language Acquisition and Use in Multilingual Contexts

acquired. For instance informant JA04 (Vocabulary size 9,000) shows no

sign of emergence for this mapping.

Accuracy percentages 100

80
60
40
20
0
0 2 4 6 8 10 12 14
Vocabulary size (x1,000)

Figure 4. 22 informants’ vocabulary size and accuracy rate of lexically non-

canonical mapping13.

100
Accuracy percentages

80
60
40
20
0
0 2 4 6 8 10 12 14
Vocabulary size (x1,000)

Figure 5. 22 informants’ vocabulary size and accuracy rate of structurally non-

canonical mapping.

Overall distribution percentages of lexically non-canonical mapping are

shown in Figure 4 against the informant’s vocabulary size. Figure 5 does the
same with structurally non-canonical mapping. As regards lexically non-
canonical mapping, in Figure 4, the 8,800 words point seems to divide
informants into two distinct groups: the ones above 8,800 with 100% ac-

13
In Figure 4, there are 21 diamonds representing 22 informants. This is because two inform-
ants of the same vocabulary size achieved the same accuracy percentage (i.e., vocab size 6,8k
with 33% accuracy). Thus one diamond represents two informants.

102
 Language Acquisition and Use in Multilingual Contexts

quisition and the others with less clear correlations between vocabulary size
and acquisitional level. Since a 5,000–6,000 vocabulary size range is deemed
to characterise successful university students of undergraduate courses in
English-speaking universities, surprisingly such informants achieved just
above 30% accuracy with lexically non-canonical mapping. Only English L2
speakers with a vocabulary size of over 8,800 managed the lexically non-
canonical mapping well. Compare the translation produced by JA22
(vocabulary size 6,200) with the one produced by JA03 (vocabulary size
12,700). JA22’s sentence exhibits wrong argument mapping (parallel
examples are found in Appendix A).
(7) JA 22: The explanation of Teacher, Yamada is confused.14
(8) JA 03: Professor Yamada's explanation always confuses his
students.

Accuracy of structurally non-canonical mapping, in Figure 5, shows, with

few exceptions, a tight relationship with vocabulary size up to 8,100 words.
Beyond this point informants reached 100% or close to 100% accuracy. This
indicates that English L2 learners learn structurally non-canonical mapping
gradually as they increase their vocabulary size. In order to test the strength
of the relationship between vocabulary size and lexically/structurally non-
canonical mapping, a Pearson correlation test was performed. Note that the
test was applied to informants whose lexical size is up to 8,100 because
beyond this point the informants hit the maximum accuracy rate (with one
exception: a vocabulary size of 9,000 and about 50% accuracy). Statistical
results are shown in Table 4. Both types of non-canonical mapping showed
statistically significant results with the one-sided Pearson test. However, the
p-value of structurally non-canonical mapping is 0.00012 while that of
lexically non-canonical mapping is 0.032. This indicates that the strength of
correlation is much higher for structural than lexical non-canonical mapping.
Table 4. Pearson correlation between vocabulary size and accuracy of non-canonical
mapping.
n df Pearson’s r value p-value (1 sided)
Lexically non-canonical 15 13 0.49 0.032
mapping
Structurally non-canonical 15 13 0.81 0.00012
mapping

14
The source sentence is: “Yamada-sensei-no setsumei-wa itsumo gakusei-o konran-saseru”
(Yamada-teacher-GEN explanation-TOPIC always students-ACC confuse-PASSIVE).

103
Language Acquisition and Use in Multilingual Contexts

It can be assumed that structurally non-canonical mapping relates more

closely to vocabulary development. I interpret this result as follows.
Structurally non-canonical mapping such as that taking place in passive and
causative constructions have a common specified syntactic frame and can be,
therefore, constructed more predictably by the learner. On the other hand,
lexically non-canonical mapping involves specific verbs. From the learning
point of view, there is no systematic way of identifying a set of such verbs a
priori. Instead, learners have to learn verb features (i.e. argument spe-
cifications) one verb at a time. This may be why some learners with a large
vocabulary are still unable to construct sentences with many intrinsically
lexical unaccusative and psych verbs. Since most verbs in the study were
selected from the first band of the English frequency list, it is evident that to
know a verb’s meaning is different from knowing how to use that verb in its
specific syntactic frame.

Conclusion
This cross-sectional study examined the acquisition of argument mapping
including both canonical and non-canonical mapping based on PT’s Lexical
Mapping Hypothesis and, secondly, it attempted to measure the relationship
between learners’ vocabulary size and the acquisition of canonical and non-
canonical mapping. The main results may be summarised in four points:
1. Canonical mapping precedes all non-canonical mapping, as predicted by
the Lexical Mapping Hypothesis.
2. Structural non-canonical mapping relates closer to developmental stages
than lexical non-canonical mapping.
3. Regarding lexical size and syntactic ability with different verb types,
only a large vocabulary size (above 9,000) may predict the grammatical
ability to produce any type of non-canonical mapping. Both informants
with small and medium-sized vocabulary showed problems with non-
canonical mapping.
4. The order of difficulty in each non-canonical subset is as follow:
For lexical non-canonical mapping: (alternating) unaccusative > psych
verb. For structural non-canonical mapping: Passive> Causative

Thus, the ability to successfully produce non-canonical mapping seems to

characterize advanced syntactic development. On the other hand, only the
highest levels of vocabulary size appear to be sensitive to advanced syntactic
development. Even a fairly large vocabulary (up to 9,000) does not guaran-
tee a parallel syntactic development. This requires further confirmation parti-
cularly in relation to psych verbs and causatives (which were not numerous
in the present study), as well as cross-linguistic validation. From the learning
and teaching point of view an awareness of the subtle role played by the

104
 Language Acquisition and Use in Multilingual Contexts

range of non-canonical mappings and the difficulties faced by learners in

using them may help towards better planning of discourse interventions and
practice in language development.

References
Bettoni, C. & Di Biase, B., In preparation. Processability Theory: theoretical bases
and universal schedules in B. Di Biase & C. Bettoni (eds.), Processability
Theory: Current issues in theory and application.
Bresnan, J. 2001. Lexical functional syntax. Oxford: Blackwell.
Burzio, L. 1986. Italian syntax: A gavernment-binding approach. Dordrecht: Reidel.
Comrie, B., Matthews, S. & Polinsky, M. 2003. The atlas of languages: The origin
and development of languages throughout the world. (Revised Edition ed.):
New Burlington Books.
Foley, W. & Van Valin, R. 1984. Functional syntax and universal grammar. Cam-
bridge: Cambridge University Press.
Givón, T. 1984. Syntax: A functional-typological introduction. Amsterdam: John
Benjamins.
Greenberg, J. 1966. Universals of language (2nd ed.). Cambridge: MIT Press.
Håkansson, G. & Norrby, C. 2007. Processability Theory applied to written and oral
Swedish. In F. Mansouri (ed.), Second language acquisition research: Theory-
construction and testing. Newcastle: Cambridge Scholars Press. 81–94.
Hirakawa, M. (2003). Unaccusativity in second language Japanese and English.
Tokyo: Hituzi Shobo.
Jackendoff, R. 1972. Semantic interpretation in generative grammar. Cambridge,
MA: MIT Press.
Kawaguchi, S. 2009. Acquiring causative constructions in Japanese as a second
language. The Journal of Japanese Studies, 29 (2). 273–291.
Keenan, E. & Comrie, B. 1977. Noun phrase accessibility and universal grammar.
Linguistic Inquiry, 8 (1). 63–99.
Levelt, W.J.M. 1989. Speaking: From intention to articulation. Cambridge, MA:
MIT Press.
Nation, P. 2006. How large a vocabulary is needed for reading and listening?
Canadian Modern Language Review, 63 (1). 59–82.
Nation, P. & Beglar, D. 2007. A vocabulary size test. The Language Teacher, 31 (7).
9–13.
Payne, T. 2011. Understanding English grammar. Cambridge: Cambridge Univer-
sity Press.
Pienemann, M., Di Biase, B. & Kawaguchi, S. 2005. Extending Processability
Theory. In M. Pienemann (ed.), Cross-linguistic aspects of processability
theory. Amsterdam: John Benjamins. 199–251.
Pinker, S. 1984. Language Learnability and language development. Cambridge,
MA: Harvard University Press.
Slobin, D.I. 1982. Universal and particular in the acquisition of language. In E.
Wanner & L.R. Gleitman (eds.), Language acquisition: State of the art. Cam-
bridge: Cambridge University Press. 128–170.
Tomasello, M. 1992. First Verbs: A case study of early grammatical development.
Cambridge: Cambridge University Press.
Tomlin, R. 1986. Basic word order: Functional principles: Croom Helm London.

105
Language Acquisition and Use in Multilingual Contexts

White, L., Brown, C., Bruhn-Garavito, J., Chen, D., Hirakawa, M. & Montrul, S.
1999. Psych verbs in second language acquisition. In E.C. Klein & G.
Martohardjono (eds.), The development of second language grammars: A
generative approach. Amsterdam: John Benjamins. 171–196.
Yokoyama, S., Miyamoto, T., Riera, J., Kim, J., Akitsuki, Y., Iwata, K., Yoshimoto,
K., Horie, K., Sato, S. & Kawashima, R. 2006. Cortical mechanisms involved in
the processing of verbs: An fMRI study. Journal of Cognitive Neuroscience, 18
(8). 1304–1313.
Yokoyama, S., Okamoto, H., Miyamoto, T., Yoshimoto, K., Kim, J., Iwata, K.,
Jeong, H., Uchida, S., Ikuta, N., Sassa, Y., Nakamura, W., Horie, K., Sato, S. &
Kawashima, R. 2006. Cortical activation in the processing of passive sentences
in L1 and L2: An fMRI study. NeuroImage, 30. 570–579.

Appendix A
Examples of translation task used in the study with example answers by JA03
(vocabulary size 12,700) and J11 (vocabulary size 3,000).
Quest.
No. 日本語の文 動詞 英語の文
JA03: My dog broke my doll.
2 私の犬が娘の人形をこわした。 break JA11: My dog breaked my doutear's doll.
山本さんの猫が、私の鳥を殺し JA03: Yamamoto's cat killed my bird.
6 た。 kill JA11: My bird killed by Yamamoto's cat.
JA03: I always close the door of my shop
私は、いつも店のドアを７時に at 7.
9 しめる。 close JA11: I close the door at 7.
JA03: A cat fell off the tree.
10 猫が木から落ちた。 fall JA11: Cat's fall down by tree.
トムのプレゼントは、たいへん JA03: I was very pleased with Tom's gift.
13 私を喜ばせた。 please JA11: I pleased by Tom's present.
JA03: The door to the shop is always
この店のドアは、いつも閉まっ closed.
14 ている。 close JA11: door's closed that store.
JA03: I was very confused after hearing
そのニュースを聞いて、私は、 the news.
15 とても混乱した。 confuse JA11: I confused about that news.
JA03: This watch is broken.
16 この時計は、壊れている。 break JA11: This watch breaked alrady.
JA03: A tree in our yard fell.
17 庭の木が倒れた。 fall JA11: fall in down gerden tree.
JA03: My mother makes me wash the
母は、毎日私にお皿を洗わせる dishes every day.
18 。 wash JA11: my mother
JA03: Tom was killed by Mary.
20 トムは、メアリーに殺された。 kill JA11: Tom killed by Mary.
JA03: Water freezes at 0 degree.
22 水は、０度で凍る。 freeze JA11: Water freeze 0°.
JA03: I am made to work until 8 by my
boss every day.
わたしは、ボスに毎日８時まで JA11: I had work at 8 every day by my
25 仕事をさせられる。 work boss.

106
Teaching English to young learners

Kristin Kersten
Andreas Rohde

Introduction
Teaching an L2 to young learners requires specific teaching approaches.
Whereas abstract rule representation is a minor issue for young children, the
communicative context is of vital importance. For this reason, approaches
based on the principle of “using English to learn it” are preferable to more
traditional approaches relying on the credo of “learning English to use it”.
Based on these assumptions, this article first addresses overall objectives of
early second language learning, different learning scenarios, and age related-
issues, before the necessary prerequisites for successful L2 learning, such as
input, interaction, output and individual factors are highlighted. As a con-
clusion to the previous sections, the final part discusses teaching principles
and means of providing input and interaction which have proven particularly
useful for young learners.

Objectives of early foreign language teaching

The Common European Framework of Reference (Council of Europe 2001)
defines intercultural communicative competence as the main goal of
language learning across European countries and their educational policies.
Intercultural communicative competence integrates linguistic competence
(the four skills focussed on in traditional language teaching, i.e. reading, wri-
ting, listening and speaking) with social and cultural competences (Doyé
1999). Communicative competence has been described as a combination of
grammatical competence, i.e. implicit knowledge about the lexicon, phon-
ology, morphology and syntax of a language; discourse competence, i.e. the
ability to create coherent text from single phrases; sociocultural competence,
i.e. knowledge of communicative rules and conventions that apply to the
specific cultural group and its societal structures, and strategic competence,
a cognitive skill which allows learners to use different ways of expression
and to change channels of communication if they lack the adequate form in
their current state of interlanguage (Savignon 2001). In a nutshell:

107
Language Acquisition and Use in Multilingual Contexts

communicative competence extends to both knowledge and expectation of

who may or may not speak in certain settings, when to speak and when to
remain silent, whom one may speak to, how one may talk to persons of
different statuses and roles […]. (Hymes 1972:279)

Intercultural competence, on the other hand, involves some factual know-

ledge about the target culture/s, social and communicative skills as well as
attitudes which enable communication and understanding between members
of different cultures (e.g. Byram 1997). Intercultural communicative com-
petence can only be attained by learners of English when they are provided
with sufficient opportunities to encounter the language in communicative
contexts. Research increasingly shows that an early start is helpful for these
goals.
Language competencies are part of the core of skills that every citizen needs
for training, employment, cultural exchange and personal fulfilment […] It is
a priority for Member States to ensure that language learning in kindergarten
and primary school is effective, for it is here that key attitudes towards other
languages and cultures are formed, and the foundations for later language
learning are laid, […] in particular by teaching at least two foreign
languages from a very early age. (European Commission 2003: 7; bold
print: highlighted by the authors)

In Europe, there is some dissatisfaction with mainstream education and the

foreign language skills which are attainable in the regular educational
programmes. An early start and a stronger focus on bilingual learning
(Burmeister 2006) or content and language integrated learning (CLIL,
Marsh & Langé 2000) have been suggested to remedy this situation. This is
why an increasing number of both bilingual kindergarten and primary school
programmes are implemented in order to enhance student opportunities to
gain intercultural communicative competence and appropriate linguistic
skills (Kersten et al. 2010a).

Different scenarios
In this article we look at young children from 3 to 10 years of age and in two
fundamentally different language learning/teaching scenarios: kindergarten
and primary school. Educational programmes in Europe differ widely with
respect to the age at which pre-primary and primary education begin. In
Germany, for instance, the term kindergarten refers to an informal setting in
which the children do not receive any formal teaching and, for that matter,
no formal second language teaching. This is true of many other forms of
European preschool education. In bilingual preschools, which introduce
English as a foreign language before formal L2 training in primary or secon-
dary education, there is usually no curriculum for the second language.

108
 Language Acquisition and Use in Multilingual Contexts

Rather, English is used as a medium of communication by L1 English edu-

cators and accompanies the everyday activities of the preschoolers, such as
having meals, taking walks, making objects etc. Our views on this scenario
is based on our own research in kindergartens from European countries such
as Germany, Belgium and Sweden (Kersten et al. 2010a); however, it is also
applicable to kindergarten and preschool scenarios in other countries which
involve formal teaching according to a specific curriculum.
There are many different types of English primary school programmes
throughout Europe. In this article we discuss two specific programmes in
Germany. The first one, referred to as early start, points to the fact that Eng-
lish is now introduced three and a half years earlier than before when the
nation-wide start was grade 5 in Germany. In two steps, the start for English
language teaching was first set at grade 3 in the early 2000s and was shifted
to the second half of the first year in the Bundesland of North-Rhine
Westphalia and even to the onset of primary school in Baden-Württemberg
in 2008, offering the 6 to 10-year-olds two hours of English per week from
grades 1 to 4. The second type is content-based English teaching, i.e. pro-
grammes in which one or more subjects are taught through the medium of
English. Such programmes are referred to as CLIL (Marsh & Langé 2000)
or, if more than 50% of the curriculum is taught in the L2, as immersion pro-
grammes (Genesee 1987). For programmes where every subject, except for
the ambient language, is taught in English the share of English amounts to
approximately 70%. While still exceptional in Europe, such programmes
have been the focus of much research interest (Burmeister 2006, Rohde &
Lepschy 2007, Kersten et al. 2010a). Because of the intensity of such progra-
mmes, research results on young children’s language learning process reveal
interesting insights and consequences for the teaching process, which may be
transferable to other types of educational programmes (Wesche 2002).

The role of age

The title of this article presupposes that young learners should be taught diff-
erently from older learners, as the learning processes involved may be diff-
erent. Learners aged 3 to 10 have as yet not acquired some of the strategies
that older learners are able to develop. Older learners have written language
as additional input and benefit from a greater metalinguistic awareness, and
may therefore be better monitor users than younger learners (Krashen 1982).
The view that young learners learn better or faster than their older peers in
general is not tenable: “[I]t is no longer possible to accept the view that
younger L2 learners are in all respects and at every stage of learning superior
to older learners […]” (Singleton & Ryan 2004:226). However, numerous
studies have found that in the long run child learners “are more likely to
reach higher levels of attainment in both pronunciation and grammar” than

109
Language Acquisition and Use in Multilingual Contexts

older learners (Ellis 2008:31; Long 2007). Another reason which justifies an
early start for L2 learning lies in the children’s ”huge learning potential”
(Cameron 2001, xii) and in the fact that our own research shows that both
bilingual kindergarten and primary school programmes can be immensely
successful without placing a burden on the children (Kersten et al. 2010a).
Especially the bilingual kindergarten scenario comes close to a naturalistic
environment where children are exposed to the L2 informally in everyday
contexts and activities. Irrespective of whether younger learners are better
language learners (which is not addressed in this article), early introduction
to L2 English in kindergarten and primary school paves the way for intro-
ducing a third language when learners enter the high school system. At that
point their L2 English proficiency allows them to start a further language in
lieu of English which, in the traditional system, would only have been intro-
duced by then.

Input / Interaction / Output

It has been shown, especially in immersion contexts, that rich and meaning-
ful input (Gass 2003) may result in above average listening comprehension;
however, input alone does not sufficiently foster the learners’ productive
skills (Cameron 2001:41). It has repeatedly been pointed out that children’s
comprehension precedes their production of the language (e.g. Edelenbos et
al. 2006). However, research suggests that all three components, input, inter-
action and output are necessary for successful second language learning.
Long (1996, 2007) shows that interaction in the shape of negotiation of mea-
ning can be facilitative for the learning process as learners may receive both
positive and negative evidence to structural elements of the L2. Furthermore,
interaction, especially in the guise of negotiation, can foster vocabulary
learning and help learners develop communicative strategies (Gass 2003).
Swain’s research (1985, 1995) suggests that input does not simply turn in-
to intake through interaction. Rather, learners have to be given the oppor-
tunity to produce the L2. On the one hand, it may be intuitively clear that
practising linguistic structures is a necessary prerequisite in order to enhance
one’s L2 skills. On the other hand, Swain’s Output Hypothesis also includes
three further essential functions: a. the noticing/triggering function which
says that L2 production caters to consciousness raising and creates an aware-
ness in the learners for gaps in their interlanguage system, b. the hypothesis-
testing function, according to which the learners have the opportunity to ex-
periment with language structures, and c. the metalinguistic function which
may be essential for older learners as it enables them to reflect on linguistic
structures, and discuss or analyse them explicitly (Swain 1995:128).
In L1 acquisition and naturalistic L2 acquisition, all three components,
input, interaction and output, are usually given. However, they are rarely

110
 Language Acquisition and Use in Multilingual Contexts

warranted in formal L2 learning contexts due to time constraints and, more

often than not, a lack of theoretical knowledge of L2 learning on the part of
the educator or teacher. A first evaluation of teaching English at primary
school in 2006 in North Rhine Westphalia revealed that teaching was
extremely teacher-centered; teachers, on average, talked for 90% of the time
(Engel et al. 2009).
As mentioned above, bilingual kindergartens in which one or more groups
are looked after by both an L1-speaking and an English-speaking educator,
provide ideal opportunities for rich input, interaction and output. Unlike in
the classroom scenario, the three components do not have to be arranged
deliberately; they are naturally given through the routines of the
kindergarten. It would be a great opportunity for traditional classroom
scenarios if they emulated these beneficial context factors from naturalistic
and bilingual language learning (Festman & Kersten 2010).

Quality and quantity of input

In a kindergarten context, the children pick up the L2 in authentic, everyday
activities and contexts. All exchanges and interactions with the native
speakers of English genuinely serve mutual understanding and are not
targeted towards particular grammatical structures (cf. Rohde 2005:157).
The children’s authentic activities are accompanied by the L2, there is no
artificial and arranged context, in contrast to schools where there is a fixed
curriculum and time frame. From that point of view, language learning in a
bilingual kindergarten is not fundamentally different from a naturalistic L2
acquisition scenario. From a quantitative point of view, the children have the
opportunity to receive L2 input at least for half the day, given that most of
the children in Germany attend the kindergarten from 8 am to 1 pm. A
similar quantity of L2 input can only be provided later in a content-based
language teaching programme at primary school. However, the quality of the
input is different then: Whereas the children acquire basic interpersonal
communicative skills (BICS) through the daily routines in kindergarten, in a
content-based teaching programme, the learners need to be provided with
cognitive academic language skills (CALP) (Cummins & Swain 1986).
Research in bilingual preschools shows that both the quality and the
quantity of L2 input, i.e. duration and intensity of L2 contact time, have a
significant effect on vocabulary and grammar comprehension at a very
young age (3–6, Weitz et al. 2010). The quality of L2 input was measured
through an observation checklist geared to principles of language use by the
L2 educators. These principles, based on best practice observations in the
preschools and on well-known teaching strategies within communicative
language teaching (Richards & Rodgers 2001:153–177), are described in
more detail below.

111
Language Acquisition and Use in Multilingual Contexts

In order to enhance the English input for children in both kindergarten

and primary level, it is desirable to also include, if possible, L1 English-
speaking children in the classroom: our experience in immersion schools
shows that the monolingual children appear to be more motivated to learn
the L2 with English-speaking peers than with an adult or adults. The
relationship among the children is symmetrical and more intimate as the peer
group shares similar interests and beliefs. The relationship between an adult
and a child, on the other hand, is asymmetrical. The adult is the child’s
supervisor; interests and beliefs differ so that interactions between child and
adult cannot be the same as between child and child (Friederici 2004:48).

Individual factors
The process and the attainment in language learning are also influenced by
personal traits, skills and aptitude of each individual learner (e.g. Dörnyei
2005). From a teacher’s perspective, it is important to keep these individual
differences in mind to create a stimulating and diversified learning environ-
ment in the classroom. Factors found to be correlated to the success of
language learning are, among others:
 intelligence (IQ) (Dörnyei 2005)
 aptitude, indicated e.g. by working memory, phonological sensitivity, or skills in
grammar analysis (Skehan 1998)
 different learning styles, e.g. perception-based, such as visual, aural or kina-
esthetic learners, or method-oriented such as concrete, analytical, communicative
or authority-oriented learners (Nunan 1999)
 motivation to learn the language, e.g. instrumental motivation which pursues an
immediate learning goal, or integrative motivation, which aims at near-native
competence and a degree of identification with the cultural community of the
foreign language (Gardner 2001)
 social factors, such as identification with a social, cultural or ethnic group
(Pavlenko 2002)
 learner beliefs, i.e. expectations on content and structure of the programme, and
convictions of which strategies are most suitable to their progress (Horwitz
1999)
 the age at which the learning process begins: an early start promotes the level of
ultimate attainment (Muñoz & Singleton 2011; Singleton & Ryan 2004);
however, in classroom contexts, older learners may have an advantage over
younger learners in the rate of learning (Singleton & Ryan 2004: 72–84)
 the amount of L1- and L2-use: frequent L2-use combined with infrequent L1-use
have been found favourable for pronunciation skills in the L2 (Piske et al. 2001)

Many studies also indicate that other personality traits may be related to the
success of language learning, but the results are inconclusive, and personal
variables are extremely difficult to control in an experiment (Dörnyei 2005).

112
 Language Acquisition and Use in Multilingual Contexts

All these factors are strongly interrelated with each other: it is not easy to
tease them apart and measure them, and each of them may be related to
different types of competence or ranges of abilities. It is, however, important
for the good language teacher to bear these factors in mind, to get to know
the individuals’ personality traits and learning styles, and to adjust the
teaching methods accordingly.

Principles of teacher-child interaction

Teacher language
Just as the learning process of young children differs from older, cognitively
more mature learners, teaching strategies have to be altered to suit more
implicit and naturalistic foreign language learning (Kersten et al. 2010b).
Weitz et al. (2010) have developed an observational tool for immersion pre-
school settings for describing differences in the nature and quality of the L2
input and further analysing the effects that these differences may have on the
children’s L2 development. Input quality turns out to have “a greater impact
on the rate of acquisition of receptive L2 grammar knowledge than the mere
amount of L2 input per week (input intensity)” (Weitz et al. 2010:37). The
comprehension of English grammar was assessed with the help of a grammar
test specifically developed for the multilateral EU Comenius Project ELIAS
(Early Language and Intercultural Acquisition Studies). This was the first
time that English grammar comprehension by bilingual preschoolers in three
European countries was tested (Steinlen et al. 2010).

Amount of L2 input
Especially for young children at the beginning of their learning process it is
essential to provide a rich, perceptually stimulating learning environment
which contains as much linguistic input as possible (e.g. Snow 1989, 1990).
Apart from the teaching materials, the teacher in a preschool or foreign
language classroom is usually the only consistent linguistic role model to the
children (Cameron 2001). Since the hours spent in the classroom are necess-
arily reduced as compared to a naturalistic ESL situation, the teacher has to
make the most of the limited contact time. Therefore, one of the most impor-
tant features of teacher language is to use language constantly, like an on-
going commentary of every action that occurs in the classroom.
Commenting on every activity also ensures that the input is lexically and
structurally rich. With language input restricted to limited topics and re-
curring activities, such as songs or games, children do generally not have
access to the range of linguistic features which covers the whole linguistic

113
Language Acquisition and Use in Multilingual Contexts

system of the L2. As a result, they have no chance to infer more complex
linguistic structures and integrate them into their own interlanguage system.
Features which have been described as desirable in early teaching contexts
are a high amount of L2 input, guaranteed through the quality of the teach-
er’s language, and frequent exposure to the second language over a long
period of time (Edelenbos et al. 2006).

Action orientation
Teaching English at primary level in Germany ideally relies on a teaching
concept referred to as action orientation. This originally German concept is
comparable to the approach of communicative language teaching (Richards
& Rodgers 2001:153ff.) but puts specific emphasis on the different senses
involved in the teaching/learning process, and is often referred to as the
head, heart and hand approach (Jank & Meyer 2002:314ff.). Other than
these three senses, there is a further dimension, the linguistic level. This
means that students are sensitized towards the expression of their own aims
and intentions. Action orientation is supposed to raise student awareness of
the effect of their utterances; they may invoke somebody’s concrete physical
action (Can you pass me the pen?) or influence somebody’s mental state (I
really like you). Action orientation that focuses on all four levels – cognitive,
emotional, physical and linguistic – has been shown to be a suitable concept
once learners are faced with formal teaching and a fixed curriculum (Rohde
2012).

Motherese
For early learners, the teachers may adapt their speech to promote a better
understanding of single words and phrases: slower rate, clearer pronun-
ciation, stronger stress and intonation and higher pitch have been observed in
teacher language (Håkansson 1986, Griffiths 1990). When language is adapt-
ed in such a way, learners have a better chance of understanding word and
phrase boundaries, and mapping single forms onto their respective meanings
as presented in the context.
Similar features of speech adaptation have been found in the speech
mothers or caretakers use to address little children to promote their L1
acquisition. This phenomenon has become known as motherese (e.g. Maty-
chuk 2005). While motherese is well suited for very young preschoolers, not
all features of motherese work with older learners. An exaggeratedly high
pitch, often used to address babies or toddlers, may for instance seem out of
place when addressing more mature students.

114
 Language Acquisition and Use in Multilingual Contexts

Gestures, mime and facial expression

Gestures, mime and facial expressions that mimic the content of the utter-
ance are other strategies which usually accompany features of motherese
(Snow 1990). They help identify the object or activity in question and pro-
vide a different way of establishing a connection between content and mean-
ing. That way, such non-verbal expressions provide a context for the other-
wise meaningless linguistic input to the children at the beginning of their
learning process.

Contextualisation
Young learners encounter the L2 as a commentary of every activity in the
classroom, without understanding every single word due to their limited L2
proficiency. Compared to adults, children are usually much more capable
and willing to cope with such a situation, which is natural to them from their
L1 acquisition. But children can only build up linguistic competence from
the limited L2 input if they are able to make sense of the stream of L2 utter-
ances. In order to create an L2 lexicon and hypotheses about the grammatical
structure of the L2, it is vital that learners are able to deduce the meaning of
the linguistic input from the context of the classroom situation. In principle,
the young learners do not have to understand exactly what the educator/
teacher says, but what he or she means. Understanding the situation, know-
ing what is going on in the group, is especially important for young children
as it guarantees their emotional stability and feeling of safety within the
classroom context (Kersten et al. 2010b).
While teachers can draw the students’ attention to specific words or phra-
ses by modifying loudness, pitch or other features of the usual intonation
patterns, they can provide additional clues to the meaning of an utterance by
establishing an easily recognizable context for it. Successful strategies for
doing this comprise the use of visual and aural stimuli such as pictures,
picture stories, CDs, videos, as well as the use of real objects, and other
hands-on materials. Ideally, such a contextualization strategy enables the
learner to understand the situation without having to rely on language at all.
Like a viewer watching an old-fashioned silent movie, a child makes sense
of a given situation by relying entirely on the non-verbal features of the
situation at hand (Burmeister 2006).
Preferably, teachers should code the content in as many ways as possible,
i.e. use different means of explanation simultaneously, such as, intonation,
facial expressions and pictures. That way, they cater to different perceptory
channels and learning preferences of the children, so-called multisensory
learning (Burmeister 2006). Once a solid basis of language competence is
established, contextual clues can be reduced.

115
Language Acquisition and Use in Multilingual Contexts

Scaffolds
Another strategy to promote understanding is scaffolding (Peregoy 1991;
Snow 1990; Massler & Iannou-Georgiou 2010). Teachers are encouraged to
establish classroom routines which are repeated every day providing re-
curring linguistic structures for the children. Such frequent repetition help
the children to quickly build a small repertoire of chunks and phrases in the
L2. Scaffolding routines not only give structure to classroom management
and activities (Edelenbos et al. 2006), they also enable the children to under-
stand and produce output from the very beginning, which usually motivates
them and helps them feel at ease with the foreign language.
Such scaffolds can be either non-verbal signals such as bells or pictures to
indicate an activity, routines such as “weather”, “date”, or “classroom
duties”, or verbal scaffolds such as recurring phrases, formulaic expressions,
rhymes or songs. Even though the children are not able to process phrases
word by word, they understand the meaning in its entirety and refine their
understanding as the language learning process proceeds.

Task-based and content-based teaching principles

As pointed out repeatedly, it is vital for children’s successful language
learning that the learning experience takes place in a genuine context
(Cameron 2001; Lorenz & Met 1989) with authentic materials (Edelenbos et
al. 2006). Such a context is provided when the focus is placed on the mean-
ing rather than on the language form used in the interaction. Task-based
activities consist of meaningful tasks, based on meaningful content. The
approaches of Task-Based Language Learning (Nunan 2004) and Content-
Based Language Teaching (Richards & Rodgers 2001) pertain to these
principles. In such approaches, the language is not the in prime focus of
attention. Instead, it is used as a means of communication.
A task has been described as a meaningful content-related classroom
activity with a specific goal, objective, or outcome, and the process of com-
municating meaning (Nunan 2004). It is carried out in the target language
with a focus on the theme rather than on the form (Legutke et al. 2009). It
should be personalised and adapted to the age and socio-cultural experience
of the children. Age appropriateness includes cognitive, linguistic, inter-
actional, metalinguistic involvement, and physical demands, which need to
be supported by the teacher using different strategies, such as: context-
ualisation, well-known routines or terms, method change and adaptation
(Cameron 2001). Content, i.e. the academic or non-academic subject matter,
should be interesting and cognitively demanding for the learners, and stim-
ulate language learning of the target language. Meaningful content should go
beyond a sole focus on language or culture (Met 1999). This is also a pre-

116
 Language Acquisition and Use in Multilingual Contexts

requisite for meaningful interaction with genuine discussions. The classroom

discourse is driven by the urge to understand the content matter, which all-
ows for negotiation of meaning among the children or between the teacher
and the child. In combination with the strategies described above, this is the
most important prerequisite to ensure holistic language learning in the class-
room (Edelenbos et al. 2006).

Focus on Form
Such meaning-driven interaction also paves the way for different types of
form-focussed teacher feedback, which becomes increasingly relevant for
older, cognitively more mature learners at more advanced stages of learning
(e.g. Lyster & Saito 2010). However, it is possible and desirable to offer age-
appropriate measures with a specific linguistic focus also in the work with
young learners (Cameron 2001). Strategies that focus on form need not, and
should not, be identified with rote learning and grammar drill.
Strategies, such as gestures, pictures, and actions, or verbal explanations,
repetitions, paraphrases and lexical networks using related vocabulary
already known to the child, are used to help the child remember new words.
All these strategies include a certain amount of mental work on the part of
the child, which in turn promotes remembering the word in focus (Cameron
2001:84). Grammar teaching can have a place in preschool and primary L2
learning as well. It is argued that due to the cognitive level of young children
the focus on grammatical features should be implicit rather than explicit:
A grammar-sensitive teacher will see the language patterns that occur in
tasks, stories, songs, rhymes and classroom talk, and will have a range of
techniques to bring these patterns to the children’s notice, and to organise
meaningful practice. (Cameron 2001:122)

In this way the teacher can raise awareness of grammatical phenomena of

the L2 without dampening the children’s motivation through meaningless
drills.
Phonological awareness is possible and helpful at an early age as well
(Edelenbos et al. 2006). Research in L2 phonology has shown that children
have an advantage over older learners when it comes to discriminating the
different foreign sounds of the L2 (e.g. Piske et al. 2001). Teachers can point
out differences and similarities between the two sound systems, and use
songs, rhymes and tongue twisters for metaphonological training (Bernhardt
& Major 2005).
Recasts, i.e. the correct repetition of a learner’s non-target-like utterance,
have been shown to be the among the most successful feedback strategies for
phonological errors (Lyster & Saito 2010). They are also the most common
form of teacher’s corrective feedback to oral errors. They can be integrated

117
Language Acquisition and Use in Multilingual Contexts

in the natural flow of interaction without interrupting the communication and

the focus on the content. Other strategies to correct learner errors are explicit
correction and prompts, which include metalinguistic clues, clarification
requests, elicitation and repetition (Lyster & Ranta 1997). While teachers
present the correct form themselves when using recasts or explicit corr-
ections, they encourage the children to modify their own output for self-
correction with the help of prompts. All three forms of corrective feedback
have been shown to have significant effects on the learners’ uptake.
However, prompts show the strongest effect, and seem to be especially
useful for grammatical errors (Lyster & Saito 2010).
However, these different teaching strategies should be subordinate to the
principle of creating genuine and meaningful activities: a stimulating learn-
ing environment is more important for the progress of very young learners
than cognitive knowledge about the language. From this point of view, there
is no difference between a four year-old preschooler or a 10-year-old child at
the end of primary schooling.

Conclusions
Introducing a second/foreign language to young children cannot start early
enough. There is ample evidence that teaching a second language at both
kindergarten and primary school level may be highly effective. However, at
primary school where there is a fixed curriculum, teaching concepts and
methods must be selected carefully and should not be based on explicit
teaching of grammar and a pronounced focus on forms (as opposed to a
focus on form). This article has discussed a number of general principles
(including concepts and methods) that have proven suitable for the teaching
of young learners. A particularly effective type of teaching programme are
CLIL or immersion programmes. Hopefully, such insights into how young
children learn will also inform the teaching of older students, e.g. at secon-
dary level, and shift the emphasis from a primarily explicit focus on
grammar, still widespread in schools in Germany and throughout Europe, to
a more communication-oriented approach

References
Bernhardt, B., & Major, E. 2005. Speech, language and literary skills 3 years later: a
follow-up study of early phonological and metaphonological intervention.
International Journal of Language and Communication Disorders, 40. 1–27.
Burmeister, P. 2006. Immersion und Sprachunterricht im Vergleich. In M.
Pienemann, J.-U. Keßler & E. Roos. (eds.), Englischerwerb in der Grundschule.
Ein Lehr- und Arbeitsbuch. Paderborn: Schöningh/UTB. 197–216.
Byram, M. 1997. Teaching and Assessing Intercultural Communicative Compe-
tence. Clevedon: Multilingual Matters.

118
 Language Acquisition and Use in Multilingual Contexts

Cameron, L. 2001. Teaching Languages to Young Learners. Cambridge: CUP.

Council of Europe 2001. Modern Languages: Learning, Teaching, Assessment. A
Common European Framework of Reference. Cambridge: CUP.
Cummins, J. & Swain, M. 1986. Bilingualism in Education. London: Longman.
Doyé, P. 1999. The Intercultural Dimension. Foreign Language Education in the
Primary School. Berlin: Cornelsen.
Dörnyei, Z. 2005. The Psychology for the Language Learner: Individual Differences
in Second Language Acquisition. Mahwah, N.J.: Lawrence Erlbaum.
Edelenbos, P., Johnstone, R. & Kubanek, A. 2006. The Main Pedagogical Princi-
ples Underlying the Teaching of Languages to Very Young Learners. Final
Report of the EAC 89/04, Lot 1 study. Brussels: European Commission.
http://ec.europa.eu/education/policies/lang/doc/young_en.pdf
Ellis, R. 2008. The Study of Second Language Acquisition. Oxford: OUP.
Engel, G., Groot-Wilken, B. & Thürmann, E. (eds.), 2009. YEAR Englisch in der
Primarstufe – Chancen und Herausforderungen. Evaluation und Erfahrungen
aus der Praxis. Berlin: Cornelsen.
European Commission 2003. Promoting Language Learning and Linguistic Diver-
sity: an Action Plan 2004–2006. http://ec.europa.eu/education/pdf_en. html
Festman, J. & Kersten, K. 2010. Kognitive Auswirkungen von Zweisprachigkeit. In
U. Massler, & P. Burmeister (eds.), CLIL und Immersion: Erfolgsbedingungen
für fremdsprachlichen Sachfachunterricht in der Grundschule. Braunschweig:
Westermann.
Friederici, A. 2004. Trampelpfade im Gehirn. Süddeutsche Zeitung 97/2004, 48.
Gardner, R. 2001. Integrative motivation and second language acquisition. In Z.
Dörnyei & R. Schmidt (eds.), Motivation and Second Language Acquisition.
Honolulu: University of Hawai’i Press. 1–19.
Gass, S.M. 2003. Input and interaction. In C.J. Doughty & M.H. Long (eds.), The
Handbook of Second Language Acquisition. Oxford: Blackwell Publishing.
224–255.
Genesee, F. 1987. Learning Through Two Languages: Studies of Immersion and
Bilingual Education. Cambridge: Newbury House.
Griffiths, R. 1990. Speech rate and NNS comprehension: a preliminary study in
time-benefit analysis. Language Learning, 40. 311–336.
Håkansson, G. 1986. Quantitative aspects of teacher talk. In G. Kasper (ed.), Learn-
ing, Teaching and Communication in the Foreign Language Classroom.
Aarhus: Aarhus University Press. 83–98.
Horwitz, E. 1999. Cultural and situation differences on foreign language learners'
beliefs about language learning: a review of BALLI studies. System, 27. 557–
576.
Hymes, D. 1972. On communicative competence. In J.B. Pride & J. Holmes (eds.),
Sociolinguistics: Selected Readings. Harmondsworth: Penguin. 269–293.
Jank, W. & Meyer, H. 2002. Didaktische Modelle. Berlin: Cornelsen Scriptor.
Kersten, K. 2012. SLA in bilingual preschools: Evidence from an international
research project. Paper at FLiPP Research Meeting, University of Amersfoort.
Kersten, K., Piske, T., Rohde, A. & Steinlen, A.K. 2010a. Bilingual Preschools. Vol
I: Learning and Development. Vol II: Best Practices. Trier: WVT.
Kersten, K., Steinlen, A.K., Tiefenthal, C., Wippermann, I. & Flyman Mattsson, A.
2010b. Guidelines for Language Use in Bilingual Preschools. In K. Kersten et
al. 2010a. 73–28.

119
Language Acquisition and Use in Multilingual Contexts

Krashen, S. 1982. Principles and Practice in Second Language Acquisition. Oxford:

Pergamon.
Legutke, M., Müller-Hartmann, A. & Schocker-von Ditfurth, M. 2009. Teaching
English in the Primary School. Stuttgart: Klett.
Long, M.H. 1996. The role of the linguistic environment in second language acqui-
sition. In W.C. Ritchie & T.K. Bhatia (eds.), Handbook of Second Language
Acquisition. San Diego, CA: Academic Press. 413–468.
Long, M. H. 2007. Problems in SLA. Mahwah, NJ: Lawrence Erlbaum.
Lorenz, E. & Met, M. 1989. What It Means to Be An Immersion Teacher. Rockville,
MD: Office of Instruction and Program Development, Montgomery County
Public Schools.
Lyster, R. & Ranta, L. 1997. Corrective feedback and learner uptake: Negotiation of
form in communicative classrooms. Studies in Second Language Acquisition,
19. 37–66.
Lyster, R. & Saito, K. 2010. Oral feedback in classroom SLA. Studies in Second
Language Acquisition, 32. 265–302.
Marsh, D. & Langé, G. 2000. Using Languages to Learn and Learning to Use
Languages. TIE-CLIL: Jyväskylä & Milan.
Massler, U. & Ioannou-Georgiou, S. 2010. Best practice: How CLIL works. In U.
Massler & P. Burmeister (eds.), CLIL und Immersion: Fremdsprachlicher
Sachfachunterricht in der Grundschule. Braunschweig: Westermann. 61–75.
Matychuk, P. 2005. The role of child-directed speech in language acquisition: a case
study. Language Sciences, 27. 301–379.
Met, M. 1999. Content-based instruction: Defining terms, making decisions. NFLC
Reports. Washington, DC: The National Foreign Language Center.
Muñoz, C. & Singleton, D. 2011. A critical review of age-related research on L2
ultimate attainment. Language Learning, 44. 1–35.
Nunan, D. 1999. Second Language Teaching and Learning. Boston: Heinle &
Heinle.
Nunan. D. 2004. Task-based language teaching. Cambridge: CUP.
Pavlenko, A. 2002. Poststructuralist approaches to the study of social factors in sec-
ond language learning and use. In V. Cook (ed.), Portraits of the L2 User.
Clevedon, UK: Multilingual Matters. 277–302.
Peregoy, S.F. 1991. Environmental scaffolds and learner responses in a two-way
Spanish immersion kindergarten. The Canadian Modern Language Review, 47
(3). 463–476.
Piske, T., MacKay, I.R.A. & Flege, J.E. 2001. Factors affecting degree of foreign
accent in an L2: A review. Journal of Phonetics, 29. 191–215.
Richards, J.C. & Rodgers, T.S. 2001. Approaches and Methods in Language Teach-
ing. Cambridge: CUP.
Rohde, A. 2005. Lexikalische Prinzipien im Erst- und Zweitsprachenerwerb. Trier:
WVT.
Rohde, A. 2012. Handlungsorientierter Englischunterricht. Nur Basteln und Singen?
Grundschulmagazin Englisch, 6. 37–38.
Rohde, A. & Lepschy, A. 2007, ‘Shoot for the moon’: Englische Immersion in der
Grundschule oder Ein Vorschlag, das Dilemma des bilingualen Sachfach-
unterrichts zu lösen. In P. Bosenius, J. Donnerstag & A. Rohde (eds.), Der
bilinguale Unterricht Englisch aus der Sicht der Fachdidaktiken. Trier: WVT.
1–17.

120
 Language Acquisition and Use in Multilingual Contexts

Savignon, S.J. 2001. Communicative Language Teaching for the twenty-first

century. In M. Celce-Murcia (ed.), Teaching English as a Foreign Language.
Boston: Heinle & Heinle. 13–28.
Singleton, D. & Ryan, L. 2004. Language Acquisition: The Age Factor. Clevedon:
Multilingual Matters.
Skehan, P. 1998. A Cognitive Approach to Language Learning. Oxford: Oxford
University Press.
Snow, M.A. 1989. Negotiation of meaning in the immersion classroom. In E.B.
Lorenz & M. Met (eds.), Negotiation of Meaning. Teacher's Activity Manual.
Rockville: Montgomery County Public Schools.
Snow, M.A. 1990. Instructional methodology in immersion foreign language
education. In A.M. Padilla, H.H. Fairchild & C.M. Valadez (eds.), Foreign
Language Education: Issues and Strategies. Newbury Park: Sage. 156–171.
Steinlen, A.K., Håkansson, G., Housen, A. & Schelletter, C. 2010. Receptive L2
grammar knowledge development in bilingual preschools. In Kersten et al.
2010a (eds.), 69–100.
Swain, M. 1985. Communicative competence: some roles of comprehensible input
and comprehensible output in its development. In S.M. Gass & C.G. Madden
(eds.), Input in Second Language Acquisition. Rowley, MA: Newbury House.
235–253.
Swain, M. 1995. Three functions of output in second language learning. In G. Cook
& B. Seidlhofer (eds.), Principles and Practice in Applied Linguistics: Studies
in Honour of H.G. Widdowson. Oxford: Oxford University Press. 125–144.
Weitz, M., Pahl, S., Flyman Mattsson, A., Buyl, A. & Kalbe, E. 2010. The Input
Quality Observation Scheme (IQOS): The nature of L2 input and its influence
on L2 development in bilingual preschools. In Kersten et al. 2010a (eds.), 5–44.
Wesche, M.B. 2002. Early French immersion: How has the original Canadian model
stood the test of time? In P. Burmeister, T. Piske & A. Rohde (eds.), An Inte-
grated View of Language Development. Papers in Honor of Henning Wode.
Trier: WVT. 357–379.

121
Flerspråkiga elevers språkutbildning

Inger Lindberg
Kenneth Hyltenstam

I detta bidrag diskuterar vi situationen för flerspråkiga elever15 i den svenska

skolan med fokus på undervisningen i svenska som andraspråk i grund-
skolan. I en tillbakablick analyserar vi utvecklingen för ämnet svenska som
andraspråk och de stora brister i implementeringen av ämnet som bidragit till
dess låga status och svårigheter att nå legitimitet. I detta sammanhang läggs
särskild vikt vid ideologiernas makt, bristperspektiv gentemot elever med
utländsk bakgrund och olika tolkningar av begreppet likvärdig utbildning.
Mot bakgrund av modeller för språkutbildning för elever med majoritets-
språket som andraspråk i andra länder presenteras slutligen tänkbara
utgångspunkter för en reformerad och mer differentierad språkutbildning för
flerspråkiga elever som i högre grad än tidigare inriktas mot denna elev-
grupps högst varierande behov av såväl mer allmänna som mer specifikt
skolrelaterade språkfärdigheter.

Skolan och flerspråkigheten

Att döma av officiella rapporter från Myndigheten för skolutveckling (2004),
Skolverket (2005, 2011), och Skolinspektionen (2010) och ett växande antal
avhandlingar (Economou 2007; Granstedt 2010; Gruber 2007; Haglund
2005; Lahdenperä 1997; Otterup 2005; Parszyk 1999) är den svenska skolan
dålig på att tillvarata och värdera flerspråkighet och flerspråkiga elevers kun-
skaper och erfarenheter. I stället förknippas flerspråkigheten ofta med pro-
blem och brister. I en kvalitetsgranskning med inriktning mot förskolors och
skolors insatser för språk- och kunskapsutveckling för barn och elever med
annat modersmål än svenska (Skolinspektionen 2010) konstaterar man t.ex.
att personalen i många fall saknar kunskap om barnens bakgrund och att man

15
Flerspråkig elev är en term som ofta används i myndighetssammanhang och i utbildnings-
politiska texter samt även i svenskspråkig forskningslitteratur. Termen refererar till elever i
skolan med annat modersmål än svenska eller annat modersmål utöver svenskan, dvs elever
med två modersmål.

122
 Language Acquisition and Use in Multilingual Contexts

i liten utsträckning i verksamheten i förskolan och undervisningen i skolan

knyter an till barnens erfarenhetsvärld. Dessutom präglas arbetet i många
förskolor i förhållande till flerspråkiga barn av inställningen att det är trygg-
het som måste prioriteras, vilket leder till att förskolans uppdrag att stimulera
och utmana de flerspråkiga barnen i deras lärande inte alltid får tillräckligt
utrymme.
Att man i förskolan och under skolans tidiga år inte lägger tydligt fokus
på att stimulera de flerspråkiga barnens språk- och kunskapsutveckling kan
få mycket negativa konsekvenser för elevernas lärande på längre sikt, då
utvecklingen av skolrelaterade språkfärdigheter är en långsiktig process som
tar många år i anspråk (Cummins 2001; Thomas & Collier 2002). Problemet
kan mycket väl ligga i att lärarna inte har tillräcklig kunskap om hur man
kan arbeta språkutvecklande med eleverna och på ett pedagogiskt sätt ta vara
på och bekräfta deras skiftande erfarenheter, frågor som ges mycket litet
utrymme i de flesta lärarutbildningar idag (Carlson 2009). Konsekvenserna
av denna brist på tidiga insatser talar sitt tydliga språk i skolstatistiken där
elever med s.k. utländsk bakgrund16 konsekvent är överrepresenterade bland
elever som har låga betyg och inte når målen (Skolverket 2005, 2011).
Men denna situation är på intet sätt ny. Redan på 1980-talet hördes kri-
tiska röster från många lärare och forskare mot vad man menade var skolans
bristfälliga bemötande av elever med andra modersmål än svenska (Ting-
björn 2004). Man ifrågasatte inte minst den stödundervisning i svenska som
andraspråk som bedrivits sedan slutet av 1960-talet som i många avseenden
visat sig bristfällig och betraktades som en högst perifer, kortsiktig och
lågprioriterad verksamhet i många skolor om den över huvudtaget erbjöds.
Enligt andraspråksforskarna Mohan, Leung och Davison (2004) präglades
bemötande av minoritetselever i många länder vid denna tid och även fram-
gent av okunniga och naiva föreställningar om dessa elevers behov som kan
sammanfattas i följande punkter:
– Language minorities will acquire an education and a second language easily and
quickly simply by exposure.
– All that language minorities need is a basic course in the second language.
– The education of language minorities can safely be isolated from the mainstream
of education.
– Educational changes for the benefit of the language minority students will
happen automatically or by the efforts of second language teachers or bilingual
teachers acting without curricular change, institutional support or professional
development (Mohan et al. 2004:2).

16
Elever med utländsk bakgrund omfattar enligt Skolverkets terminologi elever födda i och
utanför Sverige med båda föräldrarna födda utanför Sverige.

123
Language Acquisition and Use in Multilingual Contexts

I Sverige hade även modersmålsundervisningen, som genom den s.k. hem-

språksreformen infördes 1977 i svenska skolor, trots goda intentioner visat
sig ha stora implementeringsproblem och bedrevs i stor utsträckning på und-
antag (Hyltenstam & Tuomela 1996). Generellt var kunskapen om elever
med utländsk bakgrund med andra modersmål än svenska och deras utbild-
ningsbehov skrämmande låg i skolorna vid denna tid och förbehållen en liten
grupp utbildade lärare i svenska som andraspråk och modersmål. I övrigt
präglades förhållningssättet i skolorna gentemot denna elevgrupp av ett brist-
tänkande som starkt påverkade undervisningen i både modersmål och
svenska som andraspråk.

Svenska som andraspråk som eget ämne

I en strävan att bryta den enspråkiga och monokulturella hegemoni som så
starkt missgynnat många flerspråkiga elever i den svenska skolan under
många år infördes 1995 svenska som andraspråk som eget ämne med egen
kursplan, i alla avseenden likställt med det traditionella svenskämnet
(Skolverket 2011/12a, b) vilket resulterade i en hög grad av parallellitet
mellan ämnena. Förespråkare för denna lösning menade att man härigenom
gav erkännande åt den språkliga och kulturella mångfalden, vilket skulle
leda till en höjd status och professionalisering av ämnet svenska som andra-
språk som inte längre skulle betraktas som en kortsiktig stödåtgärd.
Införandet av det nya svenskämnet visade sig dock inte erbjuda den idea-
liska lösning på frågan om minoritetsspråkelevernas rätt till likvärdig utbild-
ning som många hade hoppats på. Ganska snart stod det klart att de miss-
förhållanden som rapporterats under de år som ämnet haft status som stöd-
ämne bestod även efter reformens genomförande. Enligt den kartläggning
som Myndigheten för skolutveckling (2004) senare genomförde fanns
många tänkbara orsaker till ämnets svaga ställning på skolorna som t.ex.
bristande kompetens hos skolledare och lärare, skolans monokulturella och
enspråkiga norm, synen på flerspråkighet som brist snarare än resurs, fortsatt
stödtänkande, dålig implementering av intentionerna i styrdokumenten, otyd-
liga styrdokument, svagt implementeringsstöd för lärarna, godtyckliga beslut
om vilka som ska läsa ämnet, samt brist på tillförlitliga och valida diagnos-
instrument.
I kartläggningen, som baserades på ett mycket begränsat underlag från tre
skolor, menade man att ämnet svenska som andraspråk var segregerande och
skulle tas bort för att ersättas av ett nytt, vidgat, svenskämne för alla i skolan.
Följande är exempel på frågor som inte berördes i kartläggningen, men som
givetvis måste beaktas i en mer grundläggande och seriös analys som under-
lag för framtida beslut och reformer: Hur tolkas och implementeras kurs-
planer i svenska som andraspråk på det lokala planet? Hur påverkas attityder
och värderingar i skolan av makt- och dominansförhållanden och rådande

124
 Language Acquisition and Use in Multilingual Contexts

ideologier i majoritetssamhället? Hur ser man på flerspråkighet och mång-

fald i svenska skolor? Hur påverkar rådande attityder elevernas uppfattning
om ämnet svenska som andraspråk och deras benägenhet att vilja läsa
ämnet? Vilka möjligheter har man att inom överskådlig framtid i nytt vidgat
svenskämne tillvarata de flerspråkiga elevernas högst skiftande behov av
språkundervisning mot bakgrund av svensklärarnas kompetens i svenska
som andraspråk? Hur organiseras andraspråksutbildning i länder där mino-
ritetsspråkselever når goda resultat?
Mot bakgrund av de missförhållanden som påpekades i kartläggningen
och som även dessförinnan och fortlöpande rapporterats i en rad andra
sammanhang är detta exempel på aspekter som förtjänar en särskild belys-
ning och som kan bidra med viktiga förklaringsgrunder till ämnets etable-
ringsproblem, bristande legitimitet och låga status liksom till de flerspråkiga
elevernas i många avseenden otillfredsställande skolsituation mer generellt.

Ideologiernas makt
I en rapportserie där skolsituationen för elever med utländsk bakgrund i olika
OECD-länder belyses (Taguma et al. 2010) påpekar man att intentioner och
verklighet i detta sammanhang inte alltid går hand i hand. Även om Sverige,
till skillnad från många andra länder, inte minst i Europa, har genomfört
flera viktiga utbildningspolitiska reformer med syftet att stötta, ta vara på
och utveckla den språkliga mångfalden i skolan, kan man, bl.a. till följd av
bristande implementering på olika nivåer, konstatera stora brister i det
svenska systemet, vilket också påpekas i OECD-rapporten:
Sweden has already designed a number of measures on migrant education,
but is facing a number of challenges related to implementation, especially
with a highly decentralised system. (Taguma et al. 2010:7)

Som Creese och Leung (2003) påpekat i förhållande till skolsituationen för
flerspråkiga elever i Storbritannien, är relationen mellan officiell politik,
styrdokument och implementering inte linjär. Utbildningsreformer får inte
alltid avsedd effekt på ett enkelt sätt, dvs. ”uppifrån och ner”. I stället sker
implementeringen av de intentioner som ligger bakom politiska beslut
genom tolkande processer på olika nivåer. Det innebär att det är i specifika
kontexter och genom lokala tolkningar som de officiella styrdokumenten
filtreras. Denna process påverkas i sin tur av institutionella, professionella
och individuella erfarenheter, värderingar och föreställningar. Skolpolitikers,
skolledares och lärares svar på och representationer av styrdokumentens
intentioner är ett resultat av såväl professionella överväganden som
ideologiska strömningar i samhället både på ett övergripande och på ett mer
lokalt plan. När lokala attityder och förhållningssätt avviker från den
officiella politiska hållningen och om professionaliteten hos lärarna inom ett

125
Language Acquisition and Use in Multilingual Contexts

område brister är risken stor att oreflekterat ”sunt förnuftstänkande” tar

överhanden och att intentionerna bakom reformerna inte förverkligas (jfr
Hyltenstam & Milani 2012).
Vad man inte i tillräcklig utsträckning tog i beaktande vid införandet av
det nya ämnet var att skolans förhållningssätt gentemot minoritetselever inte
enbart är en skolangelägenhet utan måste ses mot bakgrund av såväl makt-
och dominansförhållanden som rådande ideologier, attityder och värderingar
i majoritetssamhället. Så länge man i samhället i övrigt inte betraktar inte-
gration som ett resultat av en ömsesidigt berikande process av givande och
tagande där mångfalden värderas och bejakas utan som minoritetsmed-
lemmarnas ensidiga anpassning till majoritetssamhället, kommer olikhet och
avvikelser från majoriteten att ses som en brist. Följaktligen blir skolans roll
då att rätta till denna brist och påskynda en utveckling mot ”det normala”.
När majoritetssamhället sätter gränser för vad som ryms inom ”svensk-
heten” får det konsekvenser också för de utbildningssatsningar som riktas
mot minoritetseleverna. I en skola där det i första hand handlar om att bli
”riktigt svensk” och inte sticka ut, kan deltagande både i undervisning i
svenska som andraspråk och modersmål uppfattas som en bekräftelse på att
man avviker och är annorlunda och därför komma i konflikt med en strävan
att få höra till och bli accepterad. Samhällets och skolans betoning av en tra-
ditionell homogen svenskhet kan alltså leda till att minoritetseleverna väljer
att avstå från att delta i undervisning som kan främja deras tvåspråkighet och
allmänna skolframgång. Om skolan i stället genomsyrades av positiva värd-
eringar till kulturell och språklig mångfald, skulle åtgärder för att främja två-
språkig utveckling snarare bli en positiv bekräftelse på de unika möjligheter
som tillhörighet i flera kulturer och färdigheter i flera språk för med sig.
Arbetet i klassrummet återspeglar på många sätt lärarnas syn på sig själva,
eleverna och den uppgift de står inför. Lärarnas attityder och outtalade för-
väntningar spelar därför en avgörande roll för elevernas möjligheter att
lyckas i skolan och förverkliga sina mål. Inga reformer i världen rår över
värderingar, attityder och förhandsuppfattningar som i lika hög grad kan bi-
dra till att uppmuntra, stötta och stärka elever som att hålla tillbaka, tysta och
marginalisera dem. Det är därför särskilt viktigt att alla lärare såväl som
övrig personal reflekterar över sina många gånger omedvetna attityder och
värderingar och över hur t.ex. etnocentricitet och kulturella stereotypier kan
påverka synen på och bemötandet av eleverna. En positiv syn på mångfald
och en gemensam strävan att ta tillvara och bygga på alla elevers varierande
erfarenheter är en förutsättning för framgång i undervisning för elever med
annan språklig och kulturell bakgrund än den homogent svenska. Det
innebär att man ser elevernas skiftande kunskaper, flerspråkighet och varier-
ande sociala och kulturella erfarenheter och värderingar som en resurs och
tillgång som berikar undervisningen.

126
 Language Acquisition and Use in Multilingual Contexts

Den enspråkiga normen

Det bristtänkande som enligt många svenska studier karakteriserat synen på
elever med annan språklig och kulturell bakgrund än den homogent svenska
(se t.ex. Economou 2007; Granstedt 2010; Gruber 2007; Haglund 2005;
Lahdenperä 1997; Parszyk 1999) bekräftas i internationell forskning där den
ojämlika och många gånger diskriminerande behandlingen av minoritets-
elever i skolsystem världen över är väl dokumenterad (Cummins 2001;
Harklau 1994; Pacheco 2010; Valencia 1997; Vollmer 2000). Utifrån ett
sådant bristperspektiv framställs minoritetseleverna, trots sina rötter i och
rika erfarenheter av andra språk och kulturer, som problemelever med
begränsad kapacitet och låg motivation eftersom de saknar det symboliska
kapital som majoritetskulturen värderar.
Sådana uppfattningar och förhållningssätt i förhållande till språklig och
kulturell mångfald bör ses mot bakgrund av ideologier om den språkliga
homogenitetens överlägsenhet. I det sociopolitiska klimat som sedan länge
dominerat stora delar av västvärlden har nationella språk, enspråkighet och
monokulturalitet idealiserats medan språklig mångfald utmålats som hot mot
utveckling, integration och social kontinuitet och koherens (jfr May 2001;
Milani 2007; Pavlenko 2002). Det handlar här om attityder och värderingar
som ofta är omedvetna för individerna själva och sällan ifrågasätts då de
betraktas som självklara (Hyltenstam & Milani 2012).
Internationell forskning visar också att det finns en stark strävan mot
“normalisering” eller “mainstreaming” av minoritetselever som tycks gå
hand i hand med en syn på integration som minoritetsmedlemmarnas ensid-
iga anpassning till majoritetssamhället och negativa attityder gentemot icke-
majoritetsspråkstalare (Vollmer 2000). Detta kan leda till en stigmatisering
av flerspråkiga elever i skolsystemet både på ett mer övergripande plan och i
den vardagliga skolpraktiken.
Den kanadensiske språkforskaren Stephen Talmy (2008, 2009) har t.ex. i
den nordamerikanska kontexten vittnat om sådana tendenser utifrån en
kritisk etnografi som han genomfört på en gymnasieskola på Hawaii med
många elever med engelska som andraspråk. Talmy kunde där observera hur
andra generationens andraspråkselever själva spelade en mycket aktiv roll i
en återkommande projektion av vad han kallar en mainstream/ESL-hierarki17
genom ständiga uttryck för negativa värderingar av undervisningen i
engelska som andraspråk (ESL) som ett sätt att ta avstånd från en lågstatus-
identitet som ESL-elever och förneka en samhörighet med mindre språk-
kunniga nyanlända elever refererade till som FOBs, dvs. ”fresh off the boat”
eller ”just inlandstigna” som i den hawaiianska kontexten hänvisar till det
faktum att många invandrare anländer med båt från Sydostasien.

17
ESL = English as a Second Language

127
Language Acquisition and Use in Multilingual Contexts

Förenklade kategoriseringar
Talmy (ibid.) sätter här fingret på ytterligare en viktig förklaring till att
ämnet svenska som andraspråk haft svårt att vinna legitimitet i den svenska
skolan, nämligen den kategorisering i första- och andraspråk som utgör
själva grunden för ämnets berättigande. Man kan hävda att en sådan kate-
gorisering av elever var adekvat och relativt okomplicerad då ämnet fick
status som eget ämne i mitten av 1990-talet. Med tiden har den dock blivit
mer problematisk. På 1980- och 1990-talen då en majoritet av eleverna med
utländsk bakgrund i den svenska skolan var födda utomlands, var behovet av
en särskild svenskundervisning för denna elevgrupp i allmänhet uppenbart i
samband med skolstarten oavsett om den ägde rum i sjuårsåldern eller
senare. För dessa elever var svenska helt klart ett andraspråk, ett faktum som
sällan ifrågasattes av varken eleverna själva eller av deras föräldrar.
Sedan dess har situationen förändrats radikalt och idag är en majoritet av
de flerspråkiga eleverna i svenska skolor födda i Sverige med svenska som
ett av sina språk sedan tidig barndom. Det innebär ofta att de har ett fullt
utvecklat svenskt informellt språk vid sidan av ett eller flera andra språk,
vilket emellertid inte är någon garanti för att de inte kan få svårigheter i
skolan och senare även i andra utbildningssammanhang som förutsätter
behärskning av ett mer formellt standardspråk. Det som är viktigt att under-
stryka är dock att dessa elevers behov av språkutbildning skiljer sig väsent-
ligt från nyanländas, något som inte i tillräckligt hög grad beaktats. Att
tillfredsställa kommunikativt kompetenta elevers högst varierande behov av
språkutbildningsinsatser med fokus på den typ av skolrelaterade språk-
kunskaper som förutsätts för lärandet i skolans ämnen på olika stadier är en
betydligt mer grannlaga uppgift än att bedriva grundläggande språkunder-
visning för relativt nyanlända elever. Mot bakgrund av den stora brist på
lärare med formell kompetens i svenska som andraspråk som upprepade
gånger dokumenterats under många år är det rimligt att anta att stora grupper
av lärare haft svårt att gå iland med denna uppgift med förödande kon-
sekvenser för såväl elever som för ämnet.
Talmys mainstream/ESL-hierarki utgör en högst relevant förklarings-
grund till varför många elever kan uppfatta det som en kränkning och
degradering (Elmeroth 2008; Gruber 2007; Myndigheten för skolutveckling
2004) att tvingas delta i undervisning i svenska som andraspråk som för-
knippas med låg status och kvalitet, nyanlända och icke-svenskkunniga
elever och låga förväntningar. Det faktum att många vuxna i och utanför
skolan – inte minst i media – ideligen ger uttryck för sin förvåning över att
barn och ungdomar som är födda i Sverige kan behöva särskilt stöd för sin
språkutveckling, förstärker bilden av enspråkighet och homogen svenskhet
som det självklara, eftersträvansvärda och ”normala”. Paradoxalt nog kan
den parallellism som utmärker kursplanerna i svenska och svenska som
andraspråk som har sin grund i en strävan att jämställa de bägge skolämnena

128
 Language Acquisition and Use in Multilingual Contexts

och ge dem lika värde, i själva verket också ha bidragit till att förstärka en
mainstream-/andraspråkshierarki. Även om ämnena vänder sig till elever
med helt olika behov och delvis har olika innehåll kan de nästan identiskt
framskrivna kursplanerna ge en bild av två ämnen med samma innehåll med
den enda skillnaden att det ena vänder sig till elever som faller utanför
”mainstream”. I en skola som fortfarande så starkt präglas av en idealisering
av den enspråkiga normen framstår följaktligen detta ämne som det mindre
attraktiva av de två svenskämnena.
Man skulle också – med hänvisning till den sociala, demografiska och po-
litiska förvandling som Sverige genomgått under senare år – kunna tala om
en eftersläpning som kännetecknar såväl attityder och värderingar i sam-
hället som utbildningspolitiska satsningar i förhållande till flerspråkighet (jfr
Stroud 2003). Som många forskare påpekar låter sig språken i 2000-talets
flerspråkiga samhällen inte alltid uppdelas i uniforma, enhetliga och auto-
noma första- eller andraspråk (Fraurud & Boyd 2006; Jørgensen 2008; Srid-
har 1996). Många barn och ungdomar i Sverige växer upp med flera språk
som de i vardagen använder och förhåller sig till på ett mycket varierat och
dynamiskt sätt. Dessa språk låter sig inte alltid fångas i entydiga traditionella
klassifikationer som förstaspråk och andraspråk (Hyltenstam & Lindberg
2004:15–16) och för många flerspråkiga elever som vuxit upp i Sverige är
gränsen mellan första och andraspråk inte självklar. Skolans behov av att
identifiera och fastställa elevers första- och andraspråk i en ambition att
främja deras språkutveckling på bästa sätt kan alltså komma i konflikt med
elevernas uppfattning om sin egen språkliga identitet och tillhörighet.

Samma eller likvärdig utbildning?

De problem som identifierats i samband med införandet av ämnet svenska
som andraspråk i svenska skolor visar att erkännande av den språkliga och
kulturella mångfalden inte kan uppnås enbart genom kursplaner och andra
styrdokument. Hur man, utan att markera de flerspråkiga eleverna som
avvikande och ytterligare bidra till deras exkludering, ska kunna erbjuda
dessa elever en likvärdig och språkutvecklande undervisning på deras egna
villkor är på många sätt en ödesfråga inte bara för skolan utan också för den
demokratiska samhällsutvecklingen. Att erbjuda alla elever samma svensk-
undervisning vore enligt vissa röster i debatten kring ämnet (jfr Myndigheten
för skolutveckling 2004; Fridlund 2011) den självklara lösningen. I själva
verket är sådana förslag inte bara naiva och historielösa utan också djupt
ojämlika (Lewis 2001), inte minst med tanke på de brister i bemötandet av
flerspråkiga elever som avslöjats i olika studier. I stället för att neutralisera
mångfalden och förringa betydelsen av språkliga skillnader måste skolan
finna nya, flexibla och berikande sätt att representera och ta tillvara den

129
Language Acquisition and Use in Multilingual Contexts

språkliga mångfalden genom vilka elevernas olika språk tillåts samverka,

interagera och utvecklas (jfr Lindberg 2007, 2010, 2011).
Maria Westling Allodi (2002) påpekar, i en analys av hur samhället gen-
om sina utbildningssystem behandlar skillnader för att nå framgång och
minimera diskriminering, att utbildningssystemens förmåga att anpassa ut-
bildningen till varierande villkor och förutsättningar utgör en kärnfråga i
strävan för allas rätt till utbildning. Hon hänvisar här till två olika sätt att
tolka likvärdighet som ofta förväxlas i utbildningspolitiska sammanhang.
Det kan få negativa konsekvenser t.ex. då man ser avsaknad av anpassning
som bevis på likvärdighet. Westling Allodi (2002) skiljer här mellan å ena
sidan likvärdighet som samma undervisning, vilket kan ses som en strävan
uppifrån att med utgångspunkt i sociala behov och rationella motiv påverka,
forma och organisera ett homogent samhälle, och å den andra likvärdighet
som samma möjligheter, vilket snarare kan ses som ett underifrånperspektiv
som ger individer och grupper rätt att behandlas olika i syfte att uppnå lik-
värdighet. Enligt Westling Allodi uppnås inte inkludering för likvärdighet
automatiskt genom att alla elever deltar i samma undervisning oberoende av
deras behov och förutsättningar, eftersom detta förutsätter pedagogiska prak-
tiker som utgår ifrån att alla människor är unika och att man uppskattar
mångfalden. Lärarnas attityder till eleverna, deras förmåga att arbeta med
sociala relationer och anpassa undervisningen till elevernas olikheter spelar
en nyckelroll i detta sammanhang.

Mainstreaning vs. differentiering

Som McKey och Warshauer Freedman (1990) påpekat, återspeglar olika
utbildningsinsatser gentemot minoritetsspråkselever också olika antaganden
om språkutveckling och likvärdighet i utbildning. I ett internationellt per-
spektiv kan i huvudsak två kontrasterande modeller urskiljas, nämligen
mainstreaming och differentiering genom separata program. I Storbritannien
övergavs redan i slutet av 1990-talet särskild undervisning i engelska som
andraspråk i en strävan att uppnå en likvärdig icke-diskriminerande utbild-
ning. Det innebar att undervisningen i engelska som andraspråk blev alla
lärares ansvar och inte längre bedrevs av lärare med specialkompetens i
engelska som andraspråk. Minoritetsspråkseleverna antogs härmed lära sig
engelska genom att delta i ämnesundervisningen utan några särskilda språk-
undervisningsinsatser. Som en konsekvens av det nya systemet finns inte
längre någon särskild lärarutbildning eller krav på särskilda kvalifikationer
för att undervisa elever med andra modersmål än engelska. Likaså har
statligt finansierade insatser inom högre utbildning för kompetensutveckling
i engelska som andraspråk lyst med sin frånvaro sedan början på 1990-talet
(Leung 2005), vilket lett till att andelen lärare med specialkompetens i ämnet
har sjunkit till tre procent i vissa områden.

130
 Language Acquisition and Use in Multilingual Contexts

Även om en icke-diskriminerande agenda utan tvivel är av avgörande

betydelse för att uppnå social integration och likvärdig tillgång till utbildning
i mångkulturella och flerspråkiga samhällen, räcker det inte för att undanröja
minoritetselevernas språkliga hinder för skolframgång, vilket bekräftas i
internationella studier och rapporter. Mainstreammodellens underliggande
antagande om att utvecklingen av ett skolrelaterat andraspråk skulle ske helt
implicit och automatiskt bara genom att eleverna kommunicerar och deltar i
klassrumsinteraktionen utan särskilt språklig vägledning saknar stöd i forsk-
ningen. Storskaliga studier och kartläggningar (Thomas & Collier 2002;
Mohan et al. 2004; Stanat & Christensen 2006) har visat att det inte finns
några snabba, enkla sätt att eliminera språkliga skillnader och tillfredsställa
minoritetselevers språkliga behov. Frånvaron av systematiskt språkligt stöd
utgör alltså utan tvivel ett hinder för dessa elevers skolframgång.
I de jämförande studier där utbildningsresultat och skolsituationen för
elever med utländsk bakgrund analyserats i ett antal OECD-länder, däribland
Sverige (Christensen & Stanat 2007), har man funnit ett antal faktorer som
avgörande för framgångsrik utbildning för denna elevgrupp. Här betonas
bl.a. vikten av långsiktighet och kontinuitet liksom systematiskt och struk-
turerat språkligt stöd i enlighet med kursplaner baserade på centralt utveck-
lade styrdokument, dvs. riktad och specialiserad andraspråksundervisning.
Den kan genomföras som en del av, eller delvis åtskild från övrig under-
visning viss tid per dag eller vecka utifrån de enskilda elevernas behov och
förutsättningar, men bör erbjudas med en schemaläggning som möjliggör
full delaktighet i övrig undervisning. Exempel på denna typ av andra-
språksprogram finns t.ex. i Kanada och Australien, dvs. länder som enligt
internationella jämförelser i t.ex. PISA-studierna är bland dem som uppvisar
bäst resultat för elever med andra modersmål än majoritetsspråket. Här
spelar också, till skillnad från Storbritanniens mainstreamprogram, ESL-
lärarna en central roll.
Här är det självklart den professionella andraspråksläraren som bär det
yttersta ansvaret för att utveckla de flerspråkiga elevernas förutsättningar att
möta de språkliga kraven i olika ämnen genom en professionell, behovsstyrd
och ämnesintegrerad språkundervisning. I detta arbete spelar också moders-
målsundervisningen i den mån den erbjuds en viktig roll både som verktyg
för kunskapsutveckling och som länk och brygga mellan skolan och elev-
ernas övriga erfarenhetsvärldar.

Flerspråkiga elevers skolframgång – en

angelägenhet för alla lärare
Samtidigt betonar även förespråkare för differentiering och separata program
vikten av att alla lärare engageras i arbetet med att utveckla en långsiktig och
systematisk språk- och kulturmedveten undervisning i alla ämnen. Det för-

131
Language Acquisition and Use in Multilingual Contexts

utsätter naturligtvis omfattande kompetensutvecklingsinsatser för de

verksamma lärarna och en betydande satsning på dessa frågor i lärar-
utbildningarna som i Sverige, trots de senaste tjugo, trettio årens minst sagt
dramatiska demografiska utveckling, fortfarande utexaminerar lärare med
mycket begränsad beredskap inför skolans språkliga och kulturella
mångfald, som om ingenting hade hänt. Synen på ämneslärares profess-
ionalitet måste alltså vidgas till att omfatta även språkliga dimensioner på
lärande och på de ämnen de undervisar i för att rimma med de krav som
ställs på lärare i den mångspråkiga skolan.
Ett resurstänkande i förhållande till den språkliga och kulturella
mångfalden i den svenska skolan måste alltså med nödvändighet också om-
fattas av hela skolans personal för att få effekt. Det är uppenbart att den
komplexa mångfald i förhållande till både språk, etnicitet, kön och klass som
elever i dagens skola uppvisar, förutsätter en helt annan beredskap än
tidigare hos alla lärare. Det gäller alltså inte bara språklärare. Med tanke på
att det inte är i språklärarnas utan i andra lärares klassrum som eleverna
tillbringar den största delen av sin skoldag och att det är där skolframgången
avgörs, kan frågor kring flerspråkiga elevers språk- och kunskapsutveckling
inte betraktas som ett särintresse eller en angelägenhet endast för lärare i
svenska som andraspråk och modersmål. Språk- och kunskapsutveckling
utgör delar av en och samma process, vilket betyder att alla lärare spelar en
avgörande roll för den språkutveckling som krävs för lärandet i skolans olika
ämnen. I den tidigare refererade OECD-studien om invandrade elevers skol-
situation i Sverige (Taguma et al. 2010:8) slår man också fast att utbildning
av alla lärare och skolledare för språklig och kulturell mångfald är av högsta
prioritet för den svenska skolan.
I detta avseende återstår det mycket att göra i den svenska skolan där
påfallande få utbildningsinsatser i form av kompetensutveckling kring fler-
språkiga elevers språk- och kunskapsutveckling och allmänna skolsituation
genomförts och där lärarutbildningen i många avseenden fortfarande präglas
av en monokulturell och enspråkig syn på skolan.

Varierande behov
De elever som enligt Skolverkets definition karakteriseras som elever med
utländsk bakgrund är en på alla sätt heterogen grupp som för närvarande ut-
gör ca 20 procent av alla elever i den svenska skolan. Här återfinns elever
med högst olika förutsättningar och behov som sinsemellan uppvisar fler
skillnader än likheter; de har en varierande språklig, kulturell och social
bakgrund, är i skiftande åldrar och har tillbringat olika lång tid i Sverige –
många är födda i landet medan andra nyligen anlänt. De lever dessutom i
integrerade såväl som segregerade bostadsområden och har precis som alla
andra elever skiftande personligheter, intressen, drömmar och talanger.

132
 Language Acquisition and Use in Multilingual Contexts

Många har också svåra upplevelser i samband med krig och uppbrott bakom
sig. Trots denna stora variation har dessa elever dock en viktig gemensam
nämnare som gör att deras skolsituation skiljer sig från andra elevers, näm-
ligen att de i hemmet och i sin närmiljö vuxit upp med andra språk än sven-
ska. Men även i detta sammanhang är skillnaderna mellan olika elever stora.
Många elever kommer från miljöer där flera språk, och kanske även svenska,
använts sida vid sida ända sedan de var mycket små. Andra kommer från
språkligt mer homogena miljöer, vilket kan innebära att de endast har haft en
ganska sporadisk kontakt med svenska före skolstarten. Vissa elever talar ett
språk med ena föräldern, ett annat med den andra och kanske svenska med
syskonen; andra talar ett och samma språk hemma och svenska med vissa
eller alla kompisar och i skolan. Flerspråkigheten tar sig alltså många ut-
tryck, vilket bl. a. innebär att flerspråkiga elever behärskar sina språk i vari-
erande utsträckning, förhåller sig till dem på skiftande sätt och använder dem
i olika sammanhang (Håkansson 2003). Det betyder också att elevernas för-
utsättningar att kommunicera och lära på svenska, precis som deras behov av
särskilda språkliga insatser i skolan, är högst varierande. Detta ställer mycket
höga krav på skolornas förmåga att kartlägga och diagnostisera elevernas
språkfärdigheter ur en rad olika aspekter liksom på deras beredskap att er-
bjuda effektiva och språkutvecklande insatser i enlighet med de enskilda
elevernas högst varierande behov. Kartläggningar av flerspråkiga elevers
språkfärdigheter bör naturligtvis även omfatta modersmålet. Det är också
viktigt att påpeka att omfattningen och inriktningen på de språkliga insats-
erna också måste anpassas efter de språkliga krav som elever på olika stadier
möter i den övriga undervisningen.
I detta avseende ger de nuvarande styrdokumenten mycket begränsad
vägledning. Den parallellism mellan svenskämnena som tidigare refererats
till kan till och med utgöra ett hinder för skolornas beredskap inför detta vik-
tiga arbete. Här krävs en helt annan typ av kursplaner som definierar och av-
gränsar nivåer i andraspråket i förhållande till typiska målgrupper av elever.
Dessa nivåer måste vara fastställda utifrån kriterier och progressioner, bland
annat med tydlig koppling och relevans till skolrelaterade språkfärdigheter
på olika stadier. Exempel på denna typ av nivåbeskrivningar från den
kanadensiska kontexten återfinns i Alberta Government (2013).

Utgångspunkter för språkutvecklande insatser

Det är tämligen uppenbart mot bakgrund av de förhållanden vi behandlat
ovan att djupgående förändringar behövs om den svenska skolan ska förmå
tillvarata flerspråkiga elevers potential och ge dem en likvärdig och rättvis
undervisning. I det följande presenteras några tankar om i vilken riktning
sådana förändringar skulle kunna genomföras. Dessa tankar utarbetades ur-
sprungligen under vintern 2011/12 inför ett möte med politiker och tjänste-

133
Language Acquisition and Use in Multilingual Contexts

män vid Utbildningsdepartementet om flerspråkiga elevers skolsituation.18

De har därefter vidareutvecklats under ca ett års tid genom organiserade
kontakter med grupper av lärarutbildare, forskare, utbildningsadministratörer
och politiker. Grundtanken som förslaget bygger på är att behoven av språk-
utbildning för nyanlända elever skiljer sig starkt från de behov som finns hos
elever som är födda i Sverige eller som anlänt hit mycket tidigt i livet. Den
senare gruppens svårigheter i svensk skola och specifika behov av språkut-
bildning har uppmärksammats allt mer under det senaste decenniet i takt
med att heterogeniteten inom gruppen elever med annat modersmål än
svenska blivit alltmer uppenbar. Medan det torde te sig helt naturligt för
elever som nyligen kommit till landet att de behöver undervisning i svenska
som andraspråk, är det som vi visat ovan förenat med många komplikationer,
bland annat i relation till vad som kan kallas legitimitet eller trovärdighet, att
erbjuda elever som är födda i landet undervisning i ett sådant ämne. Dessa
elever har i allmänhet redan utvecklat en svenska som fungerar utan problem
i de flesta interaktiva sammanhang, men har i många fall ändå behov av en
fortsatt riktad språksatsning.

Utbildningsarrangemang för nyanlända elever

För elever som anlänt till Sverige nära skolstarten eller under skoltiden ser vi
följande arrangemang som ändamålsenliga:

1. Under den allra första tiden bör förberedelseklass eller motsvarande med
successiv inslussning i ordinarie undervisning utgöra normen. Vi föreställer
oss att det måste finnas en bortre tidsgräns för deltagande och föreslår att en
elev kan tillhöra en förberedelseklass under högst ett år. Under tiden i för-
beredelseklass ges eleven intensiv undervisning i ämnet svenska som andra-
språk, som bör vara organiserat i väl avgränsade och tydligt beskrivna nivå-
er, och eleven förväntas då utifrån sina individuella förutsättningar klara en
eller flera av dessa nivåer. För att undvika att en kunskapslucka skapas gen-
temot enspråkiga elever är det viktigt att ämnesundervisning ges med alla
medel som står till buds, varav studiehandledning och undervisning på
modersmålet under denna första tid bör vara den starkaste komponenten.
Men särskilt anpassad ämnesundervisning på svenska bör också ingå och
successivt öka. Ett erbjudande om kontinuerlig modersmålsundervisning
under tiden i förberedelseklass är också viktigt utifrån perspektivet att
modersmålet behöver utvecklas för att även fortsatt utgöra ett effektivt
instrument för kunskapsinhämtande, åtminstone fram till dess att behärsk-
ningen av svenska räcker till för att kunskap framgångsrikt ska kunna till-
ägnas via det språket.
18
Mötet var initierat av Inga-Lena Rydén, Nationellt centrum för svenska som andraspråk.

134
 Language Acquisition and Use in Multilingual Contexts

2. Efter tiden i förberedelseklass, dvs. när eleverna helt gått över till ord-
inarie övrig undervisning, får de fortsatt undervisning i ämnet svenska som
andraspråk. Som antyddes i punkten ovan ska ämnet svenska som andra-
språk då vara organiserat i ett antal väl avgränsade och åldersrelaterade
nivåer med den slutliga målsättningen att möjliggöra lärande och social int-
eraktion på svenska utan avgörande språkliga hinder. De bör alltså i detta
skede inte bara ha uppnått förmåga att klara av vardagliga sociala situa-
tioner; de ska helt enkelt behärska svenska enligt de kriterier som fastställts
för den översta nivån. Det är klart att olika individer når en sådan nivå i olika
takt, men det är helt orealistiskt att föreställa sig att ens någon kan uppnå den
under året i förberedelseklass, även om undervisningen i svenska som andra-
språk där ska vara intensiv. En organisering av ämnet i väl avgränsade nivåer
kräver att adekvata instrument utvecklas för professionell bedömning av
relevanta språkfärdigheter. Ämnet svenska som andraspråk bör i normalfallet
vara dessa elevers svenskämne tills den översta nivån uppnåtts, men man
kan tänka sig att de i individuella fall kan följa både detta ämne och ämnet
svenska. En stark modersmålskomponent med språkutveckling och studie-
handledning på modersmålet bör också ingå.

3. När eleverna uppnått den högsta och avslutande nivån i ämnet svenska
som andraspråk, ska de delta i ämnet svenska. Här är det viktigt att påpeka
att detta ämne i förhållande till hur det ser ut idag måste revideras och
anpassas till att en så stor andel elever i svenska klasser har flerspråkig bak-
grund. Eftersom det dock är så att många elever fortfarande på denna nivå
kan behöva ytterligare riktat språkligt stöd för att klara skolarbetet väl, ska
man här erbjuda deltagande i en verksamhet som skulle kunna kallas svenska
för studieframgång, språkstudio, språkhandledning, språkgym – namnfrågan
återstår att lösa. Vi har själva laborerat med etiketten språkgym huvud-
sakligen för att pedagogisera några av de karakteristiska drag som vi menar
en sådan språkstödjande verksamhet måste uppfylla. Hur som helst är det
viktigt att beteckningen kan påverka potentiella elever positivt och vara
attraherande. Även här bör en stark modersmålskomponent ingå, eventuellt
direkt kopplat till verksamheten.

Utbildningsarrangemang för elever som är födda i

Sverige eller har anlänt tidigt i livet
Denna grupp elever bör få sin svenskundervisning i (det reformerade) ämnet
svenska. De ska alltså inte ha undervisning i svenska som andraspråk, annat
än möjligtvis i individuella undantagsfall, där man trots att man bott i
Sverige under hela förskoletiden inte alls varit i kommunikativ kontakt med
svenska språket. Det kan i sådana fall t.ex. vara frågan om att man behöver
delta i ämnet svenska som andraspråk på någon av de högre nivåerna. Detta

135
Language Acquisition and Use in Multilingual Contexts

måste bedömas från fall till fall med hjälp av framtagna diagnostiska instru-
ment. Utöver den ordinarie svenskundervisningen bör denna grupp erbjudas
det ovan nämnda språkprogrammet svenska för studieframgång språkstudio
etc. Det är av största vikt att ett sådant erbjudande kommer till stånd redan
från skolstart vid en förändring av den rådande situationen, eftersom dessa
elever vid en reformerad utbildning i svenska som andraspråk, som enbart
vänder sig till elever med relativt kort tid i Sverige, inte längre kommer att
vara aktuella för undervisning i svenska som andraspråk. Även denna elev-
grupp ska erbjudas en stark modersmålskomponent.

Svenska för studieframgång, språkstudio, språkgym etc.

När det gäller denna verksamhet föreställer vi oss ett kvalificerat, varierat,
språkutvecklande program lett av högt kvalificerade och specialutbildade
lärare i ämnena svenska som andraspråk och svenska samt välutbildade
modersmålslärare. Här ska man även ha tillgång till kvalificerade ämnes-
lärare i en referensgrupp. Verksamheten ska vara ett erbjudande som elev-
erna frivilligt väljer, vilket torde vara en förutsättning för att skapa legiti-
mitet och motivation. Den behöver inte, och ska kanske inte, enbart vända
sig till flerspråkiga elever. För att undvika att den stämplas som segre-
gerande är det en fördel om alla elever får del av erbjudandet. Det är viktigt
att verksamheten är nära kopplad till skolarbetet i övrigt, och det ska också
kunna demonstreras att den har effekter på elevernas skolframgång. Vad vi
vill peka på med att använda termen språkgym är för det första att aktivi-
teterna ska vara individanpassade, så att varje elevs specifika behov beaktas.
De som ansvarar för verksamheten ska vara kvalificerade att med hjälp av
centralt utvecklade bedömningsinstrument kunna avgöra vilka språkliga
dimensioner som behöver utvecklas hos enskilda elever, eller annorlunda
uttryckt, de ska kunna bedöma varje elevs starka och svaga sidor i språket.
Vidare ska verksamheten kunna erbjuda eller sätta in ett helt spektrum av
olika program som är specifikt avpassade för den enskildes behov. Man kan
likna dessa vid stationerna i ett gym. Dessa program bör ofta vara dator-
baserade och ska i någon utsträckning kunna genomföras på distans. Bland
de typer av aktiviteter som programmen kan omfatta är följande: Utveckling
av repertoarer av skolrelevanta muntliga och skriftliga genrer; utveckling av
vissa specifika skriftspråkdrag, t.ex. grammatiska metaforer eller andra
skriftspråkliga uttryckssätt/ordförråd, både ämnesspecifikt och ämnesöver-
gripande; systematisk utveckling av ordförrådets bredd och djup; utveckling
av metaforik/bildspråk, idiomatik, kollokationer etc. i mer formella register;
utveckling av läsfärdighet och läsförståelse med fokus på ämnestexter;
utveckling av hörförståelse för olika innehåll i skilda situationer, t.ex. träna
specifikt på förståelse av svenska i bullriga miljöer (träna på simultan-
kapacitet); utveckling av studieteknik, lärstrategier, anteckningsteknik
(simultant lyssnande och bearbetning av information); träning av tal inför
publik, retorik och argumentation; litteracitetsaktiviteter på modersmålet.

136
 Language Acquisition and Use in Multilingual Contexts

Kvalitetskrav
För att systemet med ett nytt ämne svenska som andraspråk och en ny
kompletterande språkhandledning ska fungera måste det uppfylla höga krav
på legitimitet och professionalitet. Systemets legitimitet bygger på att det
innehåller evidensbaserade bedömningsinstrument som kan hantera många
dimensioner av språkfärdighet. Att dessa instrument är evidensbaserade
innebär att bedömningarna blir korrekta och har förutsägningsvärde i för-
hållande till det språk som krävs för lärande i skolans ämnen. Själva under-
visningen måste successivt vara alltmer kopplad till formellt språkbruk i
skolans ämnesundervisning. Detta gäller för både högre nivåer i svenska som
andraspråk och den kompletterande språkhandledningen.
Vidare måste skolor ha tillgång till informationsunderlag för dialoger med
föräldrar, elever och övrig skolpersonal. Det är de professionella och special-
utbildade lärarna som ska genomföra dessa samtal, men de behöver stöd för
att hantera den argumentation som de kontinuerligt måste föra. Systemet
måste också uppfylla högt ställda krav på professionalitet, vilket dels innebär
att kunskapen om flerspråkighet och lärande behöver öka hos skolledare och
alla lärare i skolan, dels att det uteslutande måste vara lärare med special-
kompetens i svenska som andraspråk som undervisar i ämnet och som ingår i
och leder den kompletterande språkverksamheten. Detta kräver utveckling
av lärarutbildningar på alla nivåer och speciellt av lärarutbildningen i
svenska som andraspråk. Den måste bättre än nu förbereda lärare i svenska
som andraspråk för att kunna undervisa elever på alla stadier i samtliga tre
spår i utbildningen (förberedelseverksamhet, svenska som andraspråk upp
till avancerad nivå, språkhandledning). Dessutom måste all lärarutbildning
förändras så att den blir bättre anpassad till den språkliga och kulturella
mångfalden i dagens skola.
Krav på kvalitet innebär också flexibilitet. Det måste finnas utrymme för
alternativa organisatoriska lösningar som speglar den variation och hetero-
genitet som präglar behoven hos eleverna. Lösningarna måste också beakta
skolornas olika förutsättningar och elevunderlag. Skolverket har här en
viktig uppgift i att stödja skolorna med modeller för alternativa lösningar till
organisation, schemaläggning och timplanering.
När det gäller betygsfrågan ser vi dock inga hinder till att betyg i svenska
som andraspråk på den aktuella nivån ersätter betyg i svenska så länge man
deltar i undervisning i svenska som andraspråk. Rimligtvis bör godkänt
betyg på den aktuella nivån i svenska som andraspråk också ge gymnasie-
behörighet varefter studierna i svenska som andraspråk kan fullföljas i gym-
nasiet. För sent anlända elever och för elever nyanlända till gymnasiet bör
naturligtvis undervisningen i svenska som andraspråk liksom för övriga
stadier anpassas efter de språkliga krav som ställs för studier i övriga ämnen
på dessa nivåer. Dessutom måste timplanen utökas och eventuellt andra spe-
ciella åtgärder vidtas för att dessa elever ska kunna nå minst godkänt

137
Language Acquisition and Use in Multilingual Contexts

gymnasiebetyg i svenska och fullständiga slutbetyg i övrigt inom ramen för

sin gymnasieutbildning.

Attitydförändringar
I ett vidare perspektiv kan det system vi föreslår inte fungera utan att vissa
attityder som råder i skolan och i samhället förändras. Det krävs helt enkelt
ett aktivt arbete mot bristtänkande och den utbredda stigmatiseringen av fler-
språkiga elever som lågpresterande. PISA-studierna visar att elever med ut-
ländsk bakgrund över lag uppvisar högre motivation, mer positiva attityder
gentemot skolan och i högre grad har planer på högskoleutbildning än elever
med majoritetsbakgrund (OECD 2007). Många forskare betonar också
vikten av att möta dessa elever med höga förväntningar (Cummins 2001).
Skolan och samhället behöver bli bättre på att tillvarata de resurser som
ligger i ett utbrett språkkunnande och rika erfarenheter av andra kulturer och
samhällen, vilket i sin tur förutsätter ett aktivt arbete för ett resurstänkande
omkring de flerspråkiga elevernas varierande språkliga och kulturella
erfarenheter. Därutöver krävs större reflektion över den idealisering av
enspråkighet som norm och homogen svenskhet som fortfarande präglar
samhället och skolan.

Slutord
Som tidigare framhållits ska de förslag till förändringar av skolans språk-
satsningar för flerspråkiga elever som här föreslås ses som möjliga utgångs-
punkter för en vidare diskussion av en reformerad språkutbildning snarare än
som en färdig, fullständig och genomarbetat modell. I detta skede återstår
många viktiga aspekter att dryfta och närmare utreda inte minst när det gäller
anpassning till förhållanden på enskilda skolor, schemaläggning och tim-
planer – frågor som vi varken har möjligheter eller de rätta förutsättningar att
gå närmare in på i detta sammanhang.
Denna text har publicerats i likartad form i följande konferensvolym: Olofsson, M. (red.),
2013. Symposium 2012. Lärarrollen i svenska som andraspråk. Stockholm: Stockholms
universitets förlag.

Litteratur
Alberta Government, Canada 2013. Benchmarks, Strategies and Resources for
Teachers of English Language Learners. http://www.learnalberta.ca/content/
eslapb/printable_benchmarks.html.
Carlson, M. 2009. Flerspråkighet i lärarutbildningen – Ett perspektiv som saknas.
Utbildning & Demokrati, 18 (2). 39–66. TEMA ”Flerspråkighet”.
Christensen, G. & Stanat, P. 2007. Language Policies and Practices for Helping
Immigrants and Second-Generation Students Succeed, The Transatlantic Task

138
 Language Acquisition and Use in Multilingual Contexts

Force on Immigration and Integration: MPI, Washington DC.

Creese, A. & Leung, C. 2003. Teachers’ discursive constructions of ethnolinguistic
difference: professional issues in working with inclusive policy. Prospect, 18
(2). 3–19.
Cummins, J. 2001. Negotiating Identities: Education for Empowerment in a Diverse
Society. 2nd ed. Los Angeles: California Association for Bilingual Education.
Economou, C. 2007. Gymnasieämnet Svenska som andraspråk – behövs det?
Malmö Högskola.
Elmeroth, E. 2008. Etnisk maktordning i skola och samhälle. Lund: Student-
litteratur.
Fraurud, K. & Boyd, S. 2006. The native-non-native speaker distinction and the
diversity of linguistic profiles of young people in Swedish multilingual urban
contexts. I F. Hinskens (red.), Language Variation – European Perspectives.
53–69. Amsterdam: John Benjamins.
Fridlund, L. 2011. Interkulturell undervisning – ett pedagogiskt dilemma talet om
undervisning i svenska som andraspråk och i förberedelseklasser. Göteborgs
universitet.
Granstedt, L. 2010. Synsätt, teman och strategier – några perspektiv på mång-
kulturella frågor i skolan i ett praktiknära projekt. Umeå universitet.
Gruber, S. 2007. Skolan gör skillnad. Etnicitet och institutionell praktik. Linköping
Studies in Arts and Science 387. Linköpings universitet.
Haglund, C. 2005. Social interaction and identification among adolescents in multi-
lingual suburban Sweden. A study of institutional order and sociocultural
change. Centrum för tvåspråkighetsforskning, Stockholms universitet.
Harklau, L. 1994. ESL versus mainstream classes: Contrasting L2 learning environ-
ments. TESOL Quarterly, 28. 241–272.
Hyltenstam, K. & Lindberg, I. 2004. Förord i K. Hyltenstam & I. Lindberg (red.),
Svenska som andraspråk – i forskning, undervisning och samhälle. 11–22.
Lund: Studentlitteratur.
Hyltenstam, K. & Milani, T. 2012. Flerspråkighetens sociopolitiska och sociokultur-
ella ramar. I K. Hyltenstam, M. Axelsson & I. Lindberg (red.), Flerspråkighet –
en forskningsöversikt. 17–152. Vetenskapsrådets rapportserie 5:2012.
Hyltenstam, K. & Tuomela, V. 1996. Hemspråksundervisningen. I K. Hyltenstam
(red.), Tvåspråkighet med förhinder? Invandrar- och minoritetsundervisning i
Sverige. 9–109. Lund: Studentlitteratur.
Håkansson, G. 2003. Tvåspråkighet hos barn i Sverige. Lund: Studentlitteratur.
Jørgensen, J. 2008. Polylingual Languaging around and among children and adole-
scents. International Journal of Multilingualism, 5 (3). 161–176.
Lahdenperä, P. 1997. Invandrarbakgrund eller skolsvårigheter? En textanalytisk
studie av åtgärdsprogram för elever med invandrarbakgrund. Lärarhögskolan i
Stockholm: HLS förlag.
Leung, C. 2005. English as an additional language policy: issues of inclusive access
and language learning in the National Curriculum. Prospect, 20 (1). 95–113.
Lewis, A. 2001. There is no “race” in the schoolyard: Color-blindness ideology in an
(almost) all-white school. American Educational Research Journal, 38. 781–
811.
Lindberg, I. 2007. Multilingualism and Education – a Swedish Perspective. I M.
Carlson, F. Gök & A. Rabo (red.), Reflections on Education in Multicultural
Societies Turkish and Swedish Perspectives. 73–92. Stockholm: Swedish

139
Language Acquisition and Use in Multilingual Contexts

Research Institute in Istanbul.

Lindberg, I. 2010. Fostering multilingualism i Swedish schools – intentions and
realities. I R. Källström & I. Lindberg (red.), Young Urban Swedish. Variation
and change in multilingual settings. 149–172. Göteborgs universitet:
Göteborgsstudier i nordisk språkvetenskap 14.
Lindberg, I. 2011. Språk och lärande i en mångspråkig skola. Forskning om
undervisning och lärande 6: Läraryrkets interkulturella dimensioner.
http://www.lararforbundet.se/web%5Cshop2.nsf/webDescription/78F536A64D
D4F091C125790B00517D80/$file/SAF_Forskning_nr_6_11.pdf 2013-02-25
May, S. 2001. Language and Minority Rights: Ethnicity, Nationalism and the
Politics of Language. Harlow, England: Longman.
Mc Kay, S. & Warshauer Freedman, S. 1990. Language Minority Education in Great
Britain: A Challenge to Current U.S. Policy. TESOL QUARTERLY, 24 (3). 385–
405.
Milani, T. 2007. Debating Swedish. Language. Politics and Ideology in Contemp-
orary Sweden. Centrum för tvåspråkighet, Stockholms universitet.
Mohan, B., Leung, C. & Davison, C. 2004. English as a second language in the
main stream. Teaching, learning and identity. London, New York: Longman.
Myndigheten för skolutveckling 2004. Kartläggning av svenska som andraspråk.
OECD 2007. PISA 2006: Science Competencies for Tomorrow’s World, 1. Paris:
OECD.
Otterup, T. 2005. “Jag känner mig begåvad bara”. Om flerspråkighet och identitets-
konstruktion bland ungdomar i ett multietniskt förortsområde. Göteborgs
universitet.
Pacheco, M. 2010. Performativity in the bilingual classroom: The plight of English
learners in the current reform context. Anthropology & Education Quarterly, 41
(1). 75–93.
Parszyk, I-M. 1999. En skola för andra. Minoritets elevers upplevelser av arbets-
och livsvillkor i grundskolan. Stockholm: HLS Publications.
Pavlenko, A. 2002. We have room for but one language here: Language and national
identity inthe US at the turn of the 20th century. Multilingua, 21. 163–196.
Skolinspektionen 2010. Språk och kunskapsutveckling för barn och elever med
annat modersmål än svenska. Kvalitetsgranskning. Rapport 2010:16
Skolverket 2005. Elever med utländsk bakgrund. Stockholm: Fritzes. Stockholm.
Skolverket 2011. Skolverkets lägesbeskrivning 2011. Del 1 – Beskrivande data.
Förskoleverksamhet, skolbarnsomsorg, skola och vuxenutbildning. Rapport 363.
Skolverket 2011/12a. Lgr 11: Kursplan i svenska för grundskolan.
Skolverket 2011/12b. Lgr 11: Kursplan i svenska som andraspråk för grundskolan.
Sridhar, K. 1996. Societal multilingualism. I S. McKay & N. Hornberger (red.),
Sociolinguistics and Language Teaching. 47–70. Cambridge University Press.
Stanat, P. & Christensen, G. 2006. Where Immigrant Students Succeed. A Compar-
ative Review of Performance and Engagement in PISA. Paris: OECD.
Stroud, C. 2003. Retheorising a politics of language and education. I K. Fraurud &
K. Hyltenstam (red.), Multilingualism in global and local perspectives. 17–30.
Centrum för tvåspråkighet, Stockholms universitet och Språkforskningsinstitutet
i Rinkeby, Stockholm stad och Stockholms universitet.
Taguma, M., Kim, M., Brink, S. & Teltemann, J. 2010. OECD Reviews of Migrant
Education: Sweden. Paris: OECD.
Talmy, S. 2008. The cultural productions of the ESL student at Tradewinds High:

140
 Language Acquisition and Use in Multilingual Contexts

Contingency, multidirectionality, and identity in L2 socialization. Applied

Linguistics, 29. 619–644.
Talmy, S. 2009. ”A very important lesson”: Respect and the socialization of order(s)
in high school ESL. Linguistics and Education, 20. 235–253.
Thomas, W. & Collier, V. 2002. A National Survey of School Effectiveness for
Language Minority Students’ Long-Term Academic Achievement: Executive
Summary. Santa Cruz, CA: Crede.
Tingbjörn, G. 2004. Svenska som andraspråk i ett utbildningspolitiskt perspektiv –
en tillbakablick. I K. Hyltenstam & I. Lindberg (red.), Svenska som andraspråk
– i forskning, undervisning och samhälle. 743–763. Lund: Studentlitteratur.
Valencia, R. 1997. Conceptualizing the notion of deficit thinking. I R. Valencia
(red.), The evolution of deficit thinking: Educational thought and practice. 1–12.
Washington, DC: Falmer.
Westling Allodi, M. 2002. Equal Opportunities in Educational Systems: the case of
Sweden. European Journal of Education, 42 (1). 133–146.
Vollmer, G. 2000. Praise and stigma: Teachers’ constructions of the ’typical ESL
student’. Journal of Intercultural Studies, 21 (1). 53–66.

141
Developmentally Moderated Transfer and
the role of the L2 in L3 acquisition19

Manfred Pienemann
Jörg-U. Keßler
Anke Lenzing

Introduction
This chapter focuses on one specific aspect of the Developmentally
Moderated Transfer Hypothesis (Pienemann et al. 2005), namely the role of
the L2 in L3 acquisition. The research presented here was prompted by the
L2 transfer hypothesis put forward by Bohnacker (2006) and Bardel & Falk
(2007). According to this hypothesis learners transfer features from the L2 to
the L3, but not from the L1 to the L3. This proposal is partly in conflict with
the Developmentally Moderated Transfer Hypothesis which predicts that
learners transfer features from the L1 or the L2 to the new language when
they are developmentally ready to acquire the features to be transferred, but
not before.
The articles by Bohnacker (2006) and Bardel & Falk (2007) are attempted
rebuttals of Håkansson, Pienemann & Sayehli’s (2002) work on L1 transfer
and aspects of the underlying theory: Processability Theory (Pienemann
1998). Håkansson et al. (2002) presented empirical evidence showing that
Swedish learners of L2 German do not transfer V2 at the initial state al-
though both are V2 languages. Bohnacker (2006) and Bardel & Falk (2007)
claim that the non-transfer of V2 is due to the influence of the L2. They
further claim to have shown in their own study that the initial L3 word order
is determined by the L2, irrespective of the structure of the L1 and in-
dependently from constraints on processability.
In their response to Bohnacker (2006) Pienemann & Håkansson (2007)
demonstrated that Bohnacker’s informants had reached an advanced level of

19
The authors would like to thank Gisela Håkansson (Lund University) and Bruno Di Biase
(University of Western Sydney) for their useful comments on this paper.

142
 Language Acquisition and Use in Multilingual Contexts

acquisition and that this set of data is not suitable to test hypotheses about
transfer at the initial state.
In this chapter we review the study by Bardel & Falk (2007) and present
the gist of an extensive replication of this study. We show that Bardel &
Falk’s study is based on a very limited database and on theoretical concepts
that lack validity, in particular the notion “strongest L2” which is crucial to
Bardel & Falk’s approach. Our replication study shows that the initial L3
word order and the initial position of negation is neither determined by the
L1 nor by the L2 and that it can be predicted on the basis of processability.

Developmental Moderation of Transfer and L2

transfer in L3 acquisition

The Developmentally Moderated Transfer Hypothesis

The Developmentally Moderated Transfer Hypothesis (DMTH) is a com-
ponent of Processability Theory (Pienemann 1998); it was spelt out in detail
with empirical support in Pienemann et al. (2005). The basic idea behind the
DMTH is the following: given the architecture of human language
processing, the L2 formulator relies on L2-specific lexical information that is
essential for grammatical processing. The learner has no a priori knowledge
of L2-specific lexical information such as the diacritic features of the L2
lexical categories. Therefore full transfer of the L1 at the initial state would
lead to very unwieldy hypotheses. Instead, it is assumed that the L2 lexicon
is annotated gradually and that this, together with the development of L2
processing procedures, permits the learner to build up the L2 in stages. As
illustrated in Figure 1, features of the L1 will be able to be utilized once the
developing L2 system can process them. For this reason all learners are
predicted to follow the same developmental trajectory irrespective of the L1,
and positive and negative effects of the L1 will be visible at predictable
points of development. In other words, the DMTH does not rule out transfer
altogether. Instead, it assumes a selective role of transfer in SLA.

143
Language Acquisition and Use in Multilingual Contexts

Figure 1. Developmentally Moderated Transfer.

Håkansson, Pienemann and Sayehli (2002)

The study by Håkansson, Pienemann & Sayehli (2002) provides empirical
support for the DMTH. The study focuses on the acquisition of German by
Swedish school children. The L1 and the L2 are typologically close and
share the following word order regularities in affirmative main clauses:
SVO, adverb fronting (ADV) and V2 (verb-second) after ADV.
The following examples of V2 in German and Swedish illustrate the word
order similarity in the two languages:
(1) German = V2 Dann kauft das Kind die Banane
(2) Swedish = V2 Sen köper barnet bananen
(Then buys the child the banana)

Note that in German and Swedish sentences without V2 are ungrammatical:

(3) * Dann das Kind kauft die Banane
(Then the child buys the banana)

Table 1 gives an overview of the acquisition of key word order patterns in

the three Germanic languages that are relevant in the context of this chapter.
These developmental patterns are displayed in relation to the corresponding
PT levels.

144
 Language Acquisition and Use in Multilingual Contexts

Table 1. L2 syntactic development in three Germanic languages (selected

structures).
PT level ESL syntax Swed. L2 syntax GSL syntax
(Meisel et al. 1981)
6 Cancel INV --- V-Final
5 Do2nd, V2 V2
Aux2nd
4 Y/N inv, --- V-Front
copula inv
3 ADV-1st, ADV-1st, ADV-1st,
WH-1st, WH-1st WH-1st
Do-1st
2 SVO SVO SVO
1 invariant forms invariant forms invariant forms

The results of this study are summarized in Table 2 below which treats all
learner samples as parts of a cross-sectional study. Therefore Table 2 re-
presents an implicational analysis of the data which demonstrates that the
learners follow the sequence (1) SVO, (2) ADV and (3) INV. In other words,
ADV and INV are not transferred from the L1 at the initial state although
these rules are contained in the L1 and the L2. This implies that for a period
of time the learners produce the following constituent order which is un-
grammatical in the L1 as well as in the L2: * adverb+S+V+O.
This finding is consistent with the DMTH because the structures which
are identical in the two languages are not transferred at the initial state.
Under the transfer assumption one would have expected to find all obliga-
tory structures to be present in all samples, particularly V2. However, 10 of
the 20 samples consistently violate the V2 rule (i.e. * adverb+ S + V + O)
despite the marked ungrammaticality of the resulting structure.

Bohnacker (2006) and response

Bohnacker claims that the late acquisition of V2 in Håkansson et al.’s study
is due to transfer from English, the L2 of all learners in the sample. She
further claims that Swedes learning German as the first L2 start with V2
because they transfer this structure from the L1. In other words, Bohnacker
assumes full transfer from the L1 to the L2 and from the L2 to the L3 if there
is an L3. To support her claims she carried out a replication of Håkansson et
al.’s study. Bohnacker’s study is based on a group of six elderly Swedes half
of whom report never to have learned English or German. The informants
learnt German mostly in order to be able to communicate with their German-
speaking grandchildren. The other three learners had English as their L2 and
learnt German as L3. Bohnacker found quantitative differences between the

145
Language Acquisition and Use in Multilingual Contexts

two groups of learners. The group without L2 English showed a higher

accuracy in the use of V2 in German. However, Pienemann & Håkansson
(2007) demonstrated that all learners in Bohnacker’s study had already
acquired V2, and the data were not suitable to make any statements about
transfer at the initial state. It may be useful to summarise Pienemann &
Håkansson’s (2007) review to reconstruct the debate.
Table 2. German as L2 by Swedish learners. Implicational scale based on all learners
in the study by Håkansson et al. (2002).
Name SVO ADV INV
Gelika (year 1) + - -
Emily (year 1) + - -
Robin (year 1) + - -
Kennet (year 1) + - -
Mats (year 1) + - -
Camilla (year 2) + - -
Johann (year 1) + + -
Cecilia (year 1) + + -
Eduard (year 1) + + -
Anna (year 1) + + -
Sandra (year 1) + + -
Erika (year 1) + + -
Mateaus (year 2) + + -
Karolin (year 2) + + -
Ceci (year 2) + + -
Peter (year 2) + + -
Johan (year 2) + + +
Sandra (year 2) + + +
Zofie (year 2) + + +
Caro (year 2) + + +

Pienemann & Håkansson (2007) subjected Bohnacker’s data to a re-analysis

based on the statistics provided in her paper. The re-analysis was necessary
because Bohnacker contrasts her claim with Håkansson’s claim that Swedes
learning German as L2 start with canonical word order. Therefore Bohn-
acker’s analysis needs to be based on the same approach to data analysis and
the same acquisition criteria (i.e. implicational scaling and the emergence
criterion). As mentioned above, Bohnacker’s own analysis focuses on quan-

146
 Language Acquisition and Use in Multilingual Contexts

titative differences between learners. Pienemann & Håkansson (2007) re-

analysed Bohnacker’s data in form of an implicational analysis using the
emergence criterion (see Table 3).
Table 3 is laid out as follows. The first column states if German is the L2
or the L3 of the informant. The second column identifies the sample by in-
formant name and data point number. The column entitled ‘SVX’ states if
the sample contains examples of canonical word order (using the emergence
criterion) where ‘+’ means ‘acquired’. The column ‘ADV’ does the same for
structures with non-subjects in initial position. The column ‘SEP’ lists the
relative frequency of two verbs (aux+V) appearing in a non-adjacent posi-
tion (i.e. XVYV). This structure occurs in German, but not in Swedish. The
last column lists the relative frequency of V2 application. In other words, the
columns from SVX to V2 are arranged in the order of acquisition initially
identified by Meisel, Clahsen & Pienemann (1981) and subsequently found
in many SLA studies.
Table 3. Re-analysis of Bohnacker’s sample using the same criteria as in Håkansson
et al. (2002).
L2 or L3? Informant SVX ADV SEP (%) V2 (%)
L2 German Marta 1 + + 12 20
100
L2 German Marta 2 + + 12 100
L2 German Marta 3 + + 70 100
L2 German Algot 1 + + 30 100
L2 German Algot 2 --- --- --- ---
L2 German Algot 3 + + 85 95
L2 German Signe 3 + + 62 100

L3 German Rune 1 + + 8 55
L3 German Rune 2 + + 8 44
L3 German Rune 3 + + 76 58
L3 German Gun 1 + + 45 55
L3 German Gun 2 --- --- --- ---
L3 German Gun 3 + + 70 57
L3 German Ulf 3 + + 61 52

It is easy to see that all four target structures meet the emergence criterion
for all informants, no cell of the implicational table is empty (apart from
missing data for Algot 2 and Gun 2), no learners slip back; thus the scalabili-
ty of Table 3 is 100%. This means that all structures under discussion, inclu-
20
The figures for Marta 1+2 and Rune 1+2 are presented as averages of the two sessions by
Bohnacker.

147
Language Acquisition and Use in Multilingual Contexts

ding V2, had already been acquired at the first point of data collection. In
other words, all informants had acquired V2 (and all other relevant struc-
tures) at the beginning of the study. This is the strongest reason why the
study is not suitable to test the initial word order of Swedish first-time
learners of German. Given that the learners had already acquired all the
structures under investigation at the beginning of the study, including V2,
they are simply too advanced to make any statement about the initial state of
their interlanguages.
One might object to this conclusion about the level of acquisition of the
six learners in Bohnacker’s corpus on logical grounds, because full transfer
from Swedish would always imply that all structures contained in Table 3
need to be present from the start. However, this hypothetical possibility
would apply only to one subgroup of Bohnacker’s sample, the learners with
L1 Swedish and L2 German. For the other subgroup with L1 Swedish, L2
English and L3 German she predicted transfer from L2 to L3. Given that V2
is not part of English, these learners should not acquire V2 at the initial state.
However, Table 3 shows that all learners from this group also display clear
evidence of V2 in the first interview. Therefore, the full transfer assumption
is not compatible with the evidence presented in Bohnacker’s study.
Nevertheless, there is one striking difference between the L2 and the L3
group. Learners without exposure to English display a native level of perfor-
mance for V2, whereas learners with previous exposure to English do not.
This is highly compatible with the DMTH, which predicts that transfer will
not appear before the structure to be transferred can be processed by the
interlanguage system. However, when structures from the L1 or L2 are pro-
cessable, they may be transferred to the target language, and this may lead to
differential patterns of language use in groups of learners with different L1s
(or L2s). We ascertained above that all informants in Bohnacker’s corpus
have reached the acquisition level marked by ‘INV’. Therefore, V2 is readily
processable by all learners in this corpus, and the group without knowledge
of English can make recourse only to their knowledge of V2 that can be
transferred at this point of development, whereas the group with English as
the first L2 can transfer two competing rules that match the structural con-
dition for V2, i.e. either XVSY or XSVY. Therefore the given learning con-
dition facilitates the accuracy with which the L2 group uses V2 compared
with the L3 group.

Bardel and Falk (2007)

Bardel & Falk (2007) also carried out a study designed to refute the DMTH.
It may be useful to first consider their critique of the study by Håkansson et
al. (2002). We will then review Bardel & Falk’s study.

148
 Language Acquisition and Use in Multilingual Contexts

Håkansson et al. (2002) argue as follows about the role of L2 transfer in

their study. If the lack of V2 transfer from Swedish to German in their data
was due to the transfer of SVO from L2 English, then why were other fea-
tures of English not transferred, such as “adverb-first” (ADV) or “particle”
(e.g. put it on)? Bardel & Falk respond to this as follows. Adverbs in initial
position are missing only in a few of the learners in Håkansson et al.’s study,
and adverbs are optional anyway. So there could be any number of reasons
for their absence. They conclude that “...the absence of a part of speech in
oral production can hardly be taken as an argument against transfer” (2007:
466).
In our view, Bardel & Falk’s line of reasoning ignores the context in
which the analysis of word order constellations was conducted in Håkansson
et al.’s study, where the following word order rules were studied as part of a
distributional analysis of the interlanguages (ILs) of all 20 informants: SVO,
ADV, V2. The corresponding table from Håkansson et al. (2002) with the
full distributional analysis is repeated above as Table 2. As Håkansson et al.
pointed out, it shows a very strong implicational relationship of the three
rules (SVO, ADV and V2) with a scalability of 100%. In other words, it
shows the following strict implicational relationship: SVO>ADV>V2. This
is supported by a comparison of the development of those learners who were
recorded at two points in time with a one-year interval.
In light of these circumstances it is unlikely that for learners who display
the feature SVO only, the absence of ADV is pure coincidence, because as
part of their IL grammar that follows canonical word order they systema-
tically place adverbs and adverbials in final position as shown in (4):
(4) Der mann gehen nach hause.

In other words, the fact that adverbs and adverbials are not fronted in a sub-
set of the data is not merely a reflection of the presence or absence of these
structures or lexical categories in the sample. Instead, the learners in this
group systematically placed adverbs and adverbials in final position,
whereas the learners with ADV alternated between initial and final position
depending on pragmatic conditions.
Bardel & Falk (B&F) present an empirical study with two distinct sets of
data to support their L2 transfer hypothesis. Data set A is based on five
learners of Swedish as L3 who were exposed to a sequence of ten 45-minute
Swedish lessons. The typology of the learners’ languages can be summarised
as follows:
# learners L1 L2 L3
3 +V2 -V2 +V2
2 -V2 +V2 +V2

149
Language Acquisition and Use in Multilingual Contexts

This design permits the specific effect of the L2 to be tested empirically.

Data set B was set up in a similar manner with respect to the typology of the
learners’ languages, only this time the learners (n=4) were recorded in one-
off individual one-to-one lessons lasting about 45 minutes. According to
B&F (p. 470), the learners were “absolute beginners” with no prior know-
ledge of Swedish. Table 4 summarizes the typology of their languages:
Table 4. The learners and their knowledge of V2 languages, data collection B.
Learner21 Sex First language Second language Target language
EN4 F Swedish +V2 English Dutch +V2
EN5 M Swedish +V2 English Dutch +V2
D/G3 M Italian German/Dutch +V2 Swedish +V2
D/G4 M Albanian German +V2 Dutch +V2
Bardel & Falk (2007:472)

The study focused on the acquisition of negation. Using example sentences,

B&F provide an implicit distribution of the position of the negator for the
languages relevant in this study (with a focus on sentence negation in de-
claratives) which can be summarized as follows:
Table 5. Distribution of the position of the negator.
S V neg Swedish Dutch German
S cop neg A Swedish Dutch German English
S Aux neg V Swedish Dutch German English
S DO neg V English
S neg V Italian Hungarian Albanian
S neg cop A Italian Hungarian Albanian
S neg aux V Italian Hungarian Albanian

B&F distinguish between two types of languages in this list: (1) those with
preverbal negation (Italian, Hungarian and Albanian) and (2) those with
postverbal negation (Swedish, Dutch, German, English). In addition they
make the following assumption about negation in English, following
Chomsky (1986):
Verb raising in English (which is not a V2 language) distinguishes thematic
from non-thematic verbs, and this has a bearing on the surface pattern of the
English negative clause. While non-thematic verbs raise to IP and leave
negation in a post-verbal position, thematic verbs remain, uninflected, in the
VP… (Bardel & Falk 2007:469)

21
The code EN is used for the learners who speak English as an L2, and the code D/G refers
to those learners whose L2 is Dutch and/or German.

150
 Language Acquisition and Use in Multilingual Contexts

They contrast this analysis with that of other Germanic languages in which,
according to their analysis, the position of the negator results from V2,
which in turn does not differentiate between thematic and non-thematic
verbs. They capitalise on this difference between English and the other Ger-
manic languages in testing their L2 transfer hypothesis. On this basis they
predict that according to the L2-transfer hypothesis where “L2 speakers of
Dutch/ German…will place negation post-verbally as in Swedish, while the
other group who have English as an L2…will distinguish between thematic
and non-thematic verbs in relation to negation placement, since this is a pro-
perty of English.” (Bardel & Falk 2007:474.)
Unfortunately, data set A yielded a very small quantity of data. Given that
the study focuses on transfer at the initial state, the first recording is of spe-
cial significance. However, this recording merely contains an average of less
than two sentences per learner and structure for the group with a V2-L2 and
an average of just over one sentence per learner and structure for the group
with a non-V2-L2. For the other recordings the data quantity was even
smaller (cf. Bardel & Falk 2007:475).
This very small amount of data is insufficient for any standard analysis of
lexical or syntactic variation aimed at excluding the use of formulae. At the
same time both groups of learners produce examples of pre-verbal negation
and post-verbal negation, although the L2 transfer hypothesis predicts a diff-
erent distribution. Therefore the relevance of this set of data for the issue of
L2 transfer remains to be demonstrated.
Data set B consists of an average of about seven relevant sentences per
learner and structure, and the distributional analysis for the four learners
shows that the Dutch/German group does not produce pre-verbal negation.
In contrast, the English group produces both pre-verbal and post-verbal neg-
ation – depending on the presence or absence of non-thematic verbs. At first
glance, this observation may be judged as support of the L2 transfer hypo-
thesis. However, there are two problems with B&F’s study in data set B: (1)
The exact status of the learners’ L2 has not been identified, and (2) the role
of repetitions and chunks in very early formal L2 learning has not been
considered.
Obviously, identifying the exact status of the learners’ L2 is vital in the
context of testing a hypothesis that assigns a special status to transfer from
the learner’s L2. The learners in data collection B and their knowledge of V2
languages are reported by B&F as shown in Table 4. For data collection A,
B&F inform the reader that they asked the learners to self-rate the proficien-
cy of their L2s, and thereby B&F identify the learners’ “strongest L2” which
is then recorded under “second language” in the language profile of the lear-
ners. For data collection B this procedure is not mentioned explicitly. One
can only assume that it was the same for both sets of data and that the L2s
shown in Table 4 are the strongest L2s of the learners. B&F do not mention
the other L2s of the learners explicitly. However, the language policies of

151
Language Acquisition and Use in Multilingual Contexts

their countries of origin and their study/work situation suggest very strongly
that they also speak other L2s – like the informants in data set A. B&F state
that learner D/G3 “was found via the European Parliament” and learner
D/G4 “…via the University of Stockholm ” (2007:472). One can assume
with near-certainty that the Italian learner recruited through the European
Parliament working in Brussels has English as one of his L2s, because
English has featured among the language subjects in Italian schools for well
over two decades, and English is also the lingua franca in and around the
European Parliament.
If the two learners of the V2-L2 group (D/G3 and D/G4) also have a non-
V2-L2 – as appears to be the case – the results of the distributional analysis
of data set B appear far more difficult to interpret than it seemed at first. The
absence of pre-verbal negation would then have to be due exclusively to the
effect of the strongest L2 which would need to override possible effects of
other L2s that do have pre-verbal negation. In fact, the same line of argu-
ment would apply to data set A as well because it also contains other L2s be-
sides the “strongest L2”.
In fact, this line of argument would be required as a matter of principle
for B&F’s L2-transfer hypothesis to be internally consistent. If one could
attribute differential effects to just any L2 in a post-factual manner, the
explanatory power of the L2 transfer hypothesis would be eroded. Alter-
natively, one would need to face a much more challenging task: to design a
testable hypothesis of partial L2 transfer that also includes the effects of
additional and typologically different L2s.
There are two issues that follow from the “strongest L2-assumption”: (1)
how does one measure and define the strongest L2 and (2) what is the theo-
retical motivation for it? Referring to the first issue, B&F admit that self-
rating – which they relied on – “may not be an objective method of identify-
ing exact proficiency in a language, but it would not have been feasible to
test proficiency level in all background languages in a precise way” (2007:
471). We would like to add that proficiency may not even be the right
concept that captures the notion of “strongest L2”. This brings us straight to
issue (2), the theoretical basis of the notion “strongest L2”. The way the term
is used by B&F is reminiscent of the notion dominant language in the con-
text of research on bilingualism. In a review article on measuring bilingua-
lism Pienemann & Keßler (2007) show that a multitude of different approa-
ches to capturing language dominance has been discussed in the past five
decades without an operational consensus. Obviously, B&F do not provide
an explicit rationale for what a strong L2 is and why it should have a privile-
ged status in non-native language acquisition. The very last sentence of their
article may hint at what they have in mind: “....in L3 acquisition, the L2 acts
like a filter, making the L1 inaccessible.” (2007:480). This begs the question
what it filters, besides the L1: The weaker L2s? All of them, and all fea-
tures? And where does this happen? In production and comprehension? In

152
 Language Acquisition and Use in Multilingual Contexts

the Formulator (e.g. Levelt 1989), in the bilingual Formulator (DeBot 1992),
in the lexicon? And how is this performance-based filter related to linguistic
knowledge in the various languages? How can all of this be represented?
Obviously, B&F’s hypothesis is not embedded in any explicit theoretical
approach, linguistic or psycholinguistic, and therefore it cannot be operation-
alised.
The second point that needs to be considered for Data set B in B&F’s
study is the status of formulae. This is relevant here because the data were
collected in one single session without any previous contact with the target
language. In very early L2 classes learners’ utterances often consist of
formulae and repetitions of the teacher's utterances, and the structures these
appear to contain are not generated by their newly developing non-native
formulator. Instead, they are unanalysed large entries in the lexicon. There-
fore special care needs to be taken to distinguish between formulae/ repeti-
tions and productive learner utterances. Pienemann (1998) showed that a
mere count of the occurrence of structures in an L2 corpus does not reveal
the underlying learner system and can be rather deceiving. He argued that
what is required instead is a test of the null-hypothesis for every structural
context. For instance, is a morpheme that marks plural in the target language
used in plural contexts only in the learner language or also in non-plural
contexts? If it is used in both contexts with a similar frequency, it is
obviously not a productive part of the IL grammar. B&F did not test the null
hypothesis.22 Therefore, we cannot rule out that the apparent distribution of
pre-verbal and post-verbal negation is based on formulae or repetitions.

The PALU23 study

Our reservations about the B&F study prompted us to replicate key aspects
of that study – with a view to overcome the shortcomings of their study. One
key aspect of the design of our replication study was to differentiate between
formulaic echoes of teacher utterances and creative L2 productions.
In the PALU study the L3 is Swedish, the L1 German. The informants
have different L2s. In keeping with the DMTH outlined above, it is hypothe-
sized that all learners follow a strictly implicational sequence of develop-
mental stages and do not transfer any structures from their first or second
language before they are developmentally ready. Also, we expect all struc-
tures they produce to be in line with the processability hierarchy for Swedish
as an L2 outlined in Pienemann (1998:190 ff.). However, learners are expec-
ted to be able to repeat phrases and sentences and to use structures as fixed
formulae.
22
In fact, data set A would be far too small for this purpose anyway.
23
PALU refers to the universities of PAderborn and LUdwigsburg.

153
Language Acquisition and Use in Multilingual Contexts

Research design
The data collection for the PALU study was conducted at the University of
Paderborn, Germany. The participants were seven German students of ling-
uistics all of whom were fluent speakers of English with high C-test scores
(cf. Grotjahn 1992). Three of the students had some prior knowledge of
Swedish: C01 and C02 attended a one-semester Swedish course and C01,
C02 and C05 took part in a comparative course of Nordic languages. The
other four students had no prior knowledge of Swedish and its structure.
However, as all seven informants were students of linguistics, and as the
curriculum includes courses in both theoretical and comparative linguistics,
they all had some meta-linguistic awareness.
The participants can be divided into two groups according to their know-
ledge of verb-second languages other than German (cf. Tables 6 and 7). The
first group (group A) consisted of four learners with English as their (first)
L2 who learned one or more Romance languages afterwards (e.g. French,
Italian). The second group (group B) comprised three learners who also had
English as their (first) L2 but who additionally learned a V2 language (e.g.
Dutch).
Table 6. Group A informants and their knowledge of languages (no V2 languages).
Learner Sex First language Additional Languages
C03 F German English, French, Latin, Arabic
C04 F German English, Latin, French, Spanish, Russian
C05 F German English, French, Italian, Chinese
C07 F German English, French, Spanish, Portuguese, Italian

Table 7. Group B informants and their knowledge of languages (+ V2 languages).

Learner Sex First language Additional Languages
C01 F German English, French, Spanish, Italian, Portuguese,
Turkish, Dutch + V2, Swedish +V2
C02 F German / English, French, Spanish , Italian, Swedish +
Russian V2
C06 F German English, Latin, Italian , French, Dutch + V2

In order to test the hypotheses outlined above, the framework for data
collection consisted of three main components: (1) a lesson in Swedish, (2) a
session with four communicative tasks which took place after the lesson and
(3) a post-test two weeks after the Swedish lesson.
Prior to the Swedish lesson, the informants listened repeatedly to a recor-
ding of forty Swedish words (nouns, verbs, adjectives, adverbs) that were re-
lated to the communicative tasks used in the subsequent lesson while looking
at picture cards illustrating these words. This was to ensure that the students

154
 Language Acquisition and Use in Multilingual Contexts

familiarized themselves with the vocabulary of the lesson and the related
tasks. After this exercise all seven informants participated in a 30-minute
one-on-one lesson conducted in Swedish by a native speaker who is a uni-
versity lecturer of Swedish. During the lesson, a dialogue was rehearsed and
a number of daily activities were described. The vocabulary introduced in
the lesson was mainly based on the recorded words. The input provided by
the teacher consisted of structures that were located at the different stages of
the PT hierarchy, including different forms of negation and the occurrence of
adverbs in varying positions in the sentence.
The overall aim of the lesson was to provide numerous contexts for the
students to repeat utterances and thus to provide an environment for the
production of formulaic speech (cf. Pienemann 2002; Aguado 2002). This
focus on formulaic speech permits us to test our hypothesis that learners of a
foreign language are able to repeat advanced L2 structures which they are
unable to produce productively.
The communicative tasks were structured so as to ensure that they would
elicit sentences that were different from the material rehearsed in the lesson.
This precaution was taken to ensure that creative L2 constructions produced
by the learners were not copies of rote-memorized sentences. The post-test
followed the same format as the session with communicative tasks.

Results
The results of the PALU study for V2 are presented in Table 8 which is laid
out as follows. The first column lists the informants; the second marks the
presence of SVO. The third column details the frequency of the structure
*advSVO which is ungrammatical in both the source and the target
language. The column headed ‘V2’ lists information about the presence (+)
or absence (-) of V2 in the sample of the individual learners. The column
‘L2=V2’ specifies if the informant acquired a V2 language as an L2 before
the study. The next column specifies if the informant has learnt Swedish
before, and the last one gives the frequency of V2 imitations in each sample.
Table 8. Swedish word order in the PALU study.
Informant SVO *adv SVO V2 L2 = V2? Swedish Imitation
before? of V2
C03 + 14 - - - 16
C05 + 25 - - - 14
C07 + - - - - 10
C04 + - - - - 20
C01 + 30 - + + 30
C02 + 15 - + + 15
C06 + 13 - + - 9

155
Language Acquisition and Use in Multilingual Contexts

Table 8 shows that all learners produced novel SVO sentences even though
the lesson consisted only of sentence repetitions. This shows that the mini-
mal input was sufficient to stimulate language production. As can be seen
from the last column, all learners were also able to repeat V2 sentences
correctly even without any previous input in Swedish. This observation con-
firms our hypothesis that learners are able to store and repeat sentences
containing advanced structures. In contrast, none of the learners produced
V2 structures spontaneously24 (column 4). As can be seen from the data in
column 3, five of seven learners produced fronted adverbs and adverbials. In
other words, these learners did not produce V2 although they produced the
structural condition for V2. The reader will recall that adverb fronting is a
structural condition for V2 in the informants’ L1 as well as in the target
language, i.e. a situation where transfer at the initial state would have been
expected under the full transfer assumption. As shown in column 5, non-
production of V2 after adverb fronting appears with learners who acquired
non-V2 languages before Swedish as well as with those who acquired V2
languages as second languages and even with the two informants who had
learned Swedish before. In other words, our corpus does not contain a single
example of V2 even though the learners know this structure from their L1
and they had plenty of opportunity to use it. This finding constitutes strong
evidence supporting the DTMH.
Apart from the focus on V2 the data were also analyzed in relation to the
position of the negator. For sentence negation the position of the negator in
declarative sentences with lexical verbs are distributed as follows in the three
Germanic languages relevant in this study:
German V+neg
English DO+neg+V
Swedish V+neg

In studies of second language acquisition forms such as don’t are often

treated as one lexical entry which serves as a negator in the interlanguage.
This approach is useful in early SLA when the corpus does not contain any
lexical or morphological variation of the negator or the auxiliary. However,
Bardel & Falk hypothesize that L3 learners are able to transfer
developmentally advanced structures from the L2. Therefore our analysis

24
We use a minimal definition of ”spontaneous production“ in this context. For the purpose
of this study we assume that structures which are not copies of the previous utterance are
produced spontaneously. This minimal definition has to be seen in the context of the
hypothesis we tested, namely that at the initial state advanced structures such as V2 can be
repeated straight after a stimulus sentence has been presented, but that learners will not be
able to produce this structure spontaneously. This minimal definition of ”spontaneous
production“ ensures that our hypothesis is highly falsifiable. It ensures that “unwanted” data
cannot be classified as formulaic copies.

156
 Language Acquisition and Use in Multilingual Contexts

treats the negator and the verbal element do as two distinct constituents as
they appear in the target language.
Table 9 displays the distributional analysis of the use of negation in the
sample. The first column identifies the informant. The next three columns
contain the counts of examples of the key structures neg+V, V+neg, aux+
neg+V and DO+neg+V. The next two columns specify if the learners have
an L2 containing V2 and if they have learned Swedish before.
Our analysis focuses on the first six informants in Table 9 because
informant C06 produced exclusively lexical forms of the negator which do
not exist in Swedish and which were not contained in the input. Learners
C01 and C02 have previously learnt Swedish. Therefore, any structures
appearing in their sample may be residual effects of their knowledge of
Swedish. Of the four remaining learners three produce preverbal negation,
i.e. the developmentally earlier structure. The only exception is C07. This
observation supports the DMTH which predicts that developmentally late
structures can only be transferred when the interlanguage is ready for it.
Table 9. Negation.
Informant neg V V neg aux+neg+ DO+neg V2=L2? Swedish
V +V before?
C03 14 + 1 - 0 0 - -
C05 17 + 0 - 0 0 - -
C07 0 - 16 + 0 0 - -
C04 12 + 4 (-) 0 0 - -
C01 1 - 16 + 0 0 + +
C02 15 + 0 - 0 0 + +
C06 indiv. indiv. indiv. indiv. + -
strategy strategy strategy strategy

Given that all learners also have English as their L2, and are highly
proficient speakers, the above data also permit a test of the L2 transfer
hypothesis. According to this hypothesis the learners would be expected to
transfer DO-insertion from English. As can be seen in column 5 of Table 9,
none of the learners transfer DO-insertion. Column 4 shows that in addition
aux+neg+V25 does not appear either. This finding boldly contradicts Bardel
& Falk’s prediction that L3 learners will transfer structures from L2 English.

25
Note that for the purpose of this analysis ‘aux’ includes auxiliaries and modals.

157
Language Acquisition and Use in Multilingual Contexts

Summary and discussion

The current debate about the role of the L2 transfer was prompted by the
study by Håkansson et al. who found that Swedish learners of German do
not transfer V2, although V2 is part of German and Swedish grammar. They
explained this phenomenon with reference to PT and the assertion that V2 is
too complex to be processable at the initial state.
Bohnacker hypothesized that the non-transfer of V2 in the study by
Håkansson et al. was due to the effect of English, the L2 of all learners in
their study. Bohnacker presented her own study of German as L2, conducted
with elderly Swedish learners of German who were reported not to have any
English as L2 as an intervening variable. She showed that these learners
were more accurate in the use of V2 than Swedish L2 learners of German
who had learned English as an L2. However, the reanalysis of her data in the
form of an implicational analysis demonstrates that all learners in her study
had already acquired V2. Therefore her study did not show that V2 is trans-
ferred at the initial state and that the previous knowledge of intervening L2 is
limited to an effect on grammatical accuracy.
Bardel & Falk (2007) repeated Bohnacker’s claim about the effect of L2
English in the acquisition of German by Swedes. They present two studies of
the acquisition of V2 languages by learners with different L1s and L2s. They
aim to demonstrate distinctive L2 effects in L3 acquisition. One of their
studies did not contain sufficient data to demonstrate their claim. The other
study, based on four informants who received 45 minutes of exposure, relied
heavily on the dominance of one of the L2s of the learner. B&F claim that it
is the dominant L2 from which the learner makes full transfer and that this
process suppresses L1 knowledge. We have shown that this position is in-
conclusive at a conceptual level; it does not specify how the mind identifies
the strongest L2, nor what to do with other L2 knowledge, how this transfer-
filter relates to linguistic knowledge and performance nor how it can be
operationalized. In addition, it remains unclear in B&F’s study which of the
limited utterances produced by the learners after a one-off 45-minute session
were mere repetitions of input and which were productive utterances.
Our own study was based on the acquisition of Swedish as a target
language by seven learners with different L2s. All the learners were highly
proficient in English, all had several other L2s, and three of them had L2s
with V2. All learners were able to repeat V2 sentences in Swedish, but none
of them were able to use this structure productively. This observation
strongly supports the DMTH and also demonstrates that formulae can easily
distort research findings in studies focusing on the initial state.
Our study also focused on the acquisition of negation. We were able to
show that except for one learner, all relevant informants acquired the deve-
lopmentally earlier structure (neg+V). We also demonstrated that none of the

158
 Language Acquisition and Use in Multilingual Contexts

learners transferred DO-insertion from English, or any similar structure.

Both these observations clearly support the DMTH.
The study focused on the restrictive effects of processability on transfer.
In this context, the DMTH might be misconstrued as a non-transfer app-
roach. This would be incorrect. As pointed out in the first section of this
chapter, the DMTH defines constraints on transfer. This implies that re-
strictive and productive effects of the L1 will materialize at predictable
points of development. Pienemann et al. (2005) reported on productive eff-
ects of L1 transfer at points of processability. Also, the DMTH does not ex-
clude the possibility that L2 features may be transferred to the L3. However,
as shown above, for any L2 transfer hypothesis to make a genuine contri-
bution to a theory of SLA it needs to be fully operationalized and theo-
retically motivated. Anything less would be no different than any of the
speculative approaches the field has experienced over the past few decades.

References
Aguado, K. 2002. Formelhafte Sequenzen und ihre Funktionen für den L2-Erwerb.
In Zeitschrift für angewandte Linguistik, 37. 27–49.
Bardel, C. & Falk, Y. 2007. The role of the second language in third language acqui-
sition: the case of Germanic syntax. Second Language Research, 23(4). 459–84.
Bohnacker, U. 2006. When Swedes begin to learn German: From V2 to V2. Second
Language Research, 22. 1–44.
Chomsky, N. 1986. Barriers. Cambridge, MA: MIT Press.
De Bot, K. 1992. A Bilingual Processing Model: Levelt’s ‘Speaking’ Model Adap-
ted. Applied Linguistics, 13. 1–24.
Grotjahn, R. (ed.), 1992. Der C-Test. Theoretische Grundlagen und praktische
Anwendungen. Bochum: AKS-Verlag.
Håkansson, G., Pienemann, M. & Sayehli, S. 2002. Transfer and typological proxi-
mity in the context of second language processing. Second Language Research,
18 (3). 250–273.
Levelt, W. 1989. Speaking. From Intention to Articulation. Cambridge, Mass.: MIT
Press.
Meisel, J., Clahsen H. & Pienemann, M. 1981. On determining developmental
sequences in natural second language acquisition. Studies in Second Language
Acquisition, 3 (2). 109–135.
Pienemann, M. 1998. Language Processing and Second Language Development:
Processability Theory. Amsterdam: John Benjamins.
Pienemann, M. 2002. Unanalysierte Einheiten und Sprachverarbeitung im Zweit-
sprachenerwerb. Zeitschrift für angewandte Linguistik, 37. 3–26.
Pienemann, M., Di Biase, B., Kawaguchi, S. & Håkansson, G. 2005. Processability,
typological distance and L1 transfer. In M. Pienemann (ed.), Cross-linguistic
Aspects of Processablity Theory. Amsterdam: John Benjamins.
Pienemann, M. & Håkansson, G. 2007. Full transfer vs. developmentally moderated
transfer. Second Language Research, 23 (4). 485–494.
Pienemann, M. & Keßler, J.-U. 2007. Measuring Bilingualism. In P. Auer & L. Wei
(eds), Handbook of Applied Linguistics, 5: Multilingualism. Mouton de Gruyter.

159
Learning in two languages
Consequences for lexical development in Swedish and
Arabic

Eva-Kristina Salameh
Ulrika Nettelbladt

Introduction
The demographic structure of Swedish society has changed significantly
over the last decades through immigration. Today 27% of the total Swedish
population of 9,6 million are either born abroad or have one or both parents
born abroad. In the city of Malmö, in the very south of Sweden, this applies
to 38% of the population of 300.000 inhabitants (Statistics, Malmö Council
2009). The proportion of language minority students in schools in Malmö is
well over 50%, and children with Arabic as their first language constitute
one of the largest groups.
In ethnically diverse residential areas the proportion of bilingual pupils in
schools is 75–100%. The academic achievement level of minority students is
significantly below average (Statistics, National School Authorities 2008). In
order to attend to the needs of these pupils, a project with bilingual Swedish-
Arabic classes was started in an ethnically very diverse area in Malmö. The
children in the bilingual classes became literate in both Arabic and Swedish,
and were also educated in school subjects in both languages. During the first
school year the children were exposed to Arabic to a high degree. By every
school year the amount of Swedish increased and Arabic decreased. Several
aspects were studied in the project regarding language development in both
languages as well as academic success. School leaders, teachers, parents and
pupils were also interviewed about their views on bilingual education.
The overall aim of the present study is to investigate possible effects of
bilingual education on lexical size and organization in both Swedish and
Arabic.

160
 Language Acquisition and Use in Multilingual Contexts

Background
Children who are educated in their second language often have difficulty
reaching higher levels of educational achievement, which has been shown in
several international studies (see e.g. Thomas & Collier 2002). They are
often faced with the challenge to simultaneously develop a new language
and complete their education mediated through the new language. As a result
they may develop a second language lexicon with insufficient proficiency in
basic vocabulary. They learn words that are difficult and abstract, since these
are the words explained in school. Basic words in their second language do
not receive much attention, since it is often taken for granted that these
words are already established (Viberg 1996).
If children who encounter a second language later in childhood are not
sufficiently exposed, they often tend to display gaps in their second language
lexicon. These gaps may affect both the basic vocabulary and the vocabulary
learned in school, mostly relating to school subjects. In a study regarding
lexical development in Turkish-Dutch children aged 4, 6, and 8 years Ver-
hallen and Schoonen (1998) showed that the size of the Dutch lexicon in-
creased across age groups also in this group. But the original gap in the basic
vocabulary in Dutch between the bilingual and monolingual children did not
disappear over time. The gap widened with age, and as a consequence the
oldest Turkish-Dutch children lagged behind most.
According to Verhallen and Schoonen (1998) children who learn a second
language lag behind not only in lexical size but also in amount and kind of
knowledge regarding individual words. Word acquisition cannot be defined
as a question of whether a word is acquired or not; rather it is a question of
how extensive the knowledge of a word is. Focus should not only be on the
number of new words in a child's lexicon, but also on in-depth knowledge of
internal relations between familiar words. Within the lexicon one or several
meanings are attached to every word, together with phonological, gram-
matical and semantic information.
Acquisition of lexical abilities can be described according to different
parameters, and in the present study a model presented by Meara (1996) will
be used. Size is a fundamental parameter in Meara’s model, but as the
lexicon grows the need for lexical organization increases steadily. One way
to identify a well-structured lexicon is to test lexical organization by using
word association tests. Lexical organization is an important tool when there
is a need to differentiate between learners with varying proficiency. Accord-
ing to Meara results on most linguistic tasks differ between persons with a
well-organized lexicon and those with the same lexical size but with an
ineffective lexical organization.
Namei (2002) questions earlier research results that second language
speakers should be more prone to organize their lexicon according to phon-
ological principles, while first language speakers use more semantic prin-

161
Language Acquisition and Use in Multilingual Contexts

ciples. She suggests that the use of phonological and semantic principles
regarding lexical organization is a product of linguistic experience. Rare and
unknown words are organized according to phonological principles, partly
known words are organized syntagmatically and frequent words paradig-
matically. This applies to speakers of both first and second languages.
In a study concerning Turkish-Dutch children Verhallen and Schoonen
(1998) studied lexical knowledge in both languages in school years 3 and 5,
by asking the children to explain the meaning of common nouns in Dutch
and Turkish. The results were compared with those of monolingual Dutch
children. The results showed that the lexical knowledge in Turkish was not
extensive enough to compensate for the poor lexical knowledge in Dutch,
since the children could not provide any explanations in either language.
One reason for this may be that the lexical knowledge in Turkish did not
become restructured and developed into academic concepts in school, since
the educational language of the children was Dutch. The Turkish lexicon
was restricted to contextualized common concepts, and in Dutch the children
did not have a stable lexical base from which they could restructure their
Dutch lexicon and increase their lexical knowledge.

Lexical development
Early lexical development is characterized by fast development, and the
ability to categorize objects and activities is a prerequisite for lexical growth.
The most basic principle in the early preschool years is based on
phonological associations, also called clang associations, for example cat –
hat or alliterations as cat – can. Later during the preschool years the child
will develop syntagmatic associations. The association to the word dog
might be ‘barks’ or ‘bad’, creating syntactic sequences with semantic con-
nections, as the dog barks and bad dog.
When a child starts school and learns to read s/he also learns new words
continuously and is confronted with new meanings and meaning relations,
which develop and enrich the knowledge of already familiar words. The
child is exposed to a continuing process of generalizing, categorizing and
abstracting, and has to apply decontextualized knowledge. The language
used in textbooks also requires an understanding of words in a paradigmatic
way since paradigmatic relations allow for generalisations. The lexicon
grows fast and thus the reorganization of associative lexical-semantic net-
works becomes a necessity in order to facilitate lexical retrieval, an essential
skill for academic success (Nelson 1977; Verhallen & Schoonen 1993, 1998;
Namei 2002; Cronin 2002; Schoonen & Verhallen 2008).
As a consequence the child will develop an increasing amount of
paradigmatic associations with coordination as well as super- and
subordination. These associations are more effective for word retrieval.

162
 Language Acquisition and Use in Multilingual Contexts

Associations to the word dog may be ‘cat’, ‘animal’, ‘spaniel’, ‘tail’. Para-
digmatic associations like these usually belong to the same word class in
contrast to syntagmatic associations that are inconsistent with the word class
of the stimuli word (Namei 2002). The shift to a mostly hierarchical lexical
organization, sometimes called the syntagmatic-paradigmatic shift, usually
occurs between 6–10 years of age, an age-range that corresponds to the first
school years with formal reading and academic instruction.
A new word entering the lexicon of a second language will usually elicit
clang associations first, and when the word is more established syntagmatic
associations will appear. Only high-frequency words will elicit paradigmatic
associations, reflecting a hierarchical organization with super- and sub-
ordination, while unfamiliar words tend to elicit clang associations even in
adults. This means that different stages of associations may exist within the
same individual (Cronin 2002).
Namei (2002, 2004) studied the syntagmatic-paradigmatic shift in 50
monolingual Swedish and 50 Persian children and in 100 bilingual Swedish-
Persian children and young people between the ages of 6 and 22 years. For
all participants clang associations were more frequent for barely known or
unknown words, partly known words had a strong syntagmatic organization
and words deeply integrated in the lexicon were paradigmatically connected
to other words. She also found that the syntagmatic-paradigmatic shift
occurred in both languages between the ages of 6–10 years of age, although
bilingual children are less exposed to each language compared to
monolinguals who are exposed only to one language. The syntagmatic-
paradigmatic shift may thus be a function of cognitive development of the
same period, also suggested by Nelson (1977).
Schoonen and Verhallen (2008) pointed out that lack of a decon-
textualized lexical knowledge is not always obvious since it can be obscured
by a superficial facility in the way a child uses words. Teachers and other
school staff are often surprised when faced with the poor word knowledge
some children display regarding words in the second language that are
considered well-known. Only when bilingual children are able to organize
their lexicon according to super- and subordination will it be possible to
reduce the difference in knowledge of words between bilingual children and
their native peers.
If a bilingual child is sufficiently exposed to both languages, a more
efficient lexical organization may develop earlier compared to monolingual
children. Sheng, McGregor and Marian (2006) studied lexical-semantic
organization in Mandarin-English children and showed that bilingualism
may enhance a paradigmatic organization of the lexicon. There is strong
empirical support that bilingual children tend to separate words and their
meanings earlier than monolinguals, since they have two words for a given
referent. This may cause an earlier development of semantic relations in the
bilingual lexicon, and initiate an interest in relationships between words.

163
Language Acquisition and Use in Multilingual Contexts

Having words for a given referent also means that the child must process and
organize more words than monolingual children, which may improve the
organizational ability of the child, thus enabling a more paradigmatic app-
roach (Peña et al. 2002).

Method

Participants
The 16 children in the project group were assessed in grade 4, at 10–11 years
of age as shown in Table 1 below. All had received bilingual education in
both Swedish and Arabic from grade 1. To enable comparison, a control
group of 33 age- and gender-matched Swedish-Arabic children in grade 4
participated. They had received education in Swedish only, but were offered
two Arabic lessons a week by a mother tongue teacher.
Table 1. Participating children.
Gender Project group Control group Total
age 10–11 yrs age 10–11 yrs
grade 4 grade 4
Boys 4 9 13
Girls 12 24 36
Total 16 33 49

All children lived in the same ethnically very diverse residential area in
Malmö. They were also considered to have a typical language development
in Arabic by teachers and parents. No monolingual Swedish-speaking
children attended any of the participating schools in this residential area.
The children were not categorised as simultaneous or successive biling-
uals. The division between simultaneous bilinguals, exposed to two lan-
guages before the age of three, and successive bilinguals, exposed to a
second language after the age of three, suggested by McLaughlin (1978), is
difficult to maintain for bilingual children from multi-ethnic areas
(Håkansson et al. 2003). The children in these areas are primarily exposed to
adult and peer models of bilingual Swedish, since competent speakers of
Swedish both as first and second language are scarce in their residential area.

Data
Lexical organization was tested with free word associations. Such tests
require that the subject responds with the first word that comes to mind,
when the stimulus word is presented. Kent and Rosanoff (1910) devised an

164
 Language Acquisition and Use in Multilingual Contexts

association list of 100 common English words that has since been used in
many studies regarding lexical development in the mother tongue, in second
language and foreign language learning (Namei 2002; Norrby & Håkansson
2007). The list contains 71 concrete and abstract nouns and 29 adjectives.
The Swedish translation of the Kent-Rosanoff list in Namei (2002) was used
in this study. For the present study the list was translated into Arabic and
reviewed by an Arabic-speaking linguist. In some cases allowances for diff-
erences in colloquial Arabic from the East Mediterranean region and Iraq
were made, by using different words for the same concept. The words were
translated back from Arabic to Swedish and with few exceptions the words
corresponded to the words in the Swedish list. The agreement was 98%.
The Arabic test of the Kent-Rosanoff list was conducted by an Arabic-
speaking teacher, and the Swedish test by two master students in speech
language pathology. For practical reasons most children were tested in
Arabic first. The interval between the test sessions was two or three weeks.
The test sessions were recorded. To ensure reliability in classification 10%
of the material was randomly chosen and the interreliability agreement
between three classifiers was 90% for both languages.

Analysis
The assessment of the associations of the Kent-Rosanoff list in both
languages was based on a total of 1600 responses in each language from the
project group and a total of 3300 responses in each language from the
control group. The word associations were classified as clang, syntagmatic
or paradigmatic associations. In some cases these categories could not be
applied, and two extra categories were added. The category ‘other’ com-
prised associations that could not be classified due to an unclear connection
with the stimulus word, misunderstandings and repetitions, and the other
extra category was ‘no answer’.
Examples of clang associations are hus – mus (‘house – mouse’), sax –
lax (‘scissors – salmon’). In some cases the response was a rhymed nonsense
word or alliteration. Examples of syntagmatic associations are frukt – äta
(‘fruit – eat’), kvinna – du (‘woman – you’). Grammatical derivations were
also counted as syntagmatic, such as blom – blomma (‘blossom – flower’),
man – manlig (‘man – manly’) according to the procedure in Namei (2002).
Examples of paradigmatic associations are pojke – flicka (‘boy – girl’), bröd
– smör (‘bread – butter’).
The following associations were categorized as ‘other’, or not class-
ifiable: repetitions of words or part of words; misunderstandings, e.g. ugn –
gammal (‘stove – old’; the word ‘ugn’ is very similar to ‘ung’ (‘young’));
incomprehensible associations as vitkål – pyjamas (‘cabbage – pyjamas’);
associations in Swedish when tested in Arabic and vice versa, and

165
Language Acquisition and Use in Multilingual Contexts

neologisms such as hammare – spikare (‘hammer – nailer’). The category

‘no answer’ included answers such as ”I don’t know” and no answers at all.

Results
A comparison of proportions regarding paradigmatic and syntagmatic
responses in Swedish in both groups is shown in Figure 1 below. The project
group (bilingual education) displays more paradigmatic than syntagmatic
responses, while the control group (monolingual education) displays more
syntagmatic than paradigmatic responses, although they are educated in
Swedish only. The difference between the proportions of syntagmatic and
paradigmatic responses in each group was not significant, however (Mann-
Whitney, p=0,105; z= -1,621).

Figur 1. Results from the Kent-Rosanoff list in Swedish.

The results in Arabic regarding paradigmatic and syntagmatic responses are

displayed in Figure 2 below and display the same pattern as in Swedish with
at higher proportion of paradigmatic responses compared to syntagmatic
responses in the project group, who received bilingual education. The diff-
erence between syntagmatic and paradigmatic responses in Arabic is signi-
ficantly larger in the project group (Mann-Whitney, p=0,021; z=-2,315). The
children in the project group thus produce significantly more paradigmatic
responses than syntagmatic compared to the control group.
The children who were educated in both Swedish and Arabic produced a
more hierarchical lexical organization in Arabic. Their results displayed
significantly more paradigmatic responses compared to the control group.

166
 Language Acquisition and Use in Multilingual Contexts

Figur 2. Results from the Kent-Rosanoff list in Arabic.

Discussion
Verhallen and Schoonen (1998) found that bilingual children could not rely
on their mother tongue lexicon for lexical development in their second
language, since lexical size and organization were insufficient in their
mother tongue. There was “a mismatch between their L1 knowledge and the
L2 requirements” (1998:456). Lexical development in the second language
needs a solid base to facilitate restructuring of common concepts into acade-
mic concepts, but this base is often lacking in the bilingual child. Verhallen
and Schoonen showed that since the bilingual children’s knowledge of
words was restricted also in Turkish, they could not rely on their mother
tongue to counteract their lexical difficulties in Dutch.
The children who were educated in both Swedish and Arabic produced a
more hierarchical lexical organization in Arabic. Their results displayed
significantly more paradigmatic responses compared to the control group.
The much larger exposure to Arabic in school in the present study gave the
children in the project group an opportunity to increase their knowledge of
words and relations between words in their Arabic lexicon to a higher extent.
The amount of exposure to both mother tongue and second language is
essential to language development, not least for lexical size and organization
(Paradis 2010). As Verhallen and Schoonen (1998) pointed out in their study
on Turkish-Dutch children, bilingual children are not always helped by
extensive exposure only to their second language. New words are not grad-
ually incorporated in their second language lexicon, and the knowledge of

167
Language Acquisition and Use in Multilingual Contexts

familiar words in the mother tongue does not increase if the exposure is
insufficient. These results may partly explain the significantly lower propor-
tion of paradigmatic responses in Arabic in the control group with education
in Swedish only. The project group, children with bilingual education, had
the possibility to develop their lexical organization in Arabic. This is reflect-
ed in their ability to develop everyday concepts into academic concepts in
Arabic to a greater extent, and transfer them to their second language.
As is shown in another study of this project concerning academic success
(Tvingstedt & Salameh 2011) the project group learned to read and write in
Arabic with no detrimental effects on their reading ability in Swedish com-
pared to the control group. The more hierarchical lexical organization is
likely to have been promoted by the children’s reading development (Cronin
2002) in both languages, enhancing their ability to organize the lexicon
according to more advanced semantic principles.
Although the results of this study are based on a limited number of school
children, they persistently point to the critical role of language exposure, and
the necessity to expand lexical size and organization in both languages. Both
in this study and in Ordónez et al. (2002) the mother tongue plays an import-
ant role in the development of paradigmatic skills. Education in the mother
tongue aimed at strengthening this language may also promote better lexical
skills in the second language.
The results in this study underline the necessity of organizational changes
to give bilingual children access to a linguistic environment adapted to
bilingual development, with education and peer interaction in both lan-
guages. Results from research into bilingual education indicate that this pro-
motes academic success to a higher degree. However, such an education
must stretch over several years to yield sustainable results (see for example
Thomas & Collier 2002).

References
Cronin, V. 2002. The syntagmatic-paradigmatic shift and reading development.
Journal of Child Language, 29. 189–204.
Håkansson, G., Salameh, E-K. & Nettelbladt, U. 2003. Measuring language
proficiency in bilingual children. Swedish-Arabic bilingual children with and
without language impairment. Linguistics, 41 (2). 255–288.
Kent, G. & Rosanoff, A. 1910. A study of association in insanity. The American
Journal of Insanity. LXVII:3.7–96.
McLaughlin, B. 1978. Second language acquisition in childhood. Vol 1. Lawrence
Erlbaum, Hillsdale, N J. 2nd ed. 1984.
Meara, P. 1996. The dimensions of lexical competence. In G. Brown, K. Malmkjaer
& J. Williams. (eds.), Performance and competence in second language
acquisition. 33–53 Cambridge: Cambridge University Press.
Namei, S. 2002. The bilingual lexicon from a developmental perspective. Center for
bilingualism, Stockholm University.

168
 Language Acquisition and Use in Multilingual Contexts

Namei, S. 2004. Bilingual lexical development: A Persian-Swedish word association

study. International Journal of Applied Linguistics, 14. 363–388.
Nelson, K. 1977. The syntagmatic-paradigmatic shift revisited: A review of research
and theory. Psychological Bulletin, 84. 93–116.
Norrby, C. & Håkansson, G. 2007. ‘Girl – lass or curl? Word associations in second
language learners’. Australian Review of Applied Linguistics, 30 (2). 22.1–
22.17. DOI: 10.2104/aral0722.
Ordónez, C., Carlo, M., Snow, C. & McLaughlin, B. 2002. Depth and breadth of
vocabulary in two languages: Which vocabulary skills transfer? Journal of
Educational Psychology, 94. 719–728.
Paradis, J. 2010. The interface between bilingual development and specific language
paradisimpairment. Applied Psycholinguistics, 31. 227–252.
Peña, E., Bedore, L. & Zlatic-Giunta, R. 2002. Category-generation performance of
bilingual children: The influence of condition, category, and language. Journal
of Speech, Language, and Hearing Research, 45. 938–947.
Schoonen, R. & Verhallen, M. 2008. The assessment of deep word knowledge in
young first and second language learners. Language Testing, 25. 211–236.
Sheng, L., McGregor, K. & Marian, V. 2006. Lexical-semantic organization in bi-
lingual children: Evidence form a repeated word association task. Journal of
Speech, Language, and Hearing Research, 49. 572–87.
Statistics, Malmö Council. 2009.
Statistics, National School Authorities 2008.
Thomas, W. & Collier, V. 2002. School effectiveness for language minority stud-
ents’ long-term academic achievement. Centre for Research on Education,
Diversity and Excellence. CREDE. http://www.usc.edu/dept/education/CMMR/
CollierThomasComplete.pdf
Tvingstedt, A-L. & Salameh, E-K. 2011. Läs- och kunskapsutveckling hos elever
som fått tvåspråkig undervisning på svenska och arabiska. [Reading skills and
academic development in pupils receiving bilingual education in Swedish and
Arabic] EDUCARE, 3. 227–267.
Verhallen, M. & Schoonen, R. 1993. Lexical knowledge of monolingual and bi-
lingual children. Applied Linguistics, 14. 344–63.
Verhallen, M. & Schoonen, R. 1998. Lexical knowledge in L1 and L2 of third and
fifth graders. Applied Linguistics, 19 (4). 452–470.
Viberg, Å. 1996. Svenska som andraspråk i skolan. In Hyltenstam K. (ed.), Två-
språkighet med förhinder? [Impeded bilingualism?] Lund: Studentlitteratur.
110–147.

169
“Flera språk – fler möjligheter” –
Immigrant children’s acquisition of
English in bilingual preschools26

Anja K. Steinlen

Introduction
To what extent is the effectiveness of early immersion education affected by
the home language background of the child? Are immersion programs at
preschool equally suited for majority language pupils (i.e. children whose
first language (L1) is the official or dominant language of the wider out-of-
preschool community) and minority language pupils (i.e. children from
family backgrounds where a different language is spoken than the official or
dominant language of the wider community)? This is a controversial issue,
particularly among policy-makers. There is general consensus, at least am-
ong researchers, that for majority language children immersion education
leads to additive bilingualism, with high levels of second language (L2) pro-
ficiency and native-like levels of monolingual L1 proficiency (e.g. Wesche
2002). There is still debate, however, whether immersion education is equ-
ally beneficial and suited for minority language children, particularly
minority language children from a so-called immigrant background. Immi-
grant children in many Western countries have been shown to have diffi-
culties acquiring the host nation’s official language (which for them
constitutes their de facto L2).27 For instance, Knapp (2006) distinguishes
three groups of immigrant children, namely immigrant children with a good
command of the L2, those with a poor command of the L2 and those with
“concealed L2 problems” which become apparent later in school. Further-
more, researchers have reported that some immigrant children occupy un-
favourable positions in the educational system of the host country and often

26
This text is based on findings presented in Rohde (2010) and Steinlen et al. (2010a).
27
In preschools, the language of the host country and the newly introduced foreign language
English may, of course, not be the children’s L2 and L3, but their L4 or L5, depending on the
languages being used at home.

170
 Language Acquisition and Use in Multilingual Contexts

achieve significantly lower overall scores than their majority language peers
(cf. Biedinger 2010). What has been examined in less detail, however, is
how immigrant children fare with respect to the learning of yet another
additional foreign language at school such as English (referred to as the
children’s L3).
For the mainstream school context in Germany, for example, the results
of DESI (Deutsch-Englisch-Schülerleistungen-International, an international
study of student performance in German and English) showed that minority
language students in grade 9 who acquired German as L2 had fewer diffi-
culties learning English and even attained higher levels of English than
majority language students who grew up in monolingual German families
(Klieme 2006). Elsner (2007) examined how Turkish children learning Eng-
lish in German primary schools fared with respect to their comprehension
skills in English as a foreign language. She reported that their English
listening comprehension skills were considerably lower than those of their
monolingual peers. She concluded that ‘multilingualism’ does not necess-
arily lead to better results in English. She attributes this result to the fact that
these Turkish children had deficits in their L2 German which in turn affected
their comprehension skills in English negatively. In Canada, there are con-
vergent findings from immersion education. In her literature review, Hurd
(1993) concluded that minority language children may benefit from early
immersion programs only as long as there is strong support for L1 develop-
ment, because otherwise, such a program could result in a subtractive
bilingualism situation for minority language students (see also Dagenais &
Day 1998; Swain and Lapkin 2005; Taylor 2006 for similar results). In gen-
eral, all authors demand more research with a greater number of children in
order to examine in more detail the challenges and successes encountered by
multilingual students.

The ELIAS project

The EU-funded project ELIAS (Early Language and Intercultural Acqui-
sition Studies), from which the data presented here originate, was a multi-
lateral EU Comenius project, carried out in Sweden, Germany, Belgium, and
England between 2008 and 2010. In total, eighteen partners were involved,
including universities and educational institutions, preschools, and a zoo.
The children participating in the project were preschool children who spoke
at least one other language at home than the majority language. In addition
to acquiring the ambient majority language of the wider community (i.e.
their L2), these children were also faced with the task of acquiring English
(often their L3) in preschool.
In the present study, the focus is on the acquisition of L2 English recep-
tive grammar and vocabulary by children with and without immigrant back-

171
Language Acquisition and Use in Multilingual Contexts

ground. As in other studies (e.g. PIRLS, Mullis et al. 2007), the term immi-
grant background will be used as the parents of these children were not born
in the countries under scrutiny (i.e. Sweden, Belgium, Germany) and their
family language/s did not correspond to the ambient language, as the ELIAS
parents’ questionnaire revealed. As there were too many home languages
involved to include a sufficient number of children for statistical analyses,
the present study focuses on the children’s immigrant background in general.
The ELIAS project nevertheless offers a unique possibility to further exa-
mine in some detail the important issue of the impact of the child’s immi-
grant background on the development of the foreign immersion language in
a bilingual preschool setting.

The organization of bilingual preschool programs

The preschools, which were part of the ELIAS project, employed teachers
from the respective countries (i.e. Sweden, Germany, and Belgium) as well
as English native speakers or teachers with high foreign language skills (in
this case English). According to the one person – one language principle
(e.g. Ronjat 1913, Baker 2000), two teachers were usually present in each
preschool class; one teacher only used the L1 of the children (i.e. Swedish,
German, or French), and the other only their L2 (i.e. English). The new lan-
guage was therefore used in everyday life as a means of communication and
was not offered as a subject (e.g. Rohde 2005; Steinlen 2008b; Wode 2010;
Weitz & Rohde 2010). This immersion type of foreign language teaching
has proven to be the most successful way to learn a foreign language (e.g.
Wesche 2002).
When they started preschool, the children usually had no or very limited
knowledge of English. Therefore, together with what was being said, the
teachers used many non-linguistic resources in order to allow the children to
make a connection between the content and its significance (context-
ualization). For example, the teacher used body language such as facial
expressions, gestures and pantomime, as well as pictures, picture stories,
CDs, videos, objects, etc. (e.g., Steinlen 2008b; Kersten et al. 2010a). This
type of language mediation is not new because children who arrive without
knowledge of the preschool’s ambient language are similarly immersed into
the language of the host country (e.g. Kolonko 2001).

The development of English language skills in preschool

When the children started preschool, they were often withdrawn because
they had to familiarize themselves with the new routines in preschool first.
The new language seemed to be only one factor among many (e.g. Schilk et
al. in press). After a few weeks, the children could easily follow the routines

172
 Language Acquisition and Use in Multilingual Contexts

of the preschool in the foreign language (e.g., good morning, tidy up time,
stop it). The children quickly developed a conversational pattern with those
teachers who only used English, where the children replied in their L1. After
about three months, the children started to use single English words, as these
examples from German preschools show: “Ich habe einen dog”. “Es ist tidy
up time”. “Let's go raus”. The children also started to translate for each other.
At the end of the first year the children responded appropriately to the
foreign language input (e.g. to more complex questions and prompts).
In the bilingual preschool programs studied prior to and within the ELIAS
study, it was obvious that the children quickly developed receptive skills in
the L2 with their L2 production lagging behind (Rohde & Tiefenthal 2002;
Schilk et al. in press). As the L1 English preschool teachers were able to
understand the ambient language outside preschool, there was never any
communicative need to produce the L2 – especially not amongst the child-
ren. “Children all share the same first language so that from their point of
view, there is no vital reason at all to take the trouble of resorting to an un-
known language” (Wode 2001:429).28
Given this situation it was decided to primarily study the children’s grow-
ing receptive knowledge, as they seemed to understand single words and
formulas after only limited exposure. However, in previous, rather small-
scale studies which examined the development of receptive and grammatical
L2 comprehension in bilingual preschools (e.g. Burmeister & Steinlen 2008;
Steinlen 2008a; Rohde 2005; Weitz & Rohde 2010), the focus was on
majority-language children. Indeed, there is a lack of systematic studies of
the acquisition of an additional language in bilingual preschools (often the
L3 or the L4) by children with an immigrant background. Experience from
previous projects indicates similar progress in the acquisition of English for
immigrant children compared to children without such a background, even
though this progress may occur later and may exhibit strong individual
variation, depending on e.g. the personality of the child, family language/s,
learning habits, to name just a few (see e.g. Burmeister & Steinlen 2008).
Children with no or little knowledge of the preschool’s ambient language
often turned to the foreign language in preschool and apparently felt that
they were “in the same boat” as the other children, as none of them were
familiar with the new language (e.g. Schilk et al. in press; Wesche 2002).

28
This situation shows why L2 acquisition in a bilingual preschool program cannot easily be
classified as either naturalistic or classroom L2 acquisition. It has naturalistic features because
the L2 is spoken in everyday situations and there is neither formal teaching nor a specific
language focus involved. On the other hand, the L2 is only spoken by the native speaker
preschool teachers and is thus not the main ambient language. In addition, it may be argued
that, due to the group structure in a preschool, activities have to be arranged and organized to
some extent and may thus be formal rather than naturalistic.

173
Language Acquisition and Use in Multilingual Contexts

Research questions
In the following section, the results of the ELIAS Grammar Test (Kersten et
al. 2010b) and of the BPVS II (British Picture Vocabulary Scale, Dunn et al.
1997), conducted in ten preschools in four European countries, are pres-
ented. The following questions are addressed:
1. What is the level of receptive English L2 grammar and vocabulary
knowledge of children in bilingual preschools at two points in time (T1
and T2), compared to a monolingual English preschool group?
6. What is the impact of the children’s immigrant background on their
levels of receptive L2 grammar and lexical knowledge and on the
amount of progress they make from T1 to T2? Do they reach similar
levels of L2 grammar and lexical knowledge as their non-immigrant
background peers?

Method
Because preschoolers are very different in terms of their cognitive and their
motor and social skills, a test format had to be found that was appropriate for
the age levels between 3 and 6 years. Picture pointing tasks have been shown
to be effective to assess preschoolers’ passive understanding of vocabulary
and grammar (e.g. Bishop 2003: TROG-2; Dunn et al. 1997: BPVS II;
Grimm et al. 2001: SETK 3-5). In the ELIAS project, the picture pointing
tasks were successful because both older and younger children took part in
the tests with joy.
All preschool children were tested individually in a quiet room they were
familiar with (see Crain & Thornton 1998 on the importance of a child-
friendly environment during an experiment). For the ELIAS Grammar Test
(Kersten et al. 2010b), the child looked at three pictures before listening to a
sentence that corresponded to one of the pictures. The child responded by
pointing to the picture it thought represented the sentence. Before the test,
the children were given two practice items with three pictures of different
objects and an appropriate single word utterance to ensure they knew how to
form the responses. The three pictures in each set differed in the following
way: two of the pictures contrasted only in the target grammatical dimension
(e.g. absence/presence of the plural inflectional marker -s: cat/cats). The
third picture was a distractor, i.e. it was semantically related to the other two
pictures and in most cases also lexically related. The distractor was used to
ensure that the child understood the grammatical phenomenon required. The

174
 Language Acquisition and Use in Multilingual Contexts

children were tested on nine grammatical phenomena,29 most of which have

been used in the Reception of Syntax Test (e.g. Au-Yeung et al. 2000;
Howell et al. 2003) and in the Kiel Picture Pointing Test (Steinlen & Wett-
laufer 2005). It is therefore possible to relate the results of the present study
to previous findings. In total, there were 54 test items (9 grammatical phe-
nomena x 3 picture pairs x 2 test presentations per picture set). The session
did not take longer than fifteen minutes.
The BPVS II (Dunn et al. 1997) is a standardized test instrument used to
determine the receptive vocabulary of 3- to 15-year-old L1 speakers of Eng-
lish as well as the vocabulary of children learning English in Great Britain as
an additional language (EAL).30 Maximally 168 words are tested, through 14
sets with 12 cards, where the individual sets are allocated to age levels. Each
card contains four pictured items, one of which is asked for when the BPVS
II is administered. The BPVS II test is very similar to the Elias Grammar
Test, but in contrast, the prompts consist of single words and the test is
discontinued if eight or more errors are made.

Subjects
In 2009 and 2010, a total of 148 children (51% girls and 49% boys) from
seven bilingual preschools in Germany, one in Sweden and one in Belgium
took the ELIAS Grammar Test twice at an interval of 5 to 12 months. Of
these, 39 children had an immigrant background. The children’s age ranged
between 3 and 6 years (mean: 54.4 months, SD = 9.4 months) and they had
been exposed to English between 1 and 42 months at the time of Test 1
(mean: 14.2 months, SD = 8.9 months). At the time of Test 2, the children
were between 4 and 7 years old (mean: 63.8 months, SD = 10.2 months) and
their contact time with English was between 10 and 51 months (mean: 24.2
months, SD = 8.6 months). It was often the case that the older the children
were, the more L2 contact they had; the younger they were, the less contact
with English they had.31 The family languages included about 35 languages,
such as Turkish, Russian, Farsi, Arabic, Polish, Punjabi, to name just a few.
In the same period, a total of 200 children, 137 children without and 63
children with an immigrant background, from the same nine bilingual pre-
schools in Sweden, Germany and Belgium, took the BPVs II twice at inter-

29
The nine grammatical phenomena included subject-verb agreement for copula and for full
verbs, presence or absence of the genitive marker ‘s and of the inflectional plural morpheme
-s, affirmative / negative sentences, 3rd person singular possessive masculine / feminine, 3 rd
person singular pronoun masculine / feminine as subject and object, and word order.
30
EAL does not necessarily mean the children’s L2 but can be any further language added to
the L1.
31
The results of a bivariate correlation analysis (Spearman's rho) showed a strong correlation
between the children's age and their exposure to English (0.387, p < 0.05).

175
Language Acquisition and Use in Multilingual Contexts

vals of 5 to 12 months. The age of the children was 34 to 88 months at the

time of Test 1 (mean 56.4 months, SD = 13.1 months), they were exposed to
English 1 to 50 months (mean 14.2 months, SD = 9.7 months). At the time of
Test 2, the children were 42 to 98 months old (mean: 67.3 months, SD =
13.3 months) and their L2 contact to the English was 10 to 61 months (mean
25.1 months, SD = 9, 3 months). Of the 63 children of immigrant back-
ground, 27 spoke the ambient language at home, for the remaining 36 child-
ren the family languages did not correspond to the ambient language.
In addition, twenty children from a monolingual English background in a
preschool in Hertfordshire, England also took the ELIAS Grammar Test and
the BPVS II twice. These monolingual English children served as a bench-
mark against which the performance of the bilingual preschool children
could be compared. The benchmark children were of approximately the
same age as the bilingual pre-schoolers. At the time of Test 1, the mono-
lingual English children were between 3 to 5 years old (mean: 52.9 months,
SD = 9.3 months) and between 4 to 6 years old at the time of Test 2 (mean:
59.4 months, SD = 9.6 months).

Results
The results show how the scores differed from Test 1 to Test 2 (i.e. five to
twelve months elapsed between Test 1 and Test 2). The focus is on the nine
preschools from Sweden, Belgium and Germany which offered a bilingual
program. Altogether, the data of 149 subjects (ELIAS Grammar Test) and of
200 subjects (BPVS II) were used, including the data of 20 monolingual
children from England. As the tests were a forced choice between three and
four pictures, respectively, 33% (for the ELIAS Grammar Test) and 25%
(for the BPVS II) represent chance level. In order to present an overview, the
results for all children in bilingual preschools were collapsed and compared
to monolingual English pre-schoolers, independent of their immigrant
background. The section Immigrant background compares the English test
scores for children with and without immigration background.

Development over time

As shown in Figure 1, the children received lower scores in Test 1 (50.2%,
SD = 14.2) than in Test 2 (57.1%, SD = 16.2) for the ELIAS Grammar Test
and lower scores at T1 (mean 79.4, SD = 15.2) than at T2 (mean 81.7, SD =
14.0) for the BPVS II. This difference is significant, as statistical analyses
showed (see Footnote 31). That is, the more exposure to their L2 English the
children received, the better their comprehension of selected grammatical
phenomena and of selected vocabulary. Interestingly, this development starts
already at an average exposure to English at 15.1 months because the child-

176
 Language Acquisition and Use in Multilingual Contexts

ren already score above the chance level of 33.3% and 25%, respectively.
The results clearly show a development in the L2 of the children who attend
a bilingual preschool, although English is not the ambient language outside
preschool. Similar results were obtained for the monolingual English control
group. They obtained higher scores in Test 2 than in Test 1, which is a clear
indication that the children’s L1 develops age-appropriately. 32

Figure 1. Scores of the ELIAS Grammar Test (left) and for the BPVS II (right) for
all bilingual preschools combined (Preschool 1–9) and for the monolingual English
children (Preschool 10), as obtained at Test 1 and Test 2.

As illustrated in Figure 1, the monolingual English preschool group scored

considerably higher than the bilingual preschool groups: differences were
significant for both test points and for both tests.33 It is not surprising that the
bilingual preschool group received lower scores in both tests as their ambient
language is not English. A more detailed analysis revealed, however, that
some of the children who had English in preschool for more than three years,
performed almost as well as their monolingual English peers (Rohde 2010,
Steinlen et al. 2010a).

32
A repeated measure analysis revealed significant differences for time for bilingual and
monolingual preschoolers (ELIAS Grammar Test: Time: F (1, 166) = 45.349, p<0.05, BPVS
II: Time: F (1, 218) = 11.167, p<0.05).
33
ANOVAs showed significant differences between the monolingual and the bilingual
preschool group for Test 1 (F (1, 167) = 69.852, p<0.05) and for Test 2 (F (1, 167) = 65.635,
p<0.05) for the ELIAS Grammar Test, similar results were noted for the BPVS II.

177
Language Acquisition and Use in Multilingual Contexts

Immigrant background
Below we examine whether the variable immigrant background affected the
results of the grammar and lexicon tests. As mentioned earlier, out of the 148
and 200 children who completed the ELIAS Grammar Test and the BPVS II,
twice, 39 and 63 had an immigrant background.

Figure 2. Scores of the ELIAS Grammar Test (left) and of the BPVS II (right) across
all preschools, as obtained at Test 1 and Test 2, the focus is on +/- immigrant
background. The results of the 20 monolingual English children are excluded.

As Figure 2 shows, all children obtained significantly better results at T2 in

comparison to T1, in both test types. This indicates that the children, irre-
spective of their background, improved with respect to their English
grammatical and lexical receptive abilities. Furthermore, no significant
difference was found with regard to the scores obtained by children with and
without immigration background for the BPVS II and for the ELIAS
Grammar Test, independent of the time of the test.34 Therefore, it seems that
immigrant background by itself is not a good indicator for predicting how
well children in bilingual preschools acquire certain grammatical and lexical
items in the new language English.

34
Repeated measure analyses for both tests with respect to +/- immigrant background did not
reveal any differences for the interaction between time and immigrant background, only for
time (ELIAS Grammar Test: Time*MigrationBackground: F (1, 146) = 2.989, p>0.05; Time:
F (1, 146) = 79.683, p<0.05; BPVS II: Time*MigrationBackground: F (1, 198) = 0.065,
p>0.05; Time: F (1, 198) = 4.814, p<0.05). As the interaction did not prove to be significant
for both tests, the children with immigrant background were not further subdivided according
to their home language.

178
 Language Acquisition and Use in Multilingual Contexts

Discussion
This study examined the development of L2 receptive grammar and lexicon
in children with and without immigrant background who were exposed to
English in a bilingual preschool context in Sweden, Belgium and Germany.
The results suggest that children, irrespective of their immigrant background,
learn a new foreign language as early as preschool and steadily improve their
receptive vocabulary and grammar, as examined twice with a 5 to 12 month
interval in between two test times. In fact, when the children were sub-
divided according to their exposure to English (see Steinlen et al. 2010a;
Rohde 2010), some of the children who had English in preschool for more
than three years, performed almost as well as their monolingual English
peers. This result clearly demonstrates the feasibility of a bilingual program
in preschools which offer English as an L2 in an immersion context (cf.
Wode 2001; Rohde & Tiefenthal 2002; Rohde 2005; Burmeister & Steinlen
2008; Steinlen 2008a; Steinlen & Rogotzki 2009).
A comparison between children with and without an immigrant back-
ground did not produce any significant differences. This is a very encoura-
ging result as it is often informally reported that children with an immigrant
background are disadvantaged in a preschool setting in which yet another
“new” language is introduced. It is these children who are reportedly more
likely to be disadvantaged in learning contexts as neither their L1 (a minority
language) nor their L2 (the ambient majority language) may be age adequate
(Apeltauer 2004). For example, Elsner (2007) reported that in German
primary schools immigrant children (in particular of Turkish background)
were less successful in foreign language learning compared to their mono-
lingual non-immigrant background German peers. Interestingly, later in their
school career, this disadvantage seems to disappear (see e.g. Klieme 2006).
Similar findings were reported for immigrant children in Canadian imm-
ersion school settings: the longer these children lived in Canada, the smaller
the gap in foreign language performance between children with and without
immigrant background, until it disappeared (e.g. McMullen 2004).
How can we account for the positive performance of immigrant children
in terms of their lexical and grammatical development of English in bilingual
preschools? First, the foreign language in an immersion setting is not taught
in a formal context (like the subject English in school) but is used as a me-
dium of communication. This approach seems very beneficial for immigrant
children who simply acquire the foreign language from the way it is being
used (e.g. Taylor 1992). Second, at least with respect to Germany, the immi-
grant children in the ELIAS Project seemed to have a high command of their
L2 German, as the results of a German language proficiency test showed
(SETK 3-5, Sprachentwicklungstest für drei- bis fünfjährige Kinder, Grimm
et al. 2001; see Steinlen et al. 2010b). As German and English are typolog-
ically closely related languages, it may be assumed that the immigrant child-

179
Language Acquisition and Use in Multilingual Contexts

ren’s comprehension of English benefits from the acquisition of German (see

also Bild and Swain 1989; Hurd 1993). It is likely that similar tendencies
may be found for immigrant children in Sweden but less so for immigrant
children in Belgium (as French and English are typologically less close).
The question of (psycho)typological relatedness in multilingual acquisi-
tion needs to be explored in more detail in a further study, which would also
include information on the children’s home language/s in order to provide a
more complete picture of their language contexts and the development of all
their languages. So far, it is unfortunately less than clear as to how the child-
ren’s home language experiences may have affected the results of the ELIAS
Grammar Test and of the BPVS II. One has to keep in mind, though, that
these children may have benefited from the early immersion program be-
cause there was strong support for their L1 literacy, because otherwise, such
a program could result in a subtractive bilingualism situation, as Canadian
studies on immersion schools has demonstrated (see e.g. Hurd 1993;
Dagenais & Day 1998; Swain & Lapkin 2005; Taylor 2006). This apparently
was not the case for the children in this study. However, such an assumption
needs to be examined in more detail using the ELIAS parents’ questionnaire
which also includes information on the parents’ use of home language
literacy practices.
What are the implications of these results? First, children are able to learn
many languages, i.e. the brain can cater for more than one or two languages
(e.g. Franceschini 2003) and bilingual preschools have been shown to be an
ideal learning environment, where children, irrespective of their language or
immigrant background, can acquire one or more language/s (Kersten et al
2010b). Second, children in bilingual preschools do not only learn a foreign
language but also acquire additional skills, such as better concept formation,
higher communicative flexibility and higher motivation for language
learning (e.g. Bialystok 2001). As Cenoz (2003) points out, multilingualism
may be associated with positive cognitive consequences as long as the child-
ren’s L1s are valued. It is therefore important to strongly advise parents
(with and without immigrant background) to use their family language/s
regularly and to encourage them to read to their children, as early literacy (in
any language) is an important predictor of later academic success in Western
societies (e.g. OECD 2005).
Third, it is important to interpret foreign language acquisition processes in
bilingual preschools in a differentiated way. An immigrant background does
not seem to be the sole predictor with respect to the development of an
additional language, as the results of this study clearly shows, where these
children performed as well as their peers without immigrant background.
Similarly, the immigrant child’s acquisition of the host country’s official
language might not be impeded by attending a bilingual preschool. Conse-
quently, it is advisable to encourage multilingual parents or parents with an
immigrant background to let their children attend immersion programs. This

180
 Language Acquisition and Use in Multilingual Contexts

is a first step within the framework of educational institutions to enable all

children to develop “flera språk – fler möjligheter”,35 as promoted by the EU
(Commission of the European Communities 2003).

Acknowledgements
The author gratefully acknowledges the support of the staff of the preschools
without whom this project could not have taken place. Most importantly, the child-
ren’s enthusiastic participation is greatly appreciated. Between 2008 and 2010,
ELIAS was a multilateral Comenius project funded by the EU (142355-LLP-1-
200B-1-DE-COMENIUS-CMP).

References
Apeltauer, E. 2004. Beobachten oder Testen? Möglichkeiten zur Erfassung des
Sprachentwicklungsstandes von Vorschulkindern mit Migrantenhintergrund.
Flensburger Papiere zur Mehrsprachigkeit und Kulturvielfalt im Unterricht;
Heft 36.
Au-Yeung, J., Howell, P., Davis, S., Sackin, S. & Cunniffe, P. 2000. Introducing the
Preschoolers' Reception of Syntax Test (ROST). In Proceedings of the
Conference on Cognitive Development. Besançon France. www.speech.psychol.
ucl.ac.uk/Publications/PAPERS/PDF/32decolage.pdf.
Baker, C. 2000. A Parents' and Teachers' Guide to Bilingualism. Clevedon: Multi-
lingual Matters.
Bialystok, E. 2001. Bilingualism in Development: Language, Literacy, and Cog-
nition. New York: CUP
Biedinger, N. 2010. Early ethnic inequality: The influence of social background and
parental involvement on preschool children’s cognitive ability in Germany.
Child Indicators Research, 3. 11–28.
Bild, E-R. & Swain, M. 1989. Minority language students in a French immersion
program: their French proficiency. Journal of Multilingual and Multicultural
Development, 10 (3). 255–274.
Bishop, D.V.M. 2003. TROG-2 Test for Reception of Grammar-2. London: Har-
court.
Burmeister, P. & Steinlen, A.K. 2008. Sprachstandserhebungen in bilingualen
Kindertagesstätten: Das erste Jahr. In G. Blell & R. Kupetz (eds.), Sammelband
des 3. Niedersächsischen Kolloquiums der Fremdsprachendidaktik vom 11. Mai
2007 in Hannover. Frankfurt: Lang. 129–146.
Cenoz, J. 2003. The additive effect of bilingualism on third language acquisition: A
review. International Journal of Bilingualism, 7. 71–87.
Commission of the European Communities, 2003. Communication from the
Commission to the Council, the European Parliament, the Economic and Social
Committee of the Regions. Promoting Language Learning and Linguistic

35
This expression has been taken from Gisela Håkansson’s presentation on YouTube with the
title ”Språkutveckling och flerspråkighet”. www.youtube.com/watch?v=kfvOJ9X_3tc
[Accessed 22 March 2013.]

181
Language Acquisition and Use in Multilingual Contexts

Diversity: An Action Plan 2004 – 2006. http://ec.europa.eu/education/doc/

official/keydoc/actlang/act_lang_en.pdf.
Crain, S. & Thornton, R. 1998. Investigations in Universal Grammar: A Guide to
Experiments in the Acquisition of Syntax and Semantics. Cambridge, MA: MIT
Press.
Dagenais, D. & Day, E. 1998. Multilingual children and classroom processes in
French immersion. The Canadian Modern Language Review, 54. 376–393.
Dunn, L[loyd] M., Dunn, L[eota] M., Whetton, C. & Burley, J. 1997. The British
Picture Vocabulary Scale II. Windsor: National Foundation for Educational
Research.
Elsner, D. 2007. Starting point “Mehrsprachigkeit”. Hörverstehen im Englisch-
unterricht der Grundschule. In D. Elsner, L. Küster & B. Viebrock (eds.),
Fremdsprachenkompetenzen für ein wachsendes Europa. Frankfurt am Main:
Lang. 91–105.
Franceschini, R. 2003. Viele Wege führen zum Ziel. Erfahrungen und Anregungen
aus der Beobachtung des Frühfranzösischunterrichts im Saarland. In
Ministerium für Bildung, Kultur und Wissenschaft & E. Hammes die Bernado:
Frühes Lernen – Bildung im Kindergarten. Eine Veranstaltung für pädagog-
isches Fachpersonal, Dokumentation. Saarbrücken: SCV. 1–16. 63–65.
Grimm, H., Aktas, M. & Frevert, S. 2001. SETK 3-5: Sprachentwicklungstest für
drei- bis fünfjährige Kinder: Diagnose von Sprachverarbeitungsfähigkeiten und
auditiven Gedächtnisleistungen [Manual]. Göttingen: Hogrefe.
Howell, P., Davis, S. & Au-Yeung, J. 2003. Syntactic development in fluent child-
ren, children who stutter, and children who have English as an additional
language. Child Language Teaching and Therapy, 19. 311–337.
Hurd, M. 1993. Minority language children and French Immersion: Additive
multilingualism or subtractive semi-lingualism? The Canadian Modern
Language Review, 49. 514–525.
Kersten, K., Rohde, A., Schelletter, C. & Steinlen, A.K. (eds.). 2010a. Bilingual
Preschools, Volume I: Learning and Development. Trier: WVT.
Kersten, K., Rohde, A., Schelletter, C. & Steinlen, A.K. (eds.). 2010b. Bilingual
Preschools. Vol. II: Best Practices. Trier: WVT.
Kersten, K., Steinlen, A.K., Tiefenthal, C., Wippermann, I. & Flyman Mattsson, A.
2010. Guidelines for language use in bilingual preschools. In K. Kersten et al.
(eds.), 2010b. 77–106.
Kersten, K., Piske, T., Rohde, A., Steinlen, A.K., Weitz, M. & Kurth, S. 2010.
ELIAS Grammar Test. University of Magdeburg: ELIAS.
Klieme, E. 2006. Zusammenfassung zentraler Ergebnisse der DESI-Studie.
http://www.kmk.org/fileadmin/veroeffentlichungen_beschluesse/2006/2006_03
_01-DESI-Ausgewaehlte-Ergebnisse.pdf.
Knapp, W. 2006. Language and Learning Disadvantages of Learners with a Migrant
Background in Germany: Preliminary Study. Council of Europe, Language
Policy Division, Strasbourg. http://www.coe.int/t/dg4/linguistic/Source/Knapp_
9oct_EN.doc.
Kolonko, B. 2001. Spracherwerb im Kindergarten. Grundlagen für die sprach-
pädagogische Arbeit von Erzieherinnen. Herbolzheim: Centaurus.
McMullen, K. 2004. Children of immigrants: How well do they do in school?
Education Matters. Statistics Canada Catalogue No. 81-004-XIE. http://www.
statcan.gc.ca//pub//81-004-x//200410//7422-eng.htm.

182
 Language Acquisition and Use in Multilingual Contexts

Mullis, I.V.S., Martin, M.O., Kenndey, A.M. & Foy, P. 2007. PIRLS 2006 Inter-
national Report. IEA’s in International Reading Literacy Study in Primary
Schools in 40 Countries. TIMSS and PIRLS International Study Center. Lynch
School of Education. Chestnut Hill, MA: Boston College.
OECD 2005. PISA 2003 Technical Report. Paris: OECD.
Rohde, A. 2005. Lexikalische Prinzipien im Erst- und Zweitspracherwerb. Trier:
WVT.
Rohde, A. 2010. Receptive L2 lexical knowledge in bilingual preschool children. In
K. Kersten et al. (eds.), 2010a. 45–68.
Rohde, A. & Tiefenthal, C. 2002. On L2 lexical learning abilities. In P. Burmeister,
T. Piske & A. Rohde (eds.), An Integrated View of Language Development.
Papers in Honor of Henning Wode. Trier: WVT. 449–471.
Ronjat, J. 1913. Enfant Bilingue. Paris: Champion.
Schilk, A., Ufert, D. & Steinlen, A.K. (in press). Erfahrungen bei der Einführung der
frühen Zweisprachigkeit in einem bilingualen Kindergarten. In T. Piske & A.K.
Steinlen (eds.), Bilinguale Programme in Kindertageseinrichtungen: Umsetz-
ungsbeispiele und Forschungsergebnisse. Erscheint beim Schneider Verlag
Hohengehren.
Steinlen, A.K. 2008a. Comprehension of L2 grammar in a German-English bilingual
preschool. In A. Nikolaev & J. Niemi (eds.), Two or More Languages: Pro-
ceedings from the 9th Nordic Conference on Bilingualism, August 10-11, 2006,
Joensuu. Studies in Languages, 43. University of Joensuu. 212–221.
Steinlen, A.K. 2008b. Immersionsmethode - wie funktioniert das? Aufbau und
Beginn einer immersiven Kita mit wissenschaftlicher Begleitung. In ESO,
Staatsministerium für Soziales, Sachsen, IFA, PONTES, Dokumentation der
Fachtagung Wege der frühen Mehrsprachigkeit in Kindertageseinrichtungen.
Stand der Wissenschaft. Erfahrungen. Austausch. Görlitz. 7–15.
Steinlen, A.K. & Rogotzki, N. 2009. Comprehension of L2 grammar in a bilingual
preschool: a developmental perspective. In T. Marinis, A. Papangeli & V.
Stojanovik (eds.), Proceedings of the Child Language Seminar 2007 – 30th
Anniversary, Reading: University of Reading. 163–173.
Steinlen, A.K. & Wettlaufer, J. 2005. Kiel Picture Pointing Test. Grammar and
Vocabulary. Online Test. Mimeo: University of Kiel.
Steinlen, A.K., Håkansson, G., Housen, A. & Schelletter, C. 2010. Receptive L2
grammar knowledge development in bilingual preschools. In K. Kersten et al.
(eds.), 2010a. 69–100.
Steinlen, A.K., Neils, K., Piske, T. & Trumpp, C. 2010. SETK 3–5: A develop-
mental language test on German for 3-to5-year-old children. In K. Kersten et al.
(eds.), 2010a. 119–135.
Swain, M. & Lapkin, S. 2005. The evolving socio-political context of immersion
education in Canada: Some implications for program development. Inter-
national Journal of Applied Linguistics, 15. 169–185.
Taylor, S. 1992. Victor: A case study of a Cantonese child in early French
immersion. Canadian Modern Language Review, 48. 736–759.
Taylor, S. 2006. ESL in bilingual education with a twist: Immigrant children in
Canadian French Immersion. In Bilingual Basics 8/1. http://www.tesol.org/
s_tesol/article.asp?vid=163&DID=6390&sid=1&cid=736&iid=6379&nid=307.
Weitz, M. & Rohde, A. 2010. German children's L2 English vocabulary in bilingual
kindergarten programmes in Germany: Why do the children's scores differ so

183
Language Acquisition and Use in Multilingual Contexts

strongly from each other? In C. Bongartz & J. Rymarczyk (eds.), Languages

across the Curriculum. Frankfurt/Main et al.: Peter Lang. 51–70.
Wesche, M.B. 2002. Early French immersion: How has the original Canadian model
stood the test of time? In P. Burmeister, T. Piske & A. Rohde (eds.), An
Integrated View of Language Development. Trier: WVT. 357–379.
Wode, H. 2001. Multilingual education in Europe: What can preschools contribute?
In S. Björklund (ed.), Language as a Tool - Immersion Research and Practices.
University of Vaasa: Proceedings of the University of Vaasa, Reports. 424–446.
Wode, H. 2010. Foreign language education in Europe: Why include preschools? In
K. Kersten et al. (eds.), 2010b. 5–33.

184
Mongolisk skrift, mongoliska staten
och mongolisk nationalism

Jan-Olof Svantesson

Bakgrund
Jag skall här försöka beskriva de yttre krafter som har bestämt valet av olika
mongoliska skriftsystem och deras vidare öden. Även om det tar emot för en
lingvist att erkänna det, tycks skriftens rent språkliga egenskaper spela en
mycket liten roll när nya skriftsystem införs (eller avskaffas). I stället verkar
det finnas tre huvudsakliga krafter som bestämmer: stat, religion och
nationalism. I det mongoliska fallet har staten haft huvudrollen, men nation-
alismen är också inblandad.
Enligt Denise Schmandt-Besserat (1997) utvecklades det äldsta kända
skriftsystemet, sumerisk kilskrift, från lerfigurer som representerar varor och
som användes för bokföring av handelstransaktioner. Skriftens uppkomst
hade i det fallet att göra med statens ekonomiska behov, även om det kan
vara svårt att skilja stat och religion åt i detta och andra antika samhällen.
Exempel på skriftsystem som har införts av en stat helt utan religiösa
baktankar, eller kanske snarare med antireligiösa baktankar, är de kyrilliska
skriftsystem som på 1930-talet utarbetades för minoritetsspråk i Sovjet-
unionen, som uzbekiska, kazakiska, azerbajdjanska, tadzjikiska och många
andra. Andra exempel är de latinska skriftsystemen för minoritetsspråk i
Kina som skapades vid början av 1950-talet, t.ex. för zhuang, bouyei, wa och
miao (se Svantesson 1991).

Skrift och religion

Religionen är antagligen den faktor som har haft störst betydelse för sprid-
andet av olika skriftsystem. Missionärers iver att sprida heliga skrifter
genom att översätta dem till folkens språk har nog varit den viktigaste
orsaken till att olika alfabet har spritts över världen.
Det arabiska alfabetet spreds med islam till åtskilliga språk, som turkiska,
persiska och malajiska. I många fall var religionen den enda orsaken till
bytet av skriftsystem, t.ex. bytet från fornmalajisk till arabisk skrift. I Syd-

185
Language Acquisition and Use in Multilingual Contexts

östasien var det buddhistiska munkar som spred de besläktade skriftsystemen

för språk som khmer, thai, lao och många andra. På liknande sätt spred
kristna missionärer det latinska alfabetet till både skriftlösa språk och till
sådana som redan hade en skrift, t.ex. runor, i alla delar av världen. Ett annat
välkänt exempel är att de makedonska munkarna Kyrillos och Methodios
under 800-talet skapade en skrift för de slaviska folken.
Denna verksamhet fortsätter än idag. Ett sentida och typiskt exempel är
den skrift som den svensk-amerikanske baptistmissionären Ola Hanson
(1864–1927), född i Åhus, skapade 1895 för att skriva kachin, ett tibeto-
burmanskt minoritetsspråk i Burma och Kina. Hans ändamål var just att
översätta kristna texter: han översatte hela Bibeln och dessutom 400 svenska
och amerikanska psalmer till kachin (Sword 1954). Hansons latinskbaserade
skrift används nu allmänt av kachinerna i Burma, och med få ändringar har
den blivit officiell skrift även för de kachiner (jingpo) som bor i Kina.

Skrift och nationalism

Uppkomsten av en nationalstat är ofta nära förknippad med skapandet av ett
skriftspråk grundat på invånarnas talade språk och använt som riksspråk i
den nya staten (Anderson 1991). Det gäller t.ex. finska, tjeckiska och norska,
där de nyskapade skriftspråken var bärare av nationalistiska strömningar.
Ibland var religion också inblandad och bibelöversättningar var ofta bland de
första texterna på de nyskapade skriftspråken.
I enstaka fall har messianska rörelser bland minoritetsfolk haft ett ny-
bildat skriftsystem som sin kärna. Det var fallet med pahawh-hmong-
skriften, skapad 1959 av en sådan rörelses ledare Shong Lue Yang (1929–
71), av sina anhängare kallad Niam Ntawv ”Skriftens Moder”. Denna helt
nyskapade – och lingvistiskt intressanta – skrift användes under en tid av
delar av hmong-folket i Laos (Smalley et al. 1990).
Något liknande var kanske orsaken till att Ola Hansons kachin-skrift i
Burma blev en sådan framgång. Även om Hanson var kristen missionär och
inte en messiansk kachin-ledare i opposition mot den burmanska majoriteten
verkar han ha fått en sådan roll. Han var ovanligt framgångsrik som missio-
när och omvände nästan halva kachinfolket till baptismen. Numera verkar
han närmast ses som en mytologisk kulturhjälte. När hundraårsjubileet av
kachinskriften firades av 30.000 personer – var tjugonde kachin – i Manda-
lay 1995 var den amerikanske baptistpastorn Tom Steller inbjuden till
jubileet som representant för Hansons gamla kyrka. Efter att ha hört
uttalanden som ”tidigare var vi vildar men tack vare Ola Hanson är vi nu
civiliserade människor” kände han sig föranlåten att hålla en predikan där
han betonade att äran i första hand tillkom Herren och inte Ola Hanson
(Steller 1996).

186
 Language Acquisition and Use in Multilingual Contexts

Uigur-mongolisk skrift
Åter till Mongoliet. Enligt den kinesiska Yuán-dynastins historia (元史
Yuánshǐ), sammanställd 1370, tillkom det första skriftsystemet för mongo-
liska på order av den mongoliska statens skapare Djingis khan (1162–1227).
En uigurisk ämbetsman Tatatong’a, som kunde det uiguriska skriftspråket
bra, tillfångatogs år 1204 och beordrades att lära prinsarna att skriva mongo-
liska med det uiguriska alfabetet. Uigurerna var ett turkiskt folk i Central-
asien, troligen inte de direkta förfäderna till de nutida uigurerna, utan till ett
litet minoritetsfolk i Kina, shira yugur, de ”gula uigurerna”. Sitt alfabet hade
de lånat från sogderna, ett iranskt folk i Centralasien, som i sin tur hade lånat
det från den syriska skriften. Syriska skrevs, liksom arabiska och andra
semitiska språk, i rader från höger till vänster, men sogderna vände skriften
ett kvarts varv så att den kom att skrivas uppifrån och ner i kolumner som
läses från vänster till höger, och detta skrivsätt togs över av uigurer och
mongoler.
Den äldsta bevarade mongoliska texten är en steninskription kallad
Djingis khans sten, daterad 1225. Det finns relativt få uigur-mongoliska
texter bevarade från 1200- och 1300-talen. De flesta är kejserliga påbud eller
brev; några har hittats i Vatikanbiblioteket. Den äldsta längre texten är
Mongolernas hemliga historia, troligen skriven år 1228 eller 1240. Den
finns bevarad endast i en version där det mongoliska språket har skrivits med
hjälp av kinesiska tecken, men de flesta forskare anser att originalet måste ha
varit skrivet med uigur-mongolisk skrift.
De äldsta religiösa (buddhistiska) skrifterna är lite senare, från början av
1300-talet, så religionen spelade ingen roll för tillkomsten av det uigur-
mongoliska skriftspråket, det var helt och hållet en statlig angelägenhet, även
om skriften senare användes för alla typer av texter, även religiösa.
Den uigur-mongoliska skriften används fortfarande av mongolerna i Kina,
men i Mongoliet ersattes den av kyrillisk skrift under 1940-talet. En reform-
erad variant av skriften skapades 1648 av den dzungariske laman Zaya
Pandita (1599–1662). Denna skrift användes för att skriva oiratiska
(västmongoliska), den mongoliska variant som talas av mongolerna i västra
Mongoliet och nordvästra Kina, och som efter hand hade utvecklats till ett
självständigt språk. Den oiratiska skriften, som kallas todo bicig ’den klara
skriften’, var mer anpassad till talspråket och nästan helt fonematisk. Den
används fortfarande av en del västmongoler i Kina. Många av dessa
mongoler utvandrade på 1600-talet till södra Ryssland och slog sig ner vid
Volgas utlopp i Kaspiska havet. De fick namnet kalmucker och använder
numera det kyrilliska alfabetet.

187
Language Acquisition and Use in Multilingual Contexts

’Phags-pa-skrift
Mindre än hundra år efter att den uigur-mongoliska skriften hade kommit till
skapades ett nytt skriftsystem på initiativ av Djingis khans sonson Kubilai
(1215–94), grundaren av den mongoliska Yuán-dynastin som erövrade hela
Kina. Denna skrift, som var baserad på det tibetanska alfabetet, designades
av laman ’Phags-pa (1235–80). ’Phags-pa-skriften gjordes 1269 till Yuán-
dynastins officiella skrift, och skulle användas i hela det mongolisk-
kinesiska imperiet för att skriva de tre officiella språken mongoliska, tibetan-
ska och kinesiska. I verkligheten kom den inte att ersätta den uiguriska
skriften för mongoliska och definitivt inte de kinesiska och tibetanska skrift-
systemen. Endast ett sextiotal mongoliska ’Phags-pa-texter är bevarade (och
inte många fler på tibetanska och kinesiska). Troligen finns det oupptäckta
texter i tibetanska och mongoliska arkiv, ett mongoliskt ’Phags-pa-dokument
dök för några år sen upp som en illustration i en tibetansk turistbroschyr,
som forskarna genast kastade sig över. De flesta ’Phags-pa-dokumenten är
kejserliga befallningar, men det finns också några religiösa steninskriptioner.
’Phags-pa-skriften användes alltså endast för officiella ändamål, det enda
undantaget är en stenkruka med inskriptionen sajin darasu ’gott vin’.
’Phags-pa-skriften verkar ha använts mer som en symbol för Yuán-statens
makt än som en praktiskt användbar skrift. Den överlevde inte dynastins fall
1368.

Andra äldre skriftsystem för mongoliska

Även om den uigur-mongoliska skriften var den enda som allmänt användes
av mongolerna fram till mitten av 1900-talet finns det enstaka äldre
mongoliska texter skrivna på andra sätt. Det mest kända exemplet är, som
redan nämnts, Mongolernas hemliga historia, där mongoliska språket skrevs
med kinesiska tecken. Kinesisk skrift används också i en del mongoliska
ordlistor och parlörer från 1300-talet. Det finns också en omfattande ordbok,
Muqaddimat al-adab ‘Inledning till litteraturen’, från 1300-talet, där mongo-
liska ord och deras motsvarigheter i centralasiatisk turkiska (tjagataiska) är
skrivna med det arabiska alfabetet. Arabisk skrift användes av de central-
asiatiska turkfolken ända fram till 1920-talet.
Mongoliska texter skrivna med dessa skriftsystem ger värdefulla upplys-
ningar om hur mongoliska uttalades i äldre tider, och språkhistorikerna är
mycket förtjusta i dem. Men de var nog aldrig avsedda för att användas i
dagligt bruk.

188
 Language Acquisition and Use in Multilingual Contexts

Kyrillisk-mongolisk skrift
Efter grundandet av Folkrepubliken Mongoliet 1924 blev Mongoliet nära
förbundet med Sovjetunionen, fastän det hela tiden förblev en självständig
stat. Detta återspeglades också i skriftpolitiken. Under 1920-talet och början
av 1930-talet skapades nya skriftsystem baserade på det latinska alfabetet för
många minoritetsspråk i Sovjetunionen, däribland burjatiska och kalmuck-
iska, som är nära besläktade med mongoliska – skillnaderna är ungefär lika
stora som mellan svenska och danska. I många fall handlade det om språk
som tidigare inte hade någon skrift, men ibland ersattes ett tidigare alfabet
med det latinska. Förutom för kalmucker och burjater gäller det exempelvis
för de muslimska centralasiatiska turkfolken vars språk tidigare skrevs med
det arabiska alfabetet.
Runt 1937 förändrades emellertid den sovjetiska skriftpolitiken drastiskt
och man övergick från latinsk till kyrillisk skrift, uppenbarligen som ett slags
manifestation av nationell enighet och rysk överhöghet.
Burjater och kalmucker hade först övergått till latinsk skrift och sedan till
kyrilliska vid slutet av 1930-talet. Detta sågs som ett steg mot ett modernare
samhälle och kultur av många mongoliska intellektuella som ansåg att den
gamla mongoliska skriften var alltför komplicerad och alltför avlägsen från
det moderna språket för att kunna tjäna som ett skriftsystem för befolk-
ningens stora flertal.
Vid början av 1930-talet gjordes ett kortvarigt försök att införa en latinsk-
baserad skrift för mongoliska på initiativ av den burjatiske intellektuelle
Tsyben Zjamtsarano (1880–1942), se Grivelet (1997). Det fick inget genom-
slag, och uigur-mongoliskan förblev den allmänt använda skriften tills det
kyrilliska alfabetet infördes genom ett regeringsbeslut av den 25 mars 1941.
Detta var troligen grundat på sovjetiska påtryckningar, men också på en
önskan från mongoliska intellektuella att utplåna den utbredda analfabet-
ismen genom ett skriftsystem som låg närmre det talade språket.
Utformningen av den kyrilliskbaserade skriften för mongoliska gjordes
till stor del av författaren och språkvetaren Tsendiin Damdins r en (1908–
86), en av de mongoliska intellektuella som ansåg att den gamla skriften
hade spelat ut sin roll. För att skynda på processen hämtades han och sattes
utan familjens vetskap i husarrest i en sommarstuga i berget Bogd Uul söder
om Ulaanbaatar och släpptes inte ut förrän han hade formulerat reglerna för
hur mongoliska skulle skrivas med det kyrilliska alfabetet.
Andra ledande intellektuella som Bjambyn Rintjen (1905–77) var starka
motståndare till den kyrilliska skriften, och ansåg att den inte lämpar sig för
att skriva mongoliska (se t.ex. Rinčen 1958). Denna uppfattning har spritts
framgångsrikt även utanför Mongoliet men är ur lingvistisk synpunkt helt
ogrundad och baserad på missuppfattningar av den moderna mongoliskans
fonologi och på en ovilja att se det moderna språket som ett självständigt
språk, på många sätt skilt från fornmongoliskan. I själva verket är den

189
Language Acquisition and Use in Multilingual Contexts

kyrilliska mongoliska skriften en i stort sett fonematisk skrift för den

normerande dialekten halh (chalcha) i Mongoliet. Den gamla (uiguriska)
skriften för mongoliska skiljer sig däremot ganska mycket från det moderna
språket och har dessutom en hel del språkliga tillkortakommanden. Flera ljud
som skiljs åt i det mongoliska uttalet (och i den kyrilliska skriften) skrivs
med samma bokstav i den gamla skriften, t.ex. skiljs inte t och d åt och inte
heller u och o. Orsaken är att dessa ljud inte skildes åt i uiguriska som
skriften ursprungligen användes för. Dessutom var det ju så att den gamla
skriften kom till under 1200-talet och språkförändringar som skett därefter
har inte fått något avtryck i skriften. Man skall emellertid inte överdriva
dessa svårigheter. Skillnaderna mellan den gamla mongoliska skriften och
det moderna språket är inte större än de som finns i en del språk som
använder det latinska alfabetet, som engelska eller franska. Den gamla
skriften har dessutom den fördelen att den är dialektneutral.
Den kyrilliska skriften ersatte efter hand den gamla mongoliska skriften
och gjordes 1946 till det enda officiella skriftsystemet. Övergången till
kyrillisk skrift hjälpte säkert till att utrota analfabetismen, vilket var huvud-
målet för de intellektuella som förespråkade skriftreformen, även om det
kanske inte var den avgörande faktorn.
Den kyrilliska skriften sågs däremot också som en symbol för sovjetisk
överhöghet. När sovjetmakten försvagades mot slutet av 1980-talet återupp-
stod den gamla skriften. Det första allmänt synliga tecknet på detta var en
neonskylt Monggol sigudan ’Mongoliska posten’ i den gamla skriften som
1986 sattes upp på Centralpostkontoret vid S hbaa tartorget i centrala Ulaan-
baatar. Utvecklingen kulminerade i ett parlamentsbeslut från 1991 att den
gamla skriften under 1994 skulle ersätta den kyrilliska för alla officiella
ändamål. Den gamla skriften infördes som förstaskrift i grundskolorna under
de tre läsåren 1991/92 till 1993/94, och kurser i den gamla skriften gavs i
den statliga televisionen och andra medier.
Fastän försöket att återinföra den gamla skriften välkomnades av många
intellektuella, som redan kunde den väl, blev det knappast någon framgång
hos den stora allmänheten, som bara kände till den kyrilliska skriften.
Planerna på att byta skriftsystem avbröts plötsligt genom ett parlaments-
beslut i juli 1994. Kyrilliska är nu den enda allmänt använda skriften, även
om den gamla skriften fortfarande är officiell skrift jämsides med kyrilliska.
Kyrilliska är helt dominerande och den gamla skriften används mest som
dekoration. De olika mongoliska skriftsystemen illustreras i bilden på nästa
sida.

190
 Language Acquisition and Use in Multilingual Contexts

Framgång och motgång

I den mongoliska skriftens historia har vi sett tre åtminstone delvis fram-
gångsrika statliga initiativ: när den gamla (uigurisk-)mongoliska skriften
infördes under 1200-talet, när ’Phags-pa-skriften infördes under 1300-talet
och när den kyrilliska skriften infördes på 1940-talet. Återinförandet av den
gamla skriften på 1990-talet misslyckades däremot.
’Phags-pa skriftens öde var nära relaterat till Yuán-staten. Den skapades
av och för staten och slutade användas när staten upphörde. Ingen annan
faktor tycks ha varit inblandad. Två framgångsrika mongoliska skriftsystem
introducerades av Djingis khans nya mongoliska stat och av 1940-talets
sovjetstödda mongoliska stat. Båda var kraftfulla regimer utan demokratiska
hänsyn, vilket nästan övertydligt framgår av den något bisarra metoden som
båda använde: att tillfångata och tvinga en ledande intellektuell att skapa ett
nytt skriftspråk.
Det intressantaste fallet är nog försöket att återinföra den gamla skriften
under 1990-talet, vilket misslyckades trots de rådande nationalistiska och
antiryska känslorna, som symboliserades av glorifierandet av Djingis khan,

191
Language Acquisition and Use in Multilingual Contexts

till exempel genom ett frenetiskt firande av olika jubileer med anknytning till
honom (se t.ex. Grivelet 2001). Sådana firanden, eller över huvud taget
positiva omnämnanden av Djingis khan var omöjliga under den sovjet-
dominerade tiden. I Ryssland var (och är) bilden av Djingis khan präglad av
föreställningen om det ”mongoliska oket”.
Uppenbarligen underskattade man den tid och ansträngning som behöv-
des för att lära sig den gamla skriften. Även högmotiverade inlärare gav upp
när de förstod att det inte bara handlade om att byta ut de kyrilliska
bokstäverna mot de uiguriska, utan att man var tvungen att lära sig ett helt
annat system med sina egna regler, mer komplicerade än de som man redan
kunde. Det fanns också ekonomiska skäl, som de investeringar som skulle
behövas för att kunna trycka böcker och tidningar i den gamla skriften. Ett
tredje skäl kan ha varit att man bekymrade sig för att den gamla mongoliska
skriften, som inte används för något annat språk än mongoliska, skulle
isolera Mongoliet från resten av världen.

Till sist
Nya skriftsystem har införts i Mongoliet nästan enbart på statligt initiativ.
Den uiguriska skriften infördes för att uppfylla den nya mongoliska statens
krav. ’Phags-pa-skriften infördes mer som en symbol för Yuán-dynastins
enhet än av något egentligt behov, eftersom det redan fanns välfungerande
skriftspråk för de olika språk som ’Phags-pa-skriften användes till, kinesiska,
mongoliska och tibetanska. Den kyrilliska skriften infördes efter sovjetiska
påtryckningar men också för att avskaffa analfabetismen. Försöket att
återinföra den gamla mongoliska skriften var en del av den nationalistiska
reaktionen mot den tidigare ryska överhögheten. De tre förstnämnda lyck-
ades, åtminstone för en tid, medan den sista misslyckades. En orsak till detta
var att den mongoliska staten under 1990-talet var både svagare och mer
demokratisk än vid de andra tillfällena.
Men jag tror att en annan faktor var den avgörande. När det kyrilliska
alfabetet infördes på 1940-talet var de flesta mongoler analfabeter, men
under 1990-talet var de allra flesta läskunniga och använde det kyrilliska
alfabetet i sina dagliga liv. Erfarenheter från förändringar av andra allmänt
använda skriftspråk visar att det är svårt för språkanvändarna att acceptera
ens ganska små förändringar, som de nyligen genomförda tämligen mar-
ginella förändringarna av tysk stavning, som väckt starkt motstånd. Den
inneboende styrkan i ett etablerat och välfungerande skriftsystem som
används aktivt av en majoritet av befolkningen kan stå emot både statliga
dekret och nationalistiska känslor, som det mongoliska exemplet visar.

192
 Language Acquisition and Use in Multilingual Contexts

Litteratur
Anderson, B. 1991. Imagined communities. Revised edition. London: Verso.
Grivelet, S. 1997. The Latinization attempt in Mongolia. In Á. Berta (ed.), The
historical and linguistic interaction between Inner-Asia and Europe: pro-
ceedings of the 39th Permanent International Altaistic Conference. Szeged:
University of Szeged. 115–120.
Grivelet, S. 2001. Digraphia in Mongolia. International Journal of the Sociology of
Language, 150. 75–93.
Rinčen, B. 1958. L’inscription sinomongole de la stèle en l’honneur de Möngke
qagan. Central Asiatic Journal, 4. 130–143.
Schmandt-Besserat, D. 1997. How writing came about. Austin: University of Texas
Press.
Smalley, W., Vang, C.K. & Yang, G.Y. 1990. Mother of writing: the origin and
development of a Hmong messianic script. Chicago: University of Chicago
Press.
Steller, T. 1996. Reflections on the centennial celebration of Kachin literacy,
Mandalay, Myanmar, Dec 30, 1995 – January 2, 1996. [Tidigare tillgängligt på
MissionNet.org]
Svantesson, J-O. 1991. Tradition and reform in China’s minority languages.
International Journal of Applied Linguistics, 1. 70–88.
Sword, G. 1954. Light in the jungle: life story of Ola Hanson of Burma. Chicago:
Baptist Conference Press.

193
The Obligatory Contour Principle in
Swedish

Joost van de Weijer

Introduction
Constraints on phonological word structure exist in all languages, that is, the
elements of the sound system of a language (the consonants and the vowels)
combine with one another in specific ways. Some elements combine
frequently, others combine rarely or never. This system of language sounds
and how they combine is known as the phonotactic structure of the language.
As an example, one of the phonotactic rules in English is that a word can
start with the combination / kl / but not with the combination / kn /. Typi-
cally, phonotactic rules are language specific. Certain combinations of
speech sounds may be forbidden in one language, but allowed in the next.
The combination / kn / from the previous example is a legitimate word onset
in other Germanic languages such as Dutch, German or Swedish.
Phonotactic rules need not be all-pervasive, that is, some rules appear to
exist, but counterexamples can also be found. The consequence is that some
combinations of speech sounds are attested in a language, but with a
frequency that is far less than would be expected given the frequencies of the
individual phonemes (e.g. Kessler and Treiman 1997; Pierrehumbert, 1994).
One example of such a not-all-pervasive phonotactic constraint is that
certain vowels tend to be combined with some consonants but not with
others (Kessler and Treiman 1997). In English, for instance, / æ / is a
common vowel and / l / is a common syllable offset, but there are remark-
ably few words that end in / æl /. Kessler and Treiman (1997) identified
several of these restrictions of vowels with subsequent consonants and took
this finding as supporting evidence that there is a natural division between
syllable onset and syllable rime. A second example is that in English approx-
imately 200 possible morpheme internal triconsonantal clusters (as in
embrace) are expected given the relatively high frequencies of the individual
consonants (Pierrehumbert 1994). However, only approximately 50 out of
200 are attested in a dictionary.

194
 Language Acquisition and Use in Multilingual Contexts

The focus of this paper is the existence of a phonotactic constraint in

Swedish, namely the avoidance of identical consonants within a word. This
characteristic, the avoidance of identical segments, is a well-known feature
of language phonology. It has been observed in other contexts and in other
languages as well. Within phonology this tendency has been formulated as
the obligatory contour principle (OCP), which states that “adjacent identical
segments are prohibited” (Clements & Hume 1995). Originally the OCP was
used to explain the absence of adjacent identical lexical tones, but at a later
stage it has also been applied to consonants.
The systematic investigation of consonant-repetition avoidance has
mainly focused on Semitic languages, Arabic and Hebrew (e.g. McCarthy
1986; Berent & Shimron 2003; Frisch et al. 2004). In Hebrew, a word root
consists of three consonants to which the vowels are added. In normal
written Hebrew, only these consonants are represented and the vowels are
omitted. The first two consonants in these roots are seldom identical and
usually not homorganic (Berent & Shimron 2003). The last two consonants
can be identical, but in these cases, presumably, the underlying form of the
root only contains two consonants. Semitic languages present a relatively
clear case because of the homogeneous structure of word roots: a skeleton
consisting of three consonants. But what about other languages?
The investigation of consonant repetition in languages such as the
Germanic languages is less straightforward. First of all, this is because the
pattern of repetition avoidance is not as clear-cut as in Semitic languages.
There are many examples of words in which a consonant occurs twice (e.g.
people, moment, dead, system, etc.), or even three times (e.g., census).
Second, syllable structure is more complex in Germanic languages. Instead
of simple consonants, syllable onsets and offsets frequently consist of
clusters of consonants. Co-occurrence restrictions within consonant clusters
are not the same as those between consonant clusters. Identical consonants
are only allowed in two different positions (syllable onset and syllable offset,
as in space), but not within the same position. There are no uninflected forms
that end in –sts, or –sks, for instance.
Nevertheless, numerous examples exist supporting the idea that com-
binations of similar or identical elements tend to be avoided in other
languages as well, albeit to a lesser extent than in Semitic roots. Clements
and Keyser (1983) and Davis (1989) pointed out that there are no
monosyllabic words in English with the form sCVC, in which the two Cs are
identical non-coronal consonants, or two nasal consonants. There are no
such words as spip, skak, or snang. Another example is that of morpho-
logical haplology (Stemberger 1981): suffixes are deleted when the adjacent
consonant in the stem is homophonous. In English, for example, the genitive
s is not realized in plural forms that end in –s (cf. the *boys’s books with the
children’s books). Similar examples exist in other languages as well. In
Swedish, the present tense markers –er or –ar are not realized when the verb

195
Language Acquisition and Use in Multilingual Contexts

stem ends in –r, e.g. jag kommer, (‘I come’), but jag hör – instead of *jag
hörer (‘I hear’). Third, there are examples of historical sound changes that
appear to be motivated by achieving greater dissimilarity between
phonological elements (Hock & Joseph 1996). The original Swedish word
for ‘key’ was lyckel, which has become nyckel in modern Swedish.
Similarly, the Italian word tartuffeli became Kartoffel (‘potato’) when
adopted into German.
On the other hand, there are also exceptions, notably in the category of
expressive words (Fudge 1970). Repeated consonants are regularly found in
(at least) three groups of words that somehow take a special status within a
language. These groups are onomatopoetic words (e.g. babble, giggle,
cackle, mumble, crack, pop); words typically used by children (e.g. mommy,
daddy, cookie, baby, nanny, etc.); taboo (slang) words or words with a
negative connotation (e.g. twit, twat, crook). On top of that, repeated
consonants are a natural characteristic of reduplicative words such as criss-
cross, zigzag, hotchpotch, hocus-pocus. These latter words, however, may be
considered as combinations of two simplex words.
Given these observations, the investigation of consonant identity avoid-
ance within non-Semitic languages appears worth the effort. Are words with
repeated consonants indeed underrepresented in the vocabulary in other
languages as well, albeit to a lesser extent than in Semitic languages, or is
the number of words with repeated consonants approximately equal to the
number that would be expected given the relative frequencies of the
consonants in the language?
A first exploration of the issue (van de Weijer 2003) showed that, in
Swedish, words with at least two repeated consonants are infrequent. The
material used in that study consisted of 8887 monomorphemic words with a
total token frequency of nearly 29 million. The type frequency of words with
identical consonants was 1001 (11.26%), but the token count was only
1.57%. In other words, there was a considerable number of word types with
repeated consonants, but since these words have low token frequency, they
are not encountered very often in everyday language.
A second study (van de Weijer 2005) suggested that listeners are sensitive
to the presence of identical consonants. In an auditory lexical-decision
experiment, Swedish listeners were presented with nonsense words and real
words. Repeated consonants occurred in one third of the real words (e.g. tält
‘tent’, lokal ‘room’), and in one third of the nonsense words (e.g. tift, lanel).
The subjects were faster in rejecting the nonsense words with repeated
consonants than those without, but, on the contrary, they were slower in
accepting real words with identical consonants than those without. This
result suggests that the pattern of identical consonants is relatively un-
common and that this is part of the implicit knowledge of native speakers.
The present study presents evidence from Swedish. The results show that,
overall, the expected numbers of words with identical consonants are higher

196
 Language Acquisition and Use in Multilingual Contexts

than the actual observed numbers. Furthermore, it is shown that the extent to
which a consonant is likely to be repeated varies. Some are very resistant to
repetition (e.g. / l / and / r /), others are less resistant, and remarkably, for
/ b / the observed frequency is higher than the expected frequency. The
method of analysis is described in the following section.

Method and materials

The observed numbers of words with identical consonants were determined,
and then compared with the expected numbers. The principle of the
calculation of the expected frequencies was as follows. Consonants such as
/ h / and / ŋ / are relatively infrequent and their occurrence is predominantly
limited to either word onset or word offset. Consequently, not many words
with repeated / h / or / ŋ / are to be expected. Other consonants, such as / p /
and / s /, are frequent and can occupy different positions within the word. It
is therefore only natural to expect more words with repeated / p / or / s /.
Secondly, the expected numbers of words with repeated consonants are de-
termined by the amount of consonants that occupy different positions within
the word. A monosyllabic word can only contain repeated consonants when
there is at least one consonant in word-onset position, and in word-offset
position. The probability that consonants become repeated increases when
more consonant ‘slots’ (that is, before or after a vowel) in a word are filled.
For monosyllabic words, there can be consonants in zero to two positions:
word onset (WO) and word offset (WF). With disyllabic words, there are
two additional slots, one for the coda of the first syllable (SF), and another
one for the onset of the second syllable (SO). These two factors were taken
into account when calculating the expected numbers of words with repeated
consonants.
The material for this study consisted of a list of 5388 monosyllabic and
disyllabic phonologically transcribed Swedish words. Each word in this list
was categorized as to whether or not the word contained one or more
consonants, and the positions (slots) that these consonants occupied. The
word brand (‘fire’), for instance, contains consonants in WO and WF
position, and the word porträtt (‘portrait’) contains consonants in all four
positions. There are 16 different possible combinations, but not all of them
occurred. Table 1 gives an overview with example words from the material.

197
Language Acquisition and Use in Multilingual Contexts

Table 1: Consonant slots. Shaded cells indicate which slot is filled with one or more
consonants.
WO SF SO WF example n
i, å 3
ko, två 134
eko, edra 76
— 0
örn, ärr 93
flykt, mynt 1824
spöke, smida 938
övrig, idrott 141
yrke, öppna 68
— 0
— 0
hustru, löfte 530
— 0
gratis, trubbig 920
anspråk, artist 122
doktrin, marknad 539

The probability (relative frequency) that a consonant occurred in any of the

four positions was calculated as the number of words with the consonant in a
certain position divided by the total number of words that contained one or
several consonants in that position. For instance, there were 4885 words with
one or more consonants in WO position, and 3639 words with one or more
consonants in WF position. The consonant / b / (bottom row of Table 2)
occurred 419 times in WO position and 29 times in WF position. The
probability that / b / occurred in WO and WF position was 419 / 4885 =
.08577, and 29 / 3639 = .00797, respectively. Table 2 shows absolute and
relative frequencies for each of the Swedish consonants in the four different
slots.
The likelihood that a consonant would occur more than once in a word
was calculated as a binomial probability based on the multiplication of the
relative frequencies of the consonant. For instance, the probability of finding
/ b / in word onset and word offset position in a WO WF word becomes
.08577  .000797 = .00068. The outcome multiplied with the number of
WO WF words yields the expected number of words with two / b /. Since
there were 1824 words with this structure in the word list (see right column
of Table 1), the expected number with two / b / was equal to
.00068  1824 = 1.25. The observed number of words in this category was
equal to 2 (i.e. bob and bomb). In a binomial distribution with p = .00068
and n = 1824, the one-tailed probability of obtaining two hits or more is .87,
which is much larger than a critical value of .05. In other words, the
observed frequency in this case was not significantly larger than the expected
frequency.

198
 Language Acquisition and Use in Multilingual Contexts

Table 2: Absolute and relative consonant frequencies in different slots.

WO SF SO WF
d 223 0.04565 38 0.03018 290 0.08698 266 0.07310
f 336 0.06878 24 0.01906 71 0.02130 53 0.01456
g 196 0.04012 33 0.02621 147 0.04409 169 0.04644
h 245 0.05015 0 0.00000 13 0.00390 0 0.00000
j 186 0.03808 23 0.01827 153 0.04589 88 0.02418
k 539 0.11034 92 0.07307 370 0.11098 414 0.11377
l 536 0.10972 140 0.11120 378 0.11338 509 0.13987
m 348 0.07124 130 0.10326 185 0.05549 219 0.06018
n 211 0.04319 224 0.17792 260 0.07798 554 0.15224
p 398 0.08147 21 0.01668 183 0.05489 137 0.03765
r 807 0.16520 243 0.19301 342 0.10258 661 0.18164
s 865 0.17707 177 0.14059 307 0.09208 548 0.15059
t 516 0.10563 59 0.04686 518 0.15537 642 0.17642
v 320 0.06551 10 0.00794 163 0.04889 83 0.02281
ɧ 127 0.02600 0 0.00000 18 0.00540 8 0.00220
ɕ 89 0.01822 0 0.00000 3 0.00090 5 0.00137
ŋ 0 0.00000 76 0.06037 47 0.01410 194 0.05331
ʃ 2 0.00041 1 0.00079 3 0.00090 48 0.01319
b 419 0.08577 4 0.00318 140 0.04199 29 0.00797
n 4885 1259 3334 3639

A consonant can, in principle, occur as many times in a word as there are

slots in the word, and sometimes it does. The probability that a consonant
would occur in the first two slots of, for instance, a WO SO WF (e.g., cider)
was calculated as the product of the relative frequencies of the consonant in
these positions, times the probability that the consonant would not occur in
the last position. This probability was calculated for all three combinations
of two groups out of three, after which the three probabilities were added.
The probability that the consonant occurred three times in the word was
equal to the product of the three relative frequencies. The total probability,
then, that a consonant occurred at least twice was the sum of the four
probabilities:
p(WO and SO but not WF) +
p(WO and WF but not SO) +
p(SO and WF but not WO) +
p(WO and SO and WF)

In the case of / b / there were four out of 1259 words with / b / in SO

position which makes p(SO) for / b / equal to .00318. The probability of

199
Language Acquisition and Use in Multilingual Contexts

finding at least two / b / in a WO SO WF word then was equal to

(.08577  .00318  .99203) + (.08577  .00797  .99682) + (.00318  .00797
 .91423) + (.08577  . 00318  .00797) = 0.00098. Again, this probability
was multiplied by the number of words with this particular structure to find
the expected number of words with identical consonants.
The probabilities for words with four groups of consonants were
calculated accordingly. Thus, this probability was equal to the probability
that a consonant occurred twice in the word, plus the probability that it
occurred three times, plus the probability that it occurred four times.

Results and discussion

The table in the appendix shows the results. Each cell contains the expected
frequency followed by the observed frequency. Cells that contain pairs
where the expected frequency is significantly larger (p < .05) than the
observed frequency are shaded. The final column shows the row totals
(collapsed over all words). The bottom row shows the column totals
(collapsed over all consonants). The table only shows those words with at
least two consonant slots.
Across all words, the expected total is more than twice as large as the
observed total (rightmost bottom cell in the table). Examining the row totals
in the right column results in a rough pattern. The consonants that are least
likely to be repeated within words are the alveolars, notably / l / and / r /, but
also / d s t n /. From the frequencies shown in Table 2, for instance, it can be
seen that / l / is a consonant that occurs regularly in word onset and word
offset. Nevertheless, there were no words in the word list with / l / in word
onset and word offset. Additionally, expected frequencies for / k /, / m / and
/ j / are substantially higher than observed frequencies. Then there are a few
consonants for which the expected frequencies are roughly equal to the
observed frequencies. These include / f g p v h ɧ ɕ ŋ /. The latter four,
however, are of low overall frequency, which reduces confidence in the
results. Finally, / b / shows up as an exceptional consonant in that the ob-
served frequency of words is larger than the expected number. Examining
the individual cells for / b / reveals that there are unexpectedly many
WO SO WF words with two / b /. These words are: babbel, bubbel, bourbon,
bebis, bebop, bibel, kebab, and babord.
In conclusion, the OCP mechanism has previously been observed in the
Semitic languages, and previous informal evidence has suggested that it is
also operational in other languages. The results of the present study provide
support for these informal observations with data from Swedish. It was
found that the frequencies of words with repeated consonants were gene-
rally, and often significantly, lower than the expected frequencies.

200
 Language Acquisition and Use in Multilingual Contexts

References
Berent, I. & Shimrom, J. 2003. Co-occurrence restrictions on identical consonants in
the Hebrew lexicon: are they due to similarity? Journal of Linguistics, 39. 31–
55.
Clements, G. & Hume, E. 1995. The internal organization of speech sounds. In J.
Goldsmith (ed.), Handbook of Phonology. Oxford: Blackwell.
Clements, G. & Keyser, S. 1983. CV Phonology: A Generative Theory of the
Syllable. Cambridge MA: The MIT Press.
Davis, S. 1989. Cross-vowel phonotactic correspondence. Computational Ling-
uistics, 15. 109–111.
Frisch, S., Pierrehumbert, J. & Broe, M. 2004. Similarity avoidance and the OCP.
Natural Language and Linguistic Theory, 22. 179–228.
Fudge, E. 1970. Phonological structure and expressiveness. Journal of Linguistics,
6. 161–188.
Hock, H. & Joseph, B. 1996. Language History, Language Change, and Language
Relationship: An Introduction to Historical and Comparative Linguistics.
Mouton de Gruyter.
Kessler, B. & Treiman, R. 1997. Syllable structure and the distribution of phonemes
in English syllables. Journal of Memory and Language, 37. 295–311.
McCarthy, J. 1986, OCP effects: Gemination and antigemination. Linguistic Inquiry
17. 207–263.
Pierrehumbert, J. 1994. Syllable structure and word structure: a study of tri-
consonantal clusters in English. In P. Keating (ed.), Phonological Structure and
Phonetic Form: Papers in Laboratory Phonology III. Cambridge University
Press.
Stemberger, J. 1981. Morphological haplology. Language, 57. 791–817.
van de Weijer, J. 2003. Consonant variation within words. In D. Archer, P. Rayson,
A. Wilson & T. McEnery (eds.), Proceedings of the Corpus Linguistics 2003
Conference. University Centre for Computer Corpus Research on Language
Technical Papers, 16. 184–190. Lancaster University, United Kingdom.
van de Weijer, J. 2005. Listeners’ sensitivity to consonant variation within words.
Lund Working Papers, 51. 225–238.

201
Language Acquisition and Use in Multilingual Contexts

Appendix
The first number in a cell indicates expected frequencies of words with two identical
consonants; the second number indicates the observed. Cells for which the
difference is significant have been shaded.
WOWF WOSO SOWF SFSO SFWF WOSF WOSFSO WOSFWF WOSOWF SFSOWF WOSFSOWF total

d 6.09–4 3.72–0 0.90–0 0.17–0 0.00–0 0.00–0 3.99–2 0.00–0 12.04–0 1.29–0 9.82–2 38.02–8.00

f 1.83–0 1.37–1 0.04–0 0.03–0 0.00–0 0.00–0 1.66–0 0.00–0 2.52–3 0.12–0 2.49–3 10.06–7.00

g 3.40–3 1.66–3 0.29–0 0.08–0 0.00–0 0.00–0 2.06–0 0.00–0 5.07–4 0.53–0 4.66–2 17.75–12.00

h 0.00–0 0.18–0 0.00–0 0.00–0 0.00–0 0.00–0 0.10–0 0.00–0 0.18–0 0.00–0 0.11–0 0.57–0.00

j 1.68–1 1.64–1 0.16–0 0.06–0 0.00–0 0.00–0 1.71–1 0.00–0 3.40–0 0.29–0 2.98–1 11.92–4.00

k 22.90–30 11.49–11 1.78–0 0.55–0 0.00–0 0.00–0 14.11–1 0.00–0 31.87–20 3.32–0 29.08–12 115.10–74.00

l 27.99–0 11.67–0 2.24–0 0.86–0 0.00–0 0.00–0 18.28–1 0.00–0 36.95–9 4.94–0 38.49–4 141.42–14.00

m 7.82–3 3.71–2 0.47–1 0.39–0 0.00–0 0.00–0 8.60–4 0.00–0 10.22–5 1.78–2 15.14–11 48.13–28.00

n 11.99–4 3.16–3 1.67–2 0.94–0 0.00–0 0.00–0 12.58–2 0.00–0 19.13–11 5.93–3 33.39–16 88.79–41.00

p 5.59–7 4.19–6 0.29–0 0.06–0 0.00–0 0.00–0 3.50–2 0.00–0 8.53–4 0.43–0 6.39–3 28.98–22.00

r 54.73–4 15.90–1 2.63–1 1.35–0 0.00–0 0.00–0 32.91–2 0.00–0 54.68–30 8.09–2 66.10–30 236.39–70.00

s 48.64–36 15.29–4 1.96–2 0.88–0 0.00–0 0.00–0 26.27–9 0.00–0 47.77–17 5.38–2 51.74–38 197.93–108.00

t 33.99–40 15.39–10 3.86–0 0.50–0 0.00–0 0.00–0 14.37–7 0.00–0 52.13–21 4.93–8 38.66–27 163.83–113.00

v 2.73–5 3.00–5 0.16–0 0.03–0 0.00–0 0.00–0 2.15–1 0.00–0 5.21–2 0.20–0 3.59–3 17.07–16.00

ɧ 0.10–0 0.13–0 0.00–0 0.00–0 0.00–0 0.00–0 0.07–0 0.00–0 0.19–0 0.00–0 0.11–0 0.60–0.00

ɕ 0.05–0 0.02–0 0.00–0 0.00–0 0.00–0 0.00–0 0.01–0 0.00–0 0.04–0 0.00–0 0.02–0 0.14–0.00

ŋ 0.00–0 0.00–0 0.11–0 0.06–0 0.00–0 0.00–0 0.45–0 0.00–0 0.69–0 0.58–0 2.55–2 4.44–2.00

ʃ 0.01–0 0.00–0 0.00–0 0.00–0 0.00–0 0.00–0 0.00–0 0.00–0 0.02–0 0.00–0 0.02–1 0.05–1.00

b 1.25–2 3.38–3 0.05–0 0.01–0 0.00–0 0.00–0 2.11–2 0.00–0 4.20–8 0.06–0 2.68–0 13.74–15.00

230.79–139 95.91–50 16.60–6 5.95–0 0.00–0 0.00–0 144.92–34 0.00–0 294.83–134 37.86–17 308.01–155 1134.87–535

202
Svenska med en touch av Aussie

Elisabeth Zetterholm

Inledning
Formell inlärning av ett nytt språk kan ske i den miljö där språket är
majoritetsspråk och talas i det omgivande samhället, vilket då betecknas som
andraspråksinlärning (L2). Om språket däremot företrädesvis lärs ut och
används enbart i klassrummet brukar det kallas för främmandespråkinlärning
(FL). Håkansson & Norrby (2010) menar att intresset för jämförande studier
med fokus på inlärningsmiljön har varit ganska lågt. Flera forskare har heller
inte gjort någon tydlig distinktion mellan L2 och FL, utan företrädesvis
använt begreppet L2. Norrby & Håkansson (2007) har också sammanfört
dessa båda perspektiv i sin studie över vuxna inlärare av svenska i Sverige
och i Australien och fokuserar på omgivningens roll för språkinlärningen.
Resultaten visar att inlärningsordningen i den grammatiska utvecklingen är
lika oavsett var man lär in sitt språk. Däremot påverkar miljön den lexikala
utvecklingen och ordförrådet är inte lika aktivt hos inlärarna i utlandet, dvs. i
en FL-miljö. Det finns också en skillnad mellan de båda inlärargrupperna i
ordassociationstest, där L2-inlärarna ligger närmare talarna i den svenska
kontrollgruppen jämfört med FL-inlärarna. Den pragmatiska utvecklingen
tycks också vara mera gynnsam när man lär sig svenska i Sverige. Collentine
(2004) gjorde en jämförelse mellan studenter som lärde sig spanska i
Spanien respektive i klassrummet i USA. Han fann inga anmärkningsvärda
skillnader i deras grammatiska kunskaper, men att de som lärt sig språket i
en L2-miljö hade lättare för att uttrycka sig på målspråket och föra ett
samtal. I föreliggande studier görs jämförelser mellan några vuxna inlärare
av svenska med fokus på uttalet, såväl de olika svenska fonemen som
prosodiska kontraster. Studenterna läser svenska vid University of Mel-
bourne respektive Lunds universitet.

Bakgrund
Bland andraspråksforskare, särskilt med fokus på uttal och fonologi,
diskuteras ofta huruvida man kan tillägna sig ett mer eller mindre brytnings-
fritt uttal vid inlärning av ett nytt språk (e.g. Abrahamsson & Hyltenstam

203
Language Acquisition and Use in Multilingual Contexts

2004; Ioup 2008; Markham 1997; McAllister 2000; Munro 2008; Pour-
hossein Gilakjani & Resa Ahmadi 2011). Några av dessa menar att det finns
en kritisk ålder, baserad på sociala och kognitiva faktorer, för att låta som en
infödd talare, medan andra ifrågasätter detta och menar att även vuxna
inlärare kan anses få ett brytningsfritt uttal. Inlärarens modersmål och
transfer från modersmålet kan också ha inverkan på uttalet av det nya
språket. Bannert (2004) gjorde under 1980-talet en stor studie som visar
vilka generella och mera specifika svårigheter det kan finnas när man lär sig
svenska som andraspråk. Utifrån inspelningar i ett s.k. brytningsarkiv med
olika modersmålstalare har Bannert kartlagt svårigheter som kan härledas till
talarens modersmål.
Det finns inte alltid ett klart samband med bokstäver och uttal av språk-
ljud. Uttalet hos vuxna inlärare påverkas ofta av hur orden stavas. Bannert
fann i sin studie ett antal uttalsfel som han kopplade till stavningen och
relationen mellan bokstav och språkljud i talarens primärspråk, exempelvis
uttalet av de svenska vokalerna /u/ och /a/ som ofta uttalades som [a]
respektive [æ] av hans informanter med engelska som modersmål. I den
svenska ortografin markeras inte de prosodiska företeelserna ordbetoning
och ordaccent, varför detta kan ställa till problem för inlärare oavsett
modersmål. Vokallängden kan oftast utläsas av enkel- eller dubbelteckning
av efterföljande konsonant.

Svenska med engelsk brytning

I de studier som Bannert beskriver förekommer samma svårigheter obe-
roende om man har amerikansk eller brittisk engelska som modersmål.
Tyvärr hade Bannert inte några talare av australiensisk engelska. Enligt Cox
(2012) är det dock inte någon anmärkningsvärd skillnad mellan varianter i
den australiensiska engelskan jämfört med brittisk och amerikansk engelska,
framför allt inte vad gäller konsonanter. Standard Australian English talas av
de flesta i Australien och anses vara förhållandevis homogen över hela kon-
tinenten (Cox 2012). Det finns några regionala och sociala varianter, som
ännu inte dokumenterats samt Australian Aboriginal English som i första
hand talas av aboriginer samt etnokulturella varianter för att uttrycka olika
etniska identiteter (Cox 2012). I föreliggande studie görs inte någon särskild
bedömning utifrån den engelska varieteten då alla informanter talar Standard
Australian English. Det Bannert tar upp i sin studie är gemensamma uttalsfel
hos informanter med samma modersmål, dvs. det han benämner som ”typisk
brytning”. När det gäller vokaler är det framför allt reduktion vid uttalet som
gör att klangfärgen förändras mot ”vokalmummel”. Exempel på detta är
uttalet av ordet flickorna [flɪkərnə]. Ett annat uttalsfel är att vokalen helt
faller bort, till exempel fågel blir [foːɡl]. Liksom många andra andra-
språksinlärare av svenska har dessa talare också problem med de främre

204
 Language Acquisition and Use in Multilingual Contexts

rundade vokalerna: y- och u-vokalen. Engelskan har kvar aspiration på

tonlös klusil efter betonad vokal, vilket inte är fallet i svenskan. Därför
menar Bannert att inlärare med engelska som modersmål ofta har mera
aspiration i den positionen, vilket en jämförelse mellan engelska letter
[lɛtʰɐ], och svenska lätta [lɛta] (a.a. 2004:41) visar. Det uttalas då mera likt
den svenska retroflexen (rt som i bort) [ʈ] som används i de flesta svenska
dialekter förutom skånska. Tonande s-ljud mellan tonande fonem är också
vanligt hos dessa inlärare. Lateralen /l/ blir ofta velariserad och uppfattas då
ofta som något ”tjockare” än ett svenskt l-ljud. I ett prosodiskt perspektiv har
inlärare med såväl amerikansk som brittisk engelska ofta problem med
distinktionen mellan lång och kort vokal i svenska samt betoningens
placering i såväl enskilda ord som fraser och satser.

Att läsa svenska som främmande språk i Melbourne

Vid ett flertal universitet i världen är det möjligt att läsa svenska som
främmande språk. Ett av dessa var till helt nyligen University of Melbourne
där man kunnat läsa svenska fram till hösten 2012. Det går fortfarande att
läsa svenska på Svenska skolan i Melbourne, men där har eleverna ofta en
annan anknytning till Sverige: ca 2/3 av eleverna har minst en svenskfödd
förälder och den sista tredjedelen utgörs av elever som varit utbytesstudenter
i Sverige under en termin eller ett år och då påbörjat svenskinlärningen där.
De eleverna är mellan 16 och18 år gamla. Studenterna på universitet är några
år äldre.
På Svenska skolan har man undervisning från kindergarten, det vill säga
när barnen är fyra år, upp till år 12, vilket motsvarar studentexamen. På
universitetet i Melbourne kan man läsa svenska i tre år.36 År ett och två
träffas man fyra timmar i veckan, fördelat på två tillfällen, medan man i år
tre endast träffas tre timmar vid ett tillfälle i veckan. Läraren vid univer-
sitetet är svenska och behörig gymnasielärare med flera års erfarenhet av
undervisning i svenska som främmande språk utomlands. Det händer att
lärare och studenter träffas vid olika evenemang med svensk anknytning, till
exempel när Svenska kyrkan har midsommarfirande eller julbasar. Student-
erna har främst valt att börja läsa svenska på grund av sitt intresse för svensk
kultur, till exempel design, film, samekulturen, och de drömmer om att
kunna åka till Sverige och studera.
I Melbourne försöker man träna det svenska uttalet mycket genom bland
annat läsning både enskilt och i kör, man lyssnar på och pratar med varandra
där läraren korrigerar uttalet när så behövs. Läraren har svenska som
modersmål. Man använder också internet som en källa för att lyssna på
36
Beskrivningen av svenskämnet vid universitetet i det följande utgår från situationen fram
tills dess att ämnet lades ner 2012.

205
Language Acquisition and Use in Multilingual Contexts

svenska. Förutom att träna på uttalet av vokaler och konsonanter tar man i
undervisningen också fram den svenska prosodin med vokalkvantitet och
betoningsmönster. Utanför klassrummet är Svenska kyrkan i Melbourne en
central och viktig mötesplats för invånare med skandinavisk anknytning, och
här ges också tillfälle för studenterna att träffas och öva sin svenska i många
olika sammanhang.

Att läsa svenska som främmande språk i Lund

Kurser i svenska som främmande språk ges vid de flesta svenska universitet
för utländska studenter och lärare. Kurserna ges i Sverige och borde kanske
därför betecknas som L2-inlärning utifrån den rådande terminologin, men
universiteten har ändå valt att kalla dessa för kurser i svenska som
främmande språk.
Vid Språk- och Litteraturcentrum på Lunds universitet har man möjlighet
att läsa svenska som främmande språk på flera olika nivåer (1–8) om vardera
7,5 hp och därutöver finns ytterligare kurser för olika målgrupper som inte
har svenska som modersmål. På introduktionskursen, 3 hp, brukar det vara
ca 1000 studenter med varierande modersmål, men betydligt färre på de
högre nivåerna. Under hösten 2012 var drygt 100 studenter registrerade på
nivå 2 och ca 30 personer på kurser som motsvarar nivå 8, som leder fram
till det s.k. Tisus-testet, vilket är ett behörighetsgivande test i svenska för
universitets- och högskolestudier.
Vid samtal med lärare i Lund menar de att man för närvarande inte har
något gemensamt särskilt fokus på uttalet i kurserna, utan det integreras i
undervisningen när man tar upp olika teman. Det är lite upp till varje lärare
hur mycket man fokuserar på uttalet och man anser att varken grammatik
eller uttal kan komma före innehållet i orden. Någon av lärarna menar att
man under några veckor går igenom olika vokaler och då samlar på ord som
innehåller dessa vokaler. De prosodiska dragen med betoningsmönster och
vokalkvantitet, liksom assimilationer och reduktioner tar man upp redan på
den första nivån. Lektionerna är ganska talspråkliga, och sedan får studenten
själv i högre grad arbeta med skriften och grammatikövningar.

Syftet med studien

Håkansson & Norrby (2010) studerade den grammatiska utvecklingen hos
inlärare av svenska i två olika språkliga miljöer, som FL- och L2- inlärare. I
den studie som presenteras här är informanterna också FL- och L2-inlärare
av svenska och har australiensisk engelska som modersmål. Den ena
gruppen läser svenska i Melbourne, den andra i Lund. Fokus ligger framför
allt på informanternas svenska uttal både vid läst och spontant tal, inte den
grammatiska utvecklingen. Beroende på inlärningsmiljön kan det finns

206
 Language Acquisition and Use in Multilingual Contexts

skillnader i uttalet av såväl de svenska fonemen som prosodin. I Melbourne

pratar studenterna i stort sett bara svenska vid undervisningstillfällena. I
Sverige omges de av det svenska språket, men några av studenterna går
kurser där undervisningen bedrivs på engelska. Det är också troligt att de
använder engelska mycket i samvaro med andra studenter eller ute i
samhället eftersom det ofta finns en attityd hos svenskar att de vill praktisera
sina egna kunskaper i engelska så snart tillfälle ges. Hur mycket man
använder det svenska språket i samtal, även utanför själva undervisningen,
spelar en viktig roll vad gäller såväl perceptionen som produktionen, men
också för den sociala interaktionen och kommunikationsmönster (Lindberg
2004). Frågeställningarna i denna studie är: Märks någon större skillnad på
uttalet hos de båda studentgrupperna? Ligger skillnaderna framför allt på
individnivå, oavsett var de studerar svenska?

Material och metod

Fem studenter som läser svenska för tredje och sista året vid University of
Melbourne har spelats in på plats i Melbourne. Fyra informanter från
Australien som läser svenska som främmande språk vid Språk- och
Litteraturcentrum i Lund har spelats in i Lund. Två av dessa läser på nivå 2,
de två andra (Frank och Georg, se presentation av informanter nedan) läser
på nivå 8 (Tisus). Antalet informanter är begränsat eftersom det endast är sju
studenter i gruppen i Melbourne och bara fyra utbytesstudenter från
Australien som läser svenska i Lund innevarande termin. Informanterna har
läst in ca 60 meningar och en kort text som innehåller svenska fonem som
anses svåra att lära in när man lär sig svenska som andraspråk, samt ett
flertal prosodiska minimala par, där vokalkvantitet, ordbetoning och ord-
accenter är betydelseskiljande. Texterna hade de inte sett i förväg, men de
fick gärna läsa igenom texterna tyst innan de läste högt och spelades in. Alla
informanterna läste dock i stort sett helt oförberedda. För att även få
möjlighet att lyssna på deras spontantal fick de beskriva ett tjugotal bilder
som föreställer bland annat djur, frukter och händelser. Under tiden de tittade
på och beskrev bilderna blev det tillfälle till spontana samtal mellan mig och
informanten, vilket också spelades in och används i analysen.
Alla informanter är mellan 21 och 32 år gamla. I Melbourne deltog fyra
kvinnor och en man i studien, i Lund var det en kvinna och tre män. Alla
som spelades in i Melbourne är födda i Australien, i eller i närheten av
Melbourne, och har läst svenska som främmandespråk för samma lärare på
Melbourne University. Ett par av studenterna har släktingar som är svenska,
men bara en av kvinnorna har varit i Sverige under några månader. Tre av
informanterna i Lund är utbytesstudenter, den fjärde är bosatt i Sverige
sedan något år tillbaka.

207
Language Acquisition and Use in Multilingual Contexts

Presentation av informanterna
Jag har avidentifierat mina informanter genom att ge dem nya namn. Nedan
följer en kort presentation av studenterna, med de fem studenterna vid
University of Melbourne först
Amy är född och uppvuxen i Melbourne. Hon har aldrig varit i Sverige,
men ska snart åka dit och studera vid ett svenskt universitet under en termin.
Bruce är född strax utanför Melbourne och bor numera inne i staden. Han
har aldrig varit i Sverige och har inte heller några svenska släktingar.
Cathy är född och uppvuxen i Melbourne. Hon har inte läst lika mycket
som de andra då hon missat någon termin tidigare. Hon har aldrig varit i
Sverige, men släktingar på hennes fars sida kom till Australien för några
generationer sedan.
Deb är född i Sverige, men flyttade till Australien när hon var 9 månader
gammal. Hennes far är svensk och hennes mor är från Australien. Modern
lärde sig lite svenska under de två år hon bodde i Sverige, men använder inte
språket numera. Deb talar numera inte svenska med sin far.
Elaine är född och uppvuxen utanför Melbourne. Hon har varit
utbytesstudent i Sverige under fem månader.
Frank är född och uppvuxen i södra delen av Australien. Han har varit
utbytesstudent i Lund under två terminer och återvänder snart till Australien.
Georg är född och uppvuxen i Queensland i Australien. Han har en
svensk fru och är bosatt i södra Sverige sedan drygt ett år tillbaka.
Hollie är från Sydney, Australien. Hon har varit utbytesstudent under en
termin i Lund och planerar nu att resa tillbaka hem.
John bor i de västra delarna Australien. Han är född i ett skandinaviskt
land men flyttade redan när han var några år gammal till Nya Zeeland och
vidare till Australien. Han har varit utbytesstudent i Lund under en termin
och återvänder nu hem.
Det är uppenbart att några av informanterna, speciellt de som nu läser och
bor i Lund har någon personlig anknytning till Sverige och därför en speciell
anledning att läsa svenska.

Resultat
Det övergripande intrycket är att informanterna befinner sig på väldigt olika
nivåer i uttalet och i sin behärskning av det svenska språket, oavsett om de
läser svenska i Melbourne eller i Lund. Ett annat gemensamt drag är att man
hör en ”typisk” engelsk brytning hos alla talarna, dock mer eller mindre
påtaglig. Exempel på detta är vokalreduktioner i obetonade stavelser,
problem med de främre rundade vokalerna samt tonande s-ljud.
I min redovisning av resultaten tar jag fram gemensamma uttalsavvikelser
(jämfört med standardsvenska) som förekommer hos de flesta av infor-

208
 Language Acquisition and Use in Multilingual Contexts

manterna som ingick i studien. Jag kommer inte att redovisa hur de enskilda
personerna talar, förutom när jag ger några konkreta exempel och sätter det i
relation till bland annat deras vistelsetid i Sverige. Jag kommenterar inte alls
de grammatiska avvikelser som förekommer.

Några uttalsdrag hos inlärare i Melbourne

Spontantal
Vokalreduktioner, framför allt i obetonad stavelse, är vanligt förekommande
i informanternas tal. Exempel på det är ord som båtar [boːʈəɾ], fågel [foːɡl].
De rundade vokalerna uttalas ofta mera orundat än i standardsvenska, t.ex.
stor [stɔːɾ] och flera av talarna har problem med vokallängd, t.ex. sol [sɔl].
Den något ovanliga främre rundade y-vokalen uttalas av flera informanter
som dess orundade motsvarighet /i/ i ord som system och mycket. De svenska
å- och ä-vokalerna förändras till en bakre a-vokal, t.ex. gräset [ɡrɑːsət], blå
[blɑː] av informanterna vid samtliga förekomster. Alla informanter använder
också ett ”tjockare” t-ljud i alla positioner där /t/ förekommer, vilket mer
liknar uttalet av retroflexen [ʈ]. Några har problem med det svenska ng-ljudet
och uttalar ord som pengar med både ng- och g-ljud. De svenska sje-ljuden
klarar alla informanter bra. Många av de här fynden stämmer väl överens
med det Bannert (2004) beskriver som en typisk brytning på svenska när
man har engelska som modersmål.

Läst text
I meningarna och texten finns ett antal prosodiska fallgropar med minimala
par beroende på betoning och vokalkvantitet. Samtliga informanter har
problem med de flesta av dessa minimala par där granen/grannen uttalas på
samma sätt [ɡɾanən] och betoningen ligger på samma stavelse i kalas/kallas
[ˈkalas]. Många sammansatta ord betonas fel. Här ges några exempel där den
(fel)betonade vokalen markeras med versal: påmInner, fördEl, fÖrdelar
(fördela uppgifter), demokrAti, avlYssna, telefonnUmmer. När /s/ före-
kommer mellan andra tonande segment, som i ordet presenter, uttalas det
som ett tonande s-ljud [z]. Ord som kyrka uttalas med en affrikata i början av
flera talare [ʧyrka], men här blir y-vokalen korrekt. I meningen Gräset var
mjukt och kittlade henne mellan tårna uttalas ordet kittlade med ett k-ljud
istället för tje-ljud. Det är möjligt att ordet inte är bekant för informanterna.

Några uttalsdrag hos inlärare i Lund

Spontantal
Den svenska långa a-vokalen får ibland ett mera ä-liknande uttal i ord som
svag [svɛːɡ] och fraser [fɹɛːsəɹ] och det är vanligt att den långa a-vokalen
byts mot kvaliteten hos den korta a-vokalen så att ord som samma sak

209
Language Acquisition and Use in Multilingual Contexts

[samasaːk], bakgrund [baːkɡɾɵnd] och mat [maːt] får samma vokalljud.

Vokalreduktioner förekommer i obetonade stavelser som till exempel i segel
[seːɡl]. Uttalet med ett något ”tjockare” t-ljud är tydligt hos alla informanter,
datorn [dɑːʈɔɾn]. Sje-ljud i ord som stjärnor, skidor och körsbär tycks inte
vålla några större problem.

Läst text
Det händer att vokalerna får fel duration i ord som duk [dɵk] som uttalas
med kort vokal och turister [tʉːrıstər] uttalas med långt u-ljud. Vokal-
reduktion i obetonade slutstavelser är vanligt och den svenska o-vokalen blir
mera öppen i ord som skogen och stationen. De fyra informanterna i Lund
använder ett tonande s-ljud mellan vokaler, inte bara i ordet presenter (som
hos talarna i Melbourne, se ovan) utan även i ord som rosor och fraser.
Endast ett par av informanterna har problem med det svenska ng-ljudet så
uttalet av många blir [mɔŋɡa]. Ordet orgel uttalas av några med ett g-ljud
istället för j-ljud. Det är möjligt att ordet är obekant och att man därför läser
som det stavas. Tje-ljudet i kör uttalas med en affrikata i början, [ʧœːr].
Liksom inlärarna i Melbourne har dessa informanter problem med de
prosodiska minimala paren i texten så att till exempel vägen/väggen uttalas
på samma sätt med kort ä-vokal. Den främre rundade y-vokalen byts ut mot
en i-vokal i mycket, fyra och flygplan och ibland blir det mera som en u-
vokal, flyga [flʉːɡa]. Flera ord i texten får fel betoning, salUhall, ledIga,
fAbrik, tOmater, fÖrteckning.

Samtal på svenska
Förutom de texter som informanterna läste och de bilder som de beskrev
ville jag ha ett kort personligt spontant samtal med var och en i syfte att höra
hur deras svenska språk med fokus på uttalet fungerade i en sådan situation.
Innehållet i de här samtalen varierade beroende på vad de själva ville berätta
för mig. Här blev det en tydlig skillnad där samtliga informanter i Lund
kunde föra en dialog helt på svenska, men med en engelsk brytning. Två av
studenterna i Melbourne, Amy som nu reser till Sverige för att studera och
Elaine som varit i Sverige nästan ett halvår, klarade också dialogen med mig
bra. För de övriga var det lite svårare att använda enbart svenska. De gjorde
omskrivningar när de inte kunde uttala orden och då jag uppfattade att de
själva tyckte det blev för besvärligt övergick de till att tala engelska med
mig. Här måste dock betonas att det var stor variation mellan informanterna.
Denna iakttagelse stämmer väl överens med studien av Collentine (2004).
De som vistats i Sverige en längre tid har lättare för att uttrycka sig på
svenska i en dialog och behärskar uttalet och prosodin så att en svensk
lyssnare förstår. Med största sannolikhet beror det på att de har haft större
möjligheter att använda språket och att de vistats i en miljö där de hört

210
 Language Acquisition and Use in Multilingual Contexts

svenska talas betydligt oftare än de som läst svenska i Melbourne. När det
gäller Amy skulle det kunna vara hennes motivation och längtan efter att få
börja studera i Sverige som gör att hon klarar konversationen så bra. Det är
dock inte alla informanter i Lund som behärskar det svenska språket lika bra
och en anledning till det kan vara att deras universitetskurser är på engelska
samt att de inte behöver använda svenska språket för att leva i Lund. De
flesta där har goda kunskaper i engelska och många vill använda sina
kunskaper när de får möjlighet till det.

Avslutande reflektioner
När man samtalar på sitt andraspråk kan man välja en strategi att före-
trädesvis använda de ord som är enkla att uttala, dvs man kan kringgå
svårare ord med hjälp av en omskrivning. Det var tydligt att några gjorde det
i det spontana samtal jag hade med var och en. När de fick i uppgift att
beskriva några bilder var det ibland svårare att använda sig av den strategin
då det alltid fanns ett motiv och ett målord, där uttalet av olika fonem eller
prosodiska egenheter stod i fokus. Det resulterade i mycket kodväxling
och/eller kortfattade beskrivningar av bilderna samt viss tveksamhet. Kod-
växlingen var naturligtvis också en hjälp när orden saknades i deras svenska
ordförråd och de upprepade alltid mitt uttal när jag sa det svenska ordet, med
olika resultat. Då det inte handlade om att förvandla det här momentet till
glosförhör på svenska blev det istället ett sätt att prata om motivet på
bilderna. I de här samtalen blev det tydligt att de som varit i Sverige hade
lättare att uttrycka sig på svenska. Detta visar att det kan ha betydelse att
vistas i en miljö där språket talas och man får möjlighet att interagera och
träna sitt uttal med modersmålstalare. Undantaget var Amy där biljetten till
ett svenskt universitet förmodligen var den största drivkraften.
När man läser en text som innehåller ord man inte förstår är det vanligt att
det finns ett samband mellan uttalet och stavningen. I svenskan framgår inte
de prosodiska dragen och fonem som uttalas olika beroende på kontexten
kan vara svåra att uttala på rätt sätt. Flera av de sammansatta orden får fel
betoning, vilket kan bero på att man inte känner till de betoningsregler som
finns för sammansatta ord på svenskan eller att man inte är bekant med
ordet. En intressant iakttagelse här är att även de studenter som har ett bra
uttal och kan föra en dialog på svenska ofta får problem med de prosodiska
minimala paren, framför allt när det gäller betoningen.
Resultaten i analyserna av de här inspelningarna stämmer överens med
det som Bannert (2004) beskrivit som ”typisk brytning” när man har
engelska som modersmål, vilket också innebär att fynden i analyserna
motsvarar de förväntningar man kan ha på såväl FL- som L2-inlärare,
oavsett inlärningsmiljö. Om svenska modersmålstalare fått göra en
bedömning av inspelningarna är det mycket troligt att alla svarat att talarna

211
Language Acquisition and Use in Multilingual Contexts

har engelska som modersmål då flera av de typiska brytningsdragen, såväl

reduktion av vokaler som de främre rundade vokalerna och de något tjockare
t-ljuden, finns representerade bland de här talarna. Det är dock inte säkert att
lyssnare kunnat avgöra talarnas engelska varietet, men vissa uttalsfel tycks
ju, enligt Bannert, vara gemensamma oberoende av engelsk varietet.

Ett särskilt tack

Tack till mina informanter i Melbourne och i Lund samt till Anna Berghamre,
University of Melbourne.

Litteratur
Abrahamsson, N. & Hyltenstam, K. 2004. Mognadsbegränsningar och den kritiska
perioden för andraspråkinlärning. I K. Hyltenstam & I. Lindberg (red.), Svenska
som andraspråk – i forskning, undervisning och samhälle. Lund: Student-
litteratur. 221–258.
Bannert, R. 2004. På väg mot svenskt uttal. Lund: Studentlitteratur.
Collentine, J. 2004. The effects of learning contexts on morphosyntactic and lexical
development. Studies in Second Language Acquisition, 26. 227–248.
Cox, F. 2012. Australian English. Pronunciation and Transcription. Cambridge
University Press.
Håkansson, G. & Norrby, C. 2010. Environmental Influence on Language Acq-
uisition: Comparing Second and Foreign Language Acquisition of Swedish.
Language Learning 60 (3). 628–650.
Ioup, G. 2008. Exploring the role of age in the acquisition of a second language
phonology. I G. Jette, E. Hansen & M.L. Zampini (red.), Phonology and Second
Language Acquisition. John Benjamins Publishing Company. 41–62.
Lindberg, I. 2004. Samtal och interaction – ett andraspråksperspektiv. I K. Hylten-
stam & I. Lindberg (red.), Svenska som andraspråk – i forskning, undervisning
och samhälle. Lund: Studentlitteratur. 461–499.
Markham, D. 1997. Phonetic Imitation, Accent, and the Learner. Travaux de
l’Institut de Linguistique de Lund 33. Lunds universitet.
McAllister, R. 2000. Perceptual Foreign Accent and its Relevance for Simultaneous
Interpreting. I B. Englund Dimitrova & K. Hyltenstam (red.), Language Pro-
cessing and Simultaneous Interpreting. Interdisciplinary perspectives. John
Benjamins Publishing Company.
Munro, M. J. 2008. Foreign accent and speech intelligibility. I G. Jette, E. Hansen,
& M.L. Zampini (red.), Phonology and Second Language Acquisition. John
Benjamins Publishing Company. 193–218.
Norrby, C. & Håkansson, G. 2007. Språkinlärning och språkanvändning. Svenska
som andraspråk i och utanför Sverige. Lund: Studentlitteratur.
Pourhossein Gilakjani, A. & Resa Ahmadi, M. 2011. Why is Pronunciation so
Difficult to Learn? English Language Teaching, 4 (3). 74–83.

212
 Language Acquisition and Use in Multilingual Contexts

Contributors37
Malin Ågren
Ph.D. Researcher in French Linguistics. Centre for Languages and Literature, Lund
University. Research areas: second language acquisition, bilingualism, French as a
second language.
Aafke Buyl
Ph.D. fellow of the Research Foundation Flanders (FWO). Centre for Linguistics,
Faculty of Arts, Languages and Literature, Free University of Brussels (VUB). Res-
earch areas: second language acquisition; grammar acquisition; English as a second
language.
Anna Flyman Mattsson
Senior Lecturer in Swedish as a second language. Centre for Languages and Litera-
ture, Lund University. Research areas: second language acquisition, multilingualism,
Swedish as a second language.
Jonas Granfeldt
Professor of French Linguistics. Centre for Languages and Literature, Lund Uni-
versity. Research areas: language acquisition, bilingualism, language didactics, the
acquisition and teaching of French language.
Marianne Gullberg
Professor of Psycholinguistics. Centre for Languages and Literature and Lund Uni-
versity Humanities Lab, Lund University. Research areas: adult second language
acquisition and use, L2/bilingual processing, gesture production and comprehension
in acquisition.
Arthur Holmer
Associate Professor of General Linguistics. Department of Linguistics and Pho-
netics, Centre for Languages and Literature, Lund University. Research areas: erga-
tivity, word class typology, Austronesian, Basque.
Merle Horne
Professor of General Linguistics. Department of Linguistics and Phonetics, Centre
for Languages and Literature, Lund University. Research areas: prosody, spoken
language processing, neurolinguistics.
Alex Housen
Professor of English Linguistics and Applied Linguistics. Department of Linguistics
and Literary Studies, University of Brussels (VUB). Research areas: second lan-
guage acquisition, multilingualism, multilingual education.
Kenneth Hyltenstam
Professor of Bilingualism. Centre for Research on Bilingualism, Department of
Swedish Language and Multilingualism, Stockholm University. Research areas: se-
cond language acquisition, bilingualism, language policy and education.

37
Lars-Åke Henningsson was in the process of writing an article for the festschrift but sadly
passed away before he was able to finish it.

213
Language Acquisition and Use in Multilingual Contexts

Victoria Johansson
Ph.D. Researcher in Linguistics. Department of Linguistics and Phonetics, Centre
for Languages and Literature, Lund University. Research areas: writing develop-
ment, cognitive processes of writing, writing and speaking.
Marie Källkvist
Associate Professor of English Linguistics. Centre for Languages and Literature,
Lund University. Research areas: EFL education, higher education pedagogy, se-
cond language acquisition, language policy and planning in education.
Satomi Kawaguchi
Senior Lecturer of Japanese and Second Language Acquisition. School of Human-
ities & Communication Arts, University of Western Sydney. Research areas: second
language acquisition, Japanese and English as a second language.
Kristin Kersten
Junior Professor of Foreign Language Teaching and Second Language Acquisition.
English Department, University of Hildesheim. Research areas: second language
acquisition, immersion/CLIL, early foreign language teaching.
Jörg-U. Keßler
Professor of English and Applied Linguistics. English Department in the Institute of
Modern Languages, Ludwigsburg University of Education. Research Areas: second
language acquisition, immersion & CLIL, applications of Processability Theory,
ESL/EFL.
Anke Lenzing
Ph.D. Lecturer in English linguistics. Department of English and American Studies,
Paderborn University. Research areas: second language acquisition, psycho-
linguistics.
Inger Lindberg
Professor of Bilingualism with special focus on second language learning. Depart-
ment of Language Education, Stockholm University. Research areas: second
language development in and out of classrooms, school-related vocabulary develop-
ment and socio-political aspects of bilingualism.
Ulrika Nettelbladt
Professor and Certified Speech and Language Pathologist. Department of
Logopedics, Phoniatrics and Audiology, Lund University. Research areas: Language
develoment in children with language impairment and typical language
development.
Catrin Norrby
Professor of Scandinavian Languages. Department of Swedish Language and Multi-
lingualism, Stockholm University. Research areas: sociolinguistics, cross-cultural
pragmatics, interactional linguistics and Swedish as a second language.
Manfred Pienemann
Professor of Linguistics and Applied Linguistics. Department of English and
American Studies, Paderborn University and Visiting Professor, Newcastle Uni-
versity. Research areas: second language acquisition, psycholinguistics and L2
processing, linguistic profiling, cross-linguistic aspects of bilingualism.

214
 Language Acquisition and Use in Multilingual Contexts

Andreas Rohde
Full Professor of Linguistics and Second Language Teaching. English Department
II, University of Cologne. Research areas: second language acquisition, bilingual
kindergarten and preschool programmes.
Eva-Kristina Salameh
Ph.D. Certified Speech Language Pathologist; Senior Lecturer in Speech Language
Pathology. Skåne University Hospital and Department of Logopedics, Phoniatrics,
and Audiology, Clinical Sciences, Lund University. Research areas: language im-
pairment in bilingual children, assessment and intervention.
Anja Steinlen
Senior lecturer. Department of Foreign Language Teaching, University of Erlangen-
Nürnberg. Research areas: first and second language acquisition, multilingualism,
psycholinguistics.
Jan-Olof Svantesson
Professor Emeritus of General linguistics. Department of Linguistics and Phonetics,
Centre for Languages and Literature, Lund University. Research interests: phono-
logy, Asian languages.
Joost van de Weijer
Ph.D. Researcher in Linguistics. Department of Linguistics and Phonetics, Centre
for Languages and Literature, Lund University. Research areas: psycholinguistics,
quantitative linguistics.
Elisabeth Zetterholm
Senior Lecturer of Swedish as a Second Language. Departement of Swedish,
Linnaeus University, Växjö. Research areas: phonetics, pronunciation, Swedish as a
second language.

215
Language Acquisition and Use in Multilingual Contexts

Gisela Håkansson – Bibliography

Håkansson, G. 2013. Att lära sig svenska i Malmö, Jyväskylä eller Melbourne. Om andra-
språk och främmandespråk. In Olofsson, M. (ed.). Lärarrollen i svenska som andraspråk.
Nationellt centrum. Stockholm: Sigma Förlag. 58–70.
Håkansson, G. 2013. Processability Theory. Explaining developmental stages. In Garcia
Mayor, M. de P., Gutierrez Mangado, M. & Martiez Adrian, M. (eds.). Contemporary
Approaches to Language Acquisition. Amsterdam: John Benjamins. 111–128.
Håkansson, G. 2013. Utveckling och variation – en tillämpning av processbarhetsteorin på
svenskt inlärarspråk. To appear in Hyltenstam, K. & Lindberg, I. (eds.). Svenska som
andraspråk. (2nd ed.). Lund: Studentlitteratur.
Håkansson, G. & Westander, J. 2013. Communication in humans and other animals. Amster-
dam: John Benjamins.
Norrby, C. & Håkansson, G. 2013. God svenska! Attityder till språkliga attityder bland gym-
nasister och deras lärare. In Bihl, B., Andersson, P. & Lötmarker, L. (eds.). Svenskans
beskrivning, 32. 208–217.
Håkansson, G. 2011. Emergence criterion The Encyclopedia of Applied Linguistics.
Håkansson, G. 2011. First Language Development of Grammar. The Encyclopedia of Applied
Linguistics.
Håkansson, G. 2011. Implicational scaling. The Encyclopedia of Applied Linguistics.
Flyman Mattsson, A. & Håkansson, G. 2010. Bedömning av svenska som andraspråk. En
analysmodell baserad på utvecklingsstadier. Lund: Studentlitteratur.
Håkansson, G. 2010. Development or impairment? Invited Commentary. Applied Psycho-
linguistics, 31 (2). 293–297.
Håkansson, G. & Norrby, C. 2010. Environmental Influence on Language Acquisition: Com-
paring Second and Foreign Language Acquisition of Swedish. Language Learning, 60.
628–650.
Norrby, C. & Håkansson, G. 2010. Introduktion till Sociolingvistik. Stockholm: Norstedts
Förlag.
Norrby, C. & Håkansson, G. 2010. Svenska här och där – ungdomar diskuterar språkbruk. In
Bo 65. Festskrift till Bo Ralph. Meijerbergs arkiv för svensk forskning, 39. Göteborg.
Steinlen, A., Håkansson, G., Housen, A. & Schelletter C. 2010. Receptive L2 Grammar
Knowledge Development in Bilingual Preschools. In Kersten, K., Rohde, A., Schelletter,
C. & Steinlen, A.K. (eds.). Bilingual Preschools. Vol 1. Learning and development. WVT.
69–100
Steinlen, A.K., Massler, U., Schelletter, C., Thomas, S., Akerman, S., Burmeister, P., Buyl,
A., Ewig, M., Flyman Mattsson, A., Gerlich, L., Håkansson, G., Housen, A., Kalbe, E.,
Kersten, H., Kersten, K., Neils, K., Pahl, S., Piske, T., Ramsey, R., Rohde A., Weitz M. &
Wippermann, I. 2010. Results of the ELIAS Research Studies: A Summary. In Kersten,
K., Rohde, A., Schelletter, C. & Steinlen, A.K. (eds.). Bilingual Preschools. Vol II. Best
Practices. WVT. 31–72.
Håkansson, G. 2009. Wh-questions, rules and constructions in L2 children. In Bernadini, P.,
Egerland, V. & Granfeldt, J. (eds.). Mélanges plurilingues offerts a Suzanne Schlyter à
l’occasion de son 65me anniversaire. Études Romanes de Lund 85. Språk och litteratur-
centrum. Lunds universitet. 117–126.
Granfeldt, J., Håkansson, G., Källqvist, M. & Schlyter, S. 2008. (eds.). Språkinlärning,
Språkdidaktik och Teknologi. Rapport från ASLA:s höstsymposium i Lund 8–9 nov 2007.
Håkansson, G. 2008. Flerspråkiga barn i dagens Sverige. In Noree, M. (ed.). Dyslexi – aktuellt
om läs- och skrivsvårigheter. 7–10.
Håkansson, G. 2008. Hur tänker studenter om grammatik? In Carlsen, C., Moe, E., Andersen,
R.O. & Tenfjord, K. (eds.). Banebryter og brobygger i andraspråksfeltet. Novus Forlag.
Oslo. 56–64.
Rahkonen, M. & Håkansson, G. 2008. Production of written L2-Swedish – processability or
input frequencies? In Kessler, J.-U. Processability Approaches to Second Language

216
 Language Acquisition and Use in Multilingual Contexts

Development and Second Language Learning. Cambridge. Cambridge Scholars Press.

Sigurd, B. & Håkansson, G. 2008. Mathematical (sigmoid) and grammatical modelling of
language learners’ progress. In Granfeldt, J., Håkansson, G., Källqvist, M. & Schlyter, S.
(eds.). Språkinlärning, Språkdidaktik och Teknologi. Rapport från ASLA:s höst-
symposium i Lund 8–9 nov 2007.
Håkansson, G. 2007. Bilingual education– the Swedish experience. Forum on Public Policy.
A Journal of the Oxford Round Table, 3 (2). 151–160.
Håkansson, G. & Hansson, K. 2007.Grammatisk utveckling. In Nettelbladt, U. & Salameh, E-
K. (eds.). Språkutveckling och språkstörning hos barn. Lund: Studentlitteratur. 135–170.
Håkansson, G. & Hansson K. 2007. Grammatiska problem hos barn med språkstörning. In
Nettelbladt, U. & Salameh, E-K. (red) Språkutveckling och språkstörning hos barn. Lund:
Studentlitteratur. 171–199.
Håkansson, G. & Norrby, C. 2007. Processability theory applied to written and oral L2
Swedish. In Mansouri, F. (ed.). Second language acquisition research: theory-
construction and testing. Cambridge: Cambridge Scholars Press. 81–94.
Herder, B., Lindgren, M., Einarsson, J. & Håkansson, G. 2007. Vad händer med läsningen?
En kunskapsöversikt om läsundervisningen i Sverige 1995–2008. Skolverkets rapport 304.
Stockholm.
Mansouri, F. & Håkansson, G. 2007. Conceptualising intra-stage sequencing in the learner
language. In Mansouri, F. (ed.). Second language acquisition research: theory-
construction and testing. Cambridge: Cambridge Scholars Press. 95–125.
Nettelbladt, U., Håkansson, G. & Salameh, E-K. 2007. Specifik intervention av fonologi,
grammatik och lexikon. In Nettelbladt, U. & Salameh, E-K. (eds.). Språkutveckling och
språkstörning hos barn. Lund: Studentlitteratur. 291–311.
Norrby, C. & Håkansson, G. 2007. Girl – lass or curl? Word associations in second language
learners. Australian Review of Applied Linguistics, 30 (2). 22.1–22.17.
Norrby, C. & Håkansson, G. 2007. Språkinlärning och språkanvändning. Svenska som andra-
språk i och utanför Sverige. Lund: Studentlitteratur.
Norrby, C. & Håkansson, G. 2007. The Interaction of Complexity and Grammatical Process-
ability – The Case of Swedish as a Foreign Language. IRAL, 45. 45–68.
Pienemann, M. & Håkansson, G. 2007. Full transfer vs. Developmentally Moderated Trans-
fer: a reply to Bohnacker. Second Language Research, 23. 485–493.
Sigurd, B. & Håkansson, G. 2007. Språk, språkinlärning och språkforskning. Lund: Student-
litteratur.
Håkansson, G. 2005. Similarities and differences in L1 and L2 development. Opening up the
perspective: including SLI. In Pienemann, M. (ed.). Cross-linguistic aspects of Processa-
bility Theory. Amsterdam: John Benjamins. 179–197.
Håkansson, G. 2005. Tvåspråkighet. Grundskoletidningen, 3. 8–14. Fortbildningsförlaget.
Håkansson, G. & Norrby, C. 2005. Flicka – dricka eller tjej? Ordassociationer hos andra-
språksinlärare. In Byrman, G., Einarsson, J., Hammarbäck, S., Lindgren, M. & Stille, P.
(eds.). Svenskans beskrivning, 27. Växjö University Press. 125–134.
Håkansson, G. & Norrby, C. 2005. Grammar and pragmatics – Swedish as a foreign language.
In Foster-Cohen, S., Garcia-Mayo, M. & Cenoz, J. (eds). EUROSLA Yearbook, 5. Amster-
dam: John Benjamins. 137–161.
Pienemann, M., Di Biase, B., Kawaguchi, S. & Håkansson, G. 2005. Processability, typo-
logical constraints and L1 transfer. In Pienemann, M. (ed.). Cross-linguistic aspects of
Processability Theory. Amsterdam: John Benjamins. 86–116.
Pienemann, M., Di Biase, B., Kawaguchi, S. & Håkansson, G. 2005. Processing constraints
on L1 transfer. In J.F. Kroll & A.M.B. DeGroot (eds.). Handbook of Bilingualism:
Psycholinguistic Approaches. New York: Oxford University Press. 128–153.
Sigurd, B., Håkansson, G. & Dahl, J. 2005. Segram – a program visualizing stages of gramm-
atical competence. Working Papers, 51. Lund University, Dept. of Linguistics.
Ekberg, L. & Håkansson, G. 2004. (eds.). Nordand 6. Sjätte konferensen om Nordens språk
som andraspråk. Lund University.
Håkansson, G. 2004. Ett lingvistiskt perspektiv på det liturgiska språket. In Glamsjö, H. (ed.).

217
Language Acquisition and Use in Multilingual Contexts

Han har gjort oss och vi är hans. Om kallet att leva i Svenska kyrkan. Arbetsgemenskapen
Kyrklig förnyelses årsbok, 2003. 68–89.
Håkansson, G. 2004. Utveckling och variation i svenska som andraspråk, enligt process-
barhetsteorin. In Hyltenstam, K. & Lindberg, I. (eds.). Svenska som andraspråk – i forsk-
ning, undervisning och samhälle. Lund: Studentlitteratur. 153–169.
Salameh, E-K., Håkansson, G. & Nettelbladt, U. 2004. Developmental perspectives on biling-
ual Swedish-Arabic children with and without language impairment: a longitudinal study.
International Journal of Language & Communication Disorders, 39. 65–91. Taylor &
Francis Group Ltd.
Håkansson, G. 2003. Tvåspråkighet hos barn i Sverige. Lund: Studentlitteratur.
Håkansson, G. & Norrby, C. 2003. Vad kostar dem?: de, dem eller dom. Språkvård. 11–16.
Språkrådet.
Håkansson, G., Salemeh, E-K. & Nettelbladt, U. 2003. Measuring language development in
bilingual children: Swedish-Arabic children with and without language impairment.
Linguistics 41. 255–288.
Josefsson, G., Platzack, C. & Håkansson, G. 2003. (eds.). The acquisition of Swedish gram-
mar. Language acquisition & language disorders, 33. John Benjamins.
Norrby, C. & Håkansson, G. 2003. Kan jag hjälpa dig med något? Om tilltal i en service-
situation. Språk och stil, 13. 1–30. Institutionen för nordiska språk, Uppsala universitet.
Håkansson, G. 2002. Learning and teaching of Swedish: A Processability Perspective. In Di
Biase, B. (ed.). Developing a second language. Acquisition, processing and pedagogy of
Arabic, Chinese, English, Italian, Japanese, Swedish. Language Australia Publications.
Håkansson, G., Pienemann, M. & Sayehli, S. 2002. Transfer and typological proximity in the
context of L2 processing. Second Language Research, 18. 250–273.
Salameh, E.-K., Nettelbladt, U., Håkansson, G. & Gullberg, B. 2002. Language impairment in
Swedish bilingual children: a comparison between bilingual and monolingual children in
Malmö. Acta Pediatrica, 91. 229–234.
Glahn, E., Håkansson, G., Hammarberg, B., Holmen, A., Hvenekilde, A, & Lund, K. 2001.
Processability in Scandinavian Second language acquisition. Studies in Second Language
Acquisition, 23. 389–416.
Håkansson, G. 2001. Against Full Transfer – evidence from Swedish learners of German.
Working Papers. Lund University, Dept. of Linguistics, 48. 67–86.
Håkansson, G. 2001. Tense morphology and verb-second in Swedish L1 children, L2 children
and children with SLI. Bilingualism: Language and Cognition, 4. 85–99.
Håkansson, G. 2001. Undervisning eller inte undervisning – gör det någon skillnad? In Nau-
cler, K. (ed.). Symposium 2000. Ett andraspråksperspektiv på lärande. Stockholm: Sigma
Förlag. 43–62.
Lastow, B. & Håkansson, G. 2001. Gramte – grammatikträning på dator. In Melander Mart-
tala, U. (ed.). ASLA-information, 27 (2). Universitetstryckeriet, Uppsala.
Håkansson, G. 2000. Svenska som förstaspråk eller som andraspråk. Likheter och skillnader.
In Åhl, H. (ed.). Svenskan i tiden – verklighet och visioner. Stockholm: HLS Förlag.
Håkansson, G. & Hansson, K. 2000. Comprehension and production of relative clauses: a
comparison between Swedish impaired and unimpaired children. Journal of Child
Language, 27. 313–333.
Håkansson, G. & Viberg, Å. 2000. (eds.). Special Issue: A selection of papers from the Ninth
Annual Conference of the European Second Language Association. Studia Linguistica, 2.
Hansson, K., Håkansson, G. & Nettelbladt, U. 2000. Carpe corpus – Om att bygga upp ett tal-
språkskorpus. In Byrman, G., Lindquist, H. & Levin, M. (eds.). Korpusar i forskning och
undervisning. Rapport från ASLA:s höstsymposium 13, Växjö nov 1999. 109–199.
Josefsson, G. & Håkansson, G. 2000. The PP-CP Parallelism Hypothesis and Language
Acquisition: Evidence from Swedish. In Hamann, C. & Powers, S. (eds.). The Acquisition
of Scrambling and Cliticization. Kluwer Academic Publishing. 397–422.
Dooley Collberg, S. & Håkansson, G. 1999. Prohibition: negative imperatives and the para-
metric typology of negation. Working Papers. Lund University, Dept. of Linguistics, 47.
25–37.

218
 Language Acquisition and Use in Multilingual Contexts

Håkansson, G. 1999. Lönar de sig att lära ut grammatiska regler?. På lätt svenska. LärarPM.
Hammarberg, B., Håkansson, G. & Martin, M. 1999. Cognitive and functional aspects of
second language development. In Pietilä, P. & Salo, O-P. (eds.). Multiple Languages –
Multiple Perspectives. AFinLA yearbook 1999. 55–82.
Pienemann, M. & Håkansson, G. 1999. A unified approach towards the development of
Swedish as L2: a processability account. Studies in Second Language Acquisition, 21.
383–420.
Håkansson, G. 1998. Language impairment and the realization of finiteness. In Greenhill, A.,
Hughes, M., Littlefield, H. & Walsh, H. (eds.). Proceedings of the 22nd Annual Boston
University Conference on Language Development, 1. Cascadilla Press. 314–324.
Håkansson, G. 1998. Modern Times in L2 Swedish. Syntax and morphology in formal and in-
formal acquisition of Swedish. In Diaz, L. & Perez, C. (eds.). Views on the acquisition
and use of a second language. EuroSLA 7 Proceedings. Barcelona: Universitat Pompeu
Fabra. 39–50.
Håkansson, G. 1998. Språkinlärning hos barn. Lund: Studentlitteratur.
Håkansson, G. 1997. Barnets väg till svensk syntax. In Söderbergh, R. (ed.). Från joller till
läsning och skrivning. Malmö: Gleerups. 47–60.
Håkansson, G. 1997. Ett ordföljdsproblem. Sprogforum. Tidskrift for sprog- og kulturpaeda-
gogik, 7. 24–30.
Håkansson, G. 1997. Language impairment from a processing perspective. Working Papers.
Lund University, Dept. of Linguistics, 46. 133–151.
Håkansson, G. & Lastow, B. 1997. GRAMTE. Datorprogram för träning av grammatisk
terminologi. Lund University, Dept. of Linguistics.
Håkansson, G., Santesson, L., Svensson, J. & Viberg, Å. 1997. (eds.). Svenskans beskrivning,
22. Lund: Studentlitteratur.
Lastow, B. & Håkansson, G. 1997. Grammatical terminology and the application Gramte.
Working Papers. Lund University, Dept. of Linguistics, 46.
Håkansson, G. 1996. Samtal i klassrummet. In Blomqvist, J. & Teleman, U. (eds.). Det gräns-
lösa samtalet. Lund University Press.
Håkansson, G. & Nettelbladt, U. 1996. Similarities between SLI and L2 children. Evidence
from the acquisition of Swedish word order. In Johnson, C.E. & Gilbert, J.H.V. (eds.).
Children's Language, 9. Lawrence Erlbaum & Associates. 135–151.
Håkansson, G. & Nettelbladt, U. 1996. (eds.). Språkförståelse. Rapport från ASLA:s höst-
symposium, Lund, 9–11 nov 1995.
Salameh, E-K., Håkansson, G. & Nettelbladt, U. 1996. The acquisition of Swedish as a
second language in a group of Arabic-speaking pre-school children: word order patterns
and phrasal morphology. Logopedics Phoniatrics Vocology, 21. 163–170.
Bruzaeus, L. & Håkansson, G. 1995. I princip: grammatikövningar med regler och
kommentarer. D.1. Scriptor.
Bruzaeus, L. & Håkansson, G. 1995. I princip: grammatikövningar med regler och
kommentarer. D.2. Scriptor.
Håkansson, G. 1995. Rapid Profile – en snabbdiagnos av grammatisk nivå i inlärarspråk. In
Linnarud, M. (ed.). Rapport från ASLA-symposiet 10–12 nov 1994.
Håkansson, G. 1995. Språkutveckling på avvikande villkor. In Söderbergh, R. (ed.). Barn-
språk vid Lunds Universitet: Föreläsningar av tolv forskare från fyra institutioner, 10.
Child Language Research Institute.
Håkansson, G. 1995. Syntax and morphology in language attrition. A study of five bilingual,
expatriate Swedes. International Journal of Applied Linguistics, 5 (2). 153–171.
Håkansson, G. 1995. Yla som vargar, tala som folk: om djurens och människornas
kommunikation. Almqvist & Wiksell.
Håkansson, G. 1994. Andraspråksundervisning – några reflektioner. Lisetten: förenings-
tidning för Riksförbundet Lärare i svenska som andraspråk, 3. 8–16.
Håkansson, G. 1994. Verb-initial sentences in the development of Swedish. Working Papers.
Lund University, Dept. of Linguistics, 42. 49–65.
Håkansson, G. & Dooley Collberg, S. 1994. The Preference for Modal + Neg. An L2 Perspec-

219
Language Acquisition and Use in Multilingual Contexts

tive Applied to L1 acquisition. Second Language Research 10 (2). 95–124.

Schlyter, S. & Håkansson, G. 1994. Word order in Swedish as the first language, second lang-
uage and weaker language in bilinguals. Scandinavian Working Papers in Bilingualism, 9.
49–66.
Håkansson, G. 1993. Språklärarspråket – en anpassningsfråga. In Cerú, E. (ed.). Svenska som
andraspråk. Mera om undervisningen. Lärarbok 3. Natur och Kultur.
Håkansson, G. 1993. Vilken roll spelar läraren? Sesam på lätt svenska. Sesam lärar-pm. 39–
40.
Håkansson, G. & Nettelbladt, U. 1993. Developmental sequences in L1 (normal and
impaired) and L2 acquisition of Swedish. International Journal of Applied Linguistics, 3.
131–157.
Håkansson, G. 1992. The role of the teacher in the second language classroom. Revue de Pho-
netique Appliquee, 103–104. 163–180.
Håkansson, G., Nettelbladt, U. & Hansson, K. 1991. Variation and Deviation in Language
Acquisition. Some hypotheses and preliminary observations. Working Papers. Lund
University, Dept. of Linguistics, 38. 83–95.
Nettelbladt, U. & Håkansson, G. 1991. Towards an integrated view of language acquisition.
Scandinavian Journal of Logopedics and Phoniatrics, 16. 29–35. Scandinavian University
Press.
Nettelbladt, U. & Håkansson, G. 1990. (eds.). Samtal och språkundervisning. Studier till
Lennart Gustavssons minne. Linköping Arts in Arts and Science, 60.
Håkansson, G. 1989. Bodily behaviour in emotive expressions. Working Papers. Lund Uni-
versity, Dept. of Linguistics, 35. 89–98.
Håkansson, G. 1989. The acquisition of negative placement in Swedish. Studia Linguistica,
43 (1). 47–58.
Håkansson, G. 1988. Hungry I Am – Breakfast I Want. On the Acquisition of Inverted Word
Order in Swedish. Working Papers. Lund University, Dept. of Linguistics, 33. 123–130.
Håkansson, G. & Lindberg, I. 1988. What's the question? Investigating questions in second
language classrooms. In Kasper, G. (ed.). Classroom research. AILA Review, 5.
Håkansson, G. 1987. Teacher Talk. How teachers modify their speech when addressing learn-
ers of Swedish as a second language. Lund University Press.
Håkansson, G. 1986. Quantitative aspects of Teacher Talk. In Kasper, G. (ed.). Learning, tea-
ching and communication in the foreign language classroom. Aarhus University Press.
Håkansson, G. 1982. Some quantitative aspects of Teacher Talk. Scandinavian Working Pap-
ers on Bilingualism, 1. Stockholm University, Centre for Research on Bilingualism.

220
 Language Acquisition and Use in Multilingual Contexts

TRAVAUX DE L’INSTITUT DE LINGUISTIQUE DE LUND

FONDÉS PAR BERTIL MALMBERG

1. Carl-Gustaf Söderberg. A Typological Study on the Phonetic Structure of English

Words with an Instrumental-Phonetic Excursus on English Stress. 1959.
2. Peter S. Green. Consonant-Vowel Transitions. A Spectrographic Study. 1959.
3. Kerstin Hadding-Koch. Acoustico-Phonetic Studies in the Intonation of Southern
Swedish. 1961.
4. Börje Segerbäck. La réalisation d’une opposition de tonèmes dans des dissyllabes
chuchotés. Étude de phonétique expérimentale. 1966.
5. Velta Ruke-Dravina. Mehrsprachigkeit im Vorschulalter. 1967.
6. Eva Gårding. Internal Juncture in Swedish. 1967.
7. Folke Strenger. Les voyelles nasales françaises. 1969.
8. Edward Carney. Hiss Transitions and their Perception. 1970.
9. Faith Ann Johansson. Immigrant Swedish Phonology. 1973.
10. Robert Bannert. Mittelbairische Phonologie auf akustischer und perzeptorischer
Grundlage. 1976.
11. Eva Gårding. The Scandinavian Word Accents. 1977.
12. Gösta Bruce. Swedish Word Accents in Sentence Perspective. 1977.
13. Eva Gårding, Gösta Bruce, Robert Bannert (eds.). Nordic Prosody. 1978.
14. Ewa Söderpalm. Speech Errors in Normal and Pathological Speech. 1979.
15. Kerstin Nauclér. Perspectives on Misspellings. 1980.
16. Per Lindblad. Svenskans sje- och tjeljud (Some Swedish sibilants). 1980.
17. Eva Magnusson. The Phonology of Language Disordered Children. 1983.
18. Jan-Olof Svantesson. Kammu Phonology and Morphology. 1983.
19. Ulrika Nettelbladt. Developmental Studies of Dysphonology in Children. 1983.
20. Gisela Håkansson. Teacher Talk. How Teachers Modify their Speech when Addressing
Learners of Swedish as a Second Language. 1987.
21. Paul Touati. Structures prosodiques du suédois et du français. Profils temporels et
configurations tonales. 1987.
22. Antonis Botinis. Stress and Prosodic Structure in Greek. A Phonological, Acoustic,
Physiological and Perceptual Study. 1989.
23. Karina Vamling. Complementation in Georgian. 1989.
24. David House. Tonal Perception in Speech. 1990.
25. Emilio Rivano Fischer. Topology and Dynamics of Interactions - with Special
Reference to Spanish and Mapudungu. 1991.
26. Magnus Olsson. Hungarian Phonology and Morphology. 1992.
27. Yasuko Nagano-Madsen. Mora and Prosodic Coordination. A Phonetic Study of
Japanese, Eskimo and Yoruba. 1992.
28. Barbara Gawronska. An MT Oriented Model of Aspect and Article Semantics. 1993.
29. Bengt Sigurd (ed.). Computerized Grammars for Analysis and Machine Translation.
1994.
30. Arthur Holmer. A Parametric Grammar of Seediq. 1996.
31. Ingmarie Mellenius. The Acquisition of Nominal Compounding in Swedish. 1997.
32. Christina Thornell. The Sango Language and Its Lexicon (Sêndâ-yângâ tî Sängö).
1997.
33. Duncan Markham. Phonetic Imitation, Accent, and the Learner. 1997.
34. Christer Johansson. A View from Language. Growth of Language in Individuals
and Populations. 1997.
35. Marianne Gullberg. Gesture as a Communication Strategy in Second Langauge
Discourse. A Study of Learners of French and Swedish. 1998.
36. Mechtild Tronnier. Nasals and Nasalisation in Speech Production. With Special
Emphasis on Methodology and Osaka Japanese. 1998.

221
 Language Acquisition and Use in Multilingual Contexts

37. Ann Lindvall. Transitivity in Discourse. A Comparison of Greek, Polish and

Swedish. 1998.
38. Kirsten Haastrup & Åke Viberg (eds.). Perspectives on Lexical Acquisition in a
Second Language. 1998.
39. Arthur Holmer, Jan-Olof Svantesson & Åke Viberg (eds). Proceedings of the 18th
Scandinavian Conference of Linguistics. 2001.
40. Caroline Willners. Antonyms in Context. A corpus-based semantic analysis of
Swedish descriptive adjectives. 2001.
41. Hong Gao. The Physical Foundation of the Patterning of Physical Action Verbs. A
Study of Chinese Verbs. 2001.
42. Anna Flyman Mattsson. Teaching, Learning, and Student Output. A study of
French in the classroom. 2003.
43. Petra Hansson. Prosodic Phrasing in Spontaneous Swedish. 2003.
44. Elisabeth Zetterholm. Voice Imitation. A Phonetic Study of Perceptual Illusions
and Acoustic Success. 2003.
45. Johan Frid. Lexical and Acoustic Modelling of Swedish Prosody. 2003.
46. Anastasia Mukhanova Karlsson. Rhythm and Intonation in Halh Mongolian. 2005.
47. Susanne Schötz. Perception, Analysis and Synthesis of Speaker Age. 2006.
48. Victoria Johansson. Developmental Aspects of Text Production in Writing and
Speech. 2009.
49. Gilbert Ambrazaitis. Nuclear Intonation in Swedish. Evidence from Experimental-
Phonetic Studies and a Comparison with German. 2009.
50. Mats Andrén. Children’s Gestures from 18 to 30 months. 2010
51. Susan Sayehli. Developmental Perspectives on Transfer in Third Language Acquisition.
2013.
52. Anna Flyman Mattsson & Catrin Norrby (eds). Language Acquisition and Use in
Multilingual Contexts. Theory and Practice. 2013.

222

View publication stats