here - CASTL

Workshop on
Multilingual
Language
Acquisition,
Processing and Use
ABSTRACTS
LIST OF ABSTRACTS
KEYNOTE SPEAKERS
Jennifer Cabrelli Amaro (University of Illinois at Chicago, USA)
Looking both ways: Examining bidirectional transfer in third language
acquisition............................................................................................... 5
Maria del Pilar García Mayo (University of the Basque Country, Spain)
The acquisition of L3 English by Basque-Spanish learners: Looking back
and moving forward ............................................................................... 5
Roumyana Slabakova (University of Southampton, UK / University of
Iowa, USA)
The Scalpel Model of Third Language Acquisition .................................. 6
David Green (University College London, UK)
Trajectories to L3 proficiency ................................................................. 7
Stefanie Wulff (University of Florida, USA)
What corpus linguistics can contribute to multilingualism
research .................................................................................................. 7
ORAL PRESENTATIONS
Form-to-meaning mapping in multilingual speakers: A test of the
predictions of the Revised Hierarchical Model
Carla A. Pastorino Campos (University of Cambridge, UK)....................... 7
Investigating the role of cognitive abilities in L3 processing and use
Laura Sanchez (Stockholm University) ..................................................... 9
Language switching: The role of linguistic relatedness and language status
Michela Mosca, Julia Festman & Harald Clahsen (University of Potsdam,
Germany) ............................................................................................. 10
On pronouns that drop (out of German)
Ylva Falk (Stockholm University, Sweden) ............................................. 11
2
Patterns of cognate facilitation effect in trilingual speakers as an index of
the structure of the trilingual mental lexicon
Agnieszka Lijewska & Hanka Błaszkowska (Adam Mickewicz University,
Poland) ................................................................................................ 12
Reflexive binding in Norwegian L2 and L3
Guro Busterud (NTNU, Norway) ........................................................... 13
Sensitivity to word order errors in L3 processing: Evidence from selfpaced reading and grammaticality judgments
Ana Stutter Garcia & Claudia Felser (Potsdam Research Institute for
Multilingualism, Germany) .................................................................... 16
Syntactic transfer from French and Basque to L3 Spanish: the null-subject
parameter
Hugues Lacroix (Université de Montréal, Canada) ................................. 17
The acquisition of anything by Catalan/Spanish bilingual and Spanish
monolingual learners of English: Evidence for the Typological Primacy
Model of L3 acquisition
Eloi Puig-Mayenco (Universitat Pompeu Fabra) & Heather Marsden (Univ.
of York, UK) .......................................................................................... 18
The role of language dominance in German-Turkish heritage speakers of
L3 English
Anika Lloyd-Smith (University of Hamburg, Germany), Henrik Gyllstad
(Lund University, Sweden) & Tanja Kupisch (University of Konstanz,
Germany) .............................................................................................. 20
Transfer from nowhere? The unknown factor in L3 learning
Christina Lindqvist (Uppsala University) & Ylva Falk (Stockholm
University) ............................................................................................. 21
What accounts for phonetic performance in a third language? A
correlational study
Magdalena Wrembel (Adam Mickiewicz University, Poland) ................ 23
POSTERS
Exploring the Mental Lexicon of the Multilingual: Testing vocabulary
knowledge and cognate recognition in the L2 and L3
3
Csaba Szabo (The Open University) ...................................................... 24
The Acquisition of L3/Ln Morphosyntax in Norwegian Learners of Turkish
Emel Turker -Van der Heiden & Gözde Mercan (University of Oslo,
MultiLing, Norway) ............................................................................... 26
Persistence of non-facilitative transfer in L3 Development
Felipe Amaro, Jennifer Cabrelli Amaro (University of Illinois at Chicago)
and Jason Rothman (University of Reading / University of Tromsø) ........ 27
Norwegian-Russian bilinguals learning English: The Linguistic Proximity
Model
Marit Westergaard (University of Tromsø, Roksolana Mykhaylyk (Harvard
University), Natalia Mitrofanova (University of Tromsø) and Yulia Rodina
(University of Oslo) ............................................................................... 29
The acquisition of Path in L4 German and the effect of previous
languages
Veronique Scheirs (Université Catholique de Louvain,
Belgium) ................................................................................................ 31
4
Looking both ways: Examining bidirectional transfer in third language acquisition
Jennifer Cabrelli Amaro
University of Illinois at Chicago
The recent upturn in formal approaches to third language (L3) acquisition is due in great part
to a growing recognition of the novel ways in which this line of research can inform theories
of multilingualism and language acquisition more generally (see, e.g., Rothman, Iverson, &
Judy, 2011 for an overview). The primary focus thus far has been on the phenomenon of
selective transfer at the L3 initial stages and the mechanisms that drive the transfer of one
system over another when the learner has multiple systems (i.e., the L1 and L2) available (see
García Mayo & Rothman, 2012, for syntax and Wrembel & Cabrelli Amaro, 2014, for
phonology). However, progressive (L1/L2 à L3) transfer is just one of the phenomena that we
can investigate to better understand the dynamic nature of multilingualism. Taking into
consideration substantial evidence of the influence of an L2 on an L1 (see e.g., Schmid,
Köpke, & de Bot, 2013), L3 regressive (L3 à L1/L2) transfer is also predicted to occur.
Examination of the L3 influence on the L1 compared with the L2 can inform longstanding
debates regarding the mental constitution of early-acquired versus late-acquired systems.
However, regressive morphosyntactic influence has not yet been investigated, and regressive
phonological influence has only been investigated in Gut (2010), and with only four
participants.
In this paper, I address L3 transfer as a bidirectional phenomenon via the investigation of L1
English/L2 Spanish and L1 Spanish/L2 English bilinguals acquiring L3 Brazilian Portuguese
(BP). First, I present data in support of L3 initial stages transfer driven by structural similarity
(i.e., evidence that Spanish, rather than English, transfers to L3 BP regardless of age of
acquisition). Second, I compare the stability of L1 and L2 Spanish morphosyntactic and
phonological mental representations at different levels of L3 BP proficiency, providing
evidence that an adult-acquired Spanish system is more vulnerable to L3 influence than an
early-acquired system.
The acquisition of L3 English by Basque-Spanish learners: Looking back and moving
forward
María del Pilar García Mayo
Universidad del País Vasco (UPV/EHU)
Over the past two decades the study of third language (L3) acquisition from a formal
perspective has evolved considerably (García Mayo & Rothman, 2012; Rothman, Iverson &
Judy, 2011). Although a good amount of research has focused on assessing which previous
language determines the initial state of L3 acquisition and has proposed different L3 initial
state models, some studies have considered various morphosyntactic aspects in learners that
are beyond the initial state. This talk will provide an overview of research on the
interlanguage development of BasqueSpanish bilinguals learning L3 English in a school
context. The linguistic topics under investigation include placeholders, subject pronouns and
agreement morphemes, suppletive vs. affixal inflection and third person singular possessives.
More recent research on the syntax-semantics interface and morphological processing of
compounds will also be reviewed. On the methodogical level, cross-sectional and
5
longitudinal data and the use of online and offline tasks will be considered in an effort to
inform research on formal non-native syntax.
The Scalpel Model of Third Language Acquisition
Roumyana Slabakova
University of Southampton/University of Iowa
This presentation will revisit the current theoretical models attempting to explain the
development process of third language acquisition and the factors affecting it. There are
currently three active models on the market, each with some empirical support. The critical
research question is whether the L1 or the L2 structure has an effect when acquiring the
L3/Ln and whether the totality of the impacting grammar is transferred as an L3/Ln initial
stage. The L2 Status model (Bardel & Falk 2007, 2013) proposes that the second language
has a decisive influence on L3 development, due to its recency of acquisition. The
Cumulative Enhancement model (CE, Flynn et al 2004) and the Typological Primacy model
(TP, Rothman, 2010, 2011, in press) both hypothesize that all previously acquired languages
might be a source of transfer.
CE contends that any previously acquired properties are potentially available to the L3/Ln
learner to use in parsing, comprehending and producing language. Crucially, prior language
experiences can either enhance subsequent language acquisition or remain neutral. The TP
model shares the claim that neither the L1 nor the L2 have a privileged status for initial state
L3/Ln morphosyntactic transfer. However, Rothman (in press) has argued that interlanguage
development is constrained by what the internal parser takes to be the (actual or perceived)
structural similarity, among the three grammars. A further claim of this model is that the
entirety of the grammar perceived to be typologically similar is transferred as the L3 initial
state.
The newly proposed Scalpel model shares features with the CE and TP models. Together
with these proposals, it argues that neither the L1 nor the L2 have a dramatically privileged
status in L3/Ln acquisition. However, contrary to CE, I will argue that previous linguistic
experience can not only enhance but also impede L3A. I will further argue that there is no
need for a wholesale transfer of the grammar perceived as similar, even at initial stages.
Instead, the activated grammatical possibilities of the L1-plus-L2 sum act with a scalpel-like
precision, rather than as a blunt object, to extract the enhancing options of L1 or L2
parameter values. Multilingual grammars are sufficiently sophisticated not to need wholesale
transfer. At the same time, other factors such as processing complexity, misleading input and
construction frequency also play an important role in that process, with the effect of
oftentimes thwarting the potential cumulative enhancement. This talk will offer experimental
evidence for these claims, and make testable predictions for further research.
6
Trajectories to L3 proficiency
David W. Green
University College London
Learning and using a third (L3) language increases demands on the network of neural regions
representing the different domains of language. Prior representations undoubtedly constrain
L3 learning and current accounts of the early stages of L3 learning articulate how they may
do so. This talk proposes that only by also considering the processes of language control that
act on these emerging representations will accounts be sufficient to capture the trajectories of
acquisition in real-world interactional contexts.
What corpus linguistics can contribute to multilingualism research
Stefanie Wulff
University of Florida
Over the past two decades, corpus linguists have become increasingly interested in applying
their methodological tool box to the various areas that fall under the umbrella term
multilingualism, including contrastive studies of native and non-native language, learner
language development, and studies into third language acquisition (Granger et al. 2015).
However, both outside of the corpus linguistics community and – somewhat surprisingly –
within the field itself, corpus linguistic research that adopts a decidedly quantitative
perspective is often conceived as too shallow for, or even hindering, meaningful analysis and
interpretation. In this paper, I aim to address this misconception of what corpus linguistics is,
what it has to offer, and to whom. In order to convey my line of reasoning, I will first define
corpus linguistics as a method (rather than a theory) that is inherently quantitative in nature. I
will then give an overview of recent learner corpus research that employs increasingly
sophisticated quantitative methods to demonstrate how these stand to further our
understanding of many of the current hot topics in multilingualism research: appropriate
characterizations of what constitutes the L1 input and/or target norm; the adequate modeling
of the intrinsically complex as well as highly L1-specific nature of learner language
development; and the increasingly recognized role of individual variation in the acquisition
process.
Form-to-meaning mapping in multilingual speakers: A test of the predictions of the
Revised Hierarchical Model
Carla A. Pastorino Campos (University of Cambridge, UK)
The Revised Hierarchical Model (RHM) (Kroll & Stewart, 1994; Kroll, Van Hell, Tokowicz,
& Green, 2010) has served from its inception as a means to theoretically understand bilingual
lexical processing, providing insights regarding the possible mechanisms of form-to-meaning
7
mapping for bilingual speakers. Its core assumptions have been repeatedly tested with
varying results (e.g. Kroll, Michael, Tokowicz, & Dufour, 2002; Sholl, Sankaranarayanan, &
Kroll, 1995; Talamas, Kroll, & Dufour, 1999). However, whereas research focusing on
bilingual speakers abound, these studies have rarely addressed the question of multilingual
lexical processing (MLP).
Considering this, the purpose of this study was twofold: (a) to isolate the RHM’s main
assumptions and, using these premises, to propose hypotheses regarding MLP; and (b) to test
these hypotheses to ascertain if they sustain when trilingual speakers are tested.
Taking into account the propositions of the RHM regarding the nature and strength of lexical
links in different translation directions, it was hypothesized that unbalanced trilingual
speakers (dominant first language [L1], less dominant second language [L2] and least
dominant third language [L3]) would exhibit stronger semantic links between their L2 and L3
and stronger lexical links between the L3 and both the L1 and L2.
These hypotheses were tested using a translation recognition task in which reaction-times
(RT) were recorded. The participants were 37 consecutive Spanish-Italian-English trilinguals.
The task included three translation directions: L3 to L1, L3 to L2, L2 to L1 and L2 to L3.
There were three types of translation pairs: - Correct translations (e.g. cheese / queso); Form/lexical distracters: the prime was orthographically related to the target (e.g. cheese /
quiso [wanted]) - Semantic distracter (e.g. cheese / leche [milk]): the prime was related in
meaning to the target
Both the participants’ mean RTs and errors were compared in order to observe the nature of
the lexical links established in these multilinguals' minds.
Considering the RT data, participants were significantly slower at rejecting incorrect
translations in two cases: (a) in the L3 to L2 direction for form distracters; and (b) in the L3
to L1 direction for semantic distracters. The error data showed that the participants were
susceptible to semantic interference in the L3 to L2, L3 to L1 and L2 to L1 directions and to
form interference in the L3 to L1 direction.
These results suggest that semantic interference was more pervasive than initially expected,
but it was not found in the predicted direction. Furthermore, form mediation, although
present, may not underlie processing in all instances of weaker to stronger language
translation, as hypothesized.
Overall, these findings highlight some of the difficulties of adapting the RHM to multilingual
speakers, showing that, if the model’s assumptions are followed in a strict way, the mere
addition of lexicon to the current bilingual model may not be sufficient to account for MLP.
References
Kroll, J. F., Michael, E., Tokowicz, N., & Dufour, R. (2002). The development of lexical
fluency in a second language. Second Language Research, 18(2), 137–171.
Kroll, J. F., & Stewart, E. (1994). Category interference in translation and picture naming:
Evidence for asymmetric connections between bilingual memory representations. Journal
of Memory and Language, (33), 149–174.
8
Kroll, J. F., Van Hell, J. G., Tokowicz, N., & Green, D. W. (2010). The Revised Hierarchical
Model: A critical review and assessment. Bilingualism: Language and Cognition, 13(3),
373–381. doi:10.1017/S136672891000009X.
Sholl, A., Sankaranarayanan, A., & Kroll, J. (1995). Transfer between Picture Naming and
Translation: A Test of Asymmetries in Bilingual Memory. Psychological Science, 6(1),
45–49.
Talamas, A., Kroll, J. F., & Dufour, R. (1999). From form to meaning: Stages in the
acquisition of second-language vocabulary. Bilingualism: Language and Cognition, 2,
45–58.
Investigating the role of cognitive abilities in L3 processing and use
Laura Sanchez (Stockholm University)
In the last decade research on the effects of cognitive factors on interlanguage performance
has increased substantially. Above all, the effects of these factors are investigated on the
basis of measures such as phonological memory, short-term memory, working memory,
attention control and attention switching (e.g. French and O’Brien, 2008; Gilabert and
Muñoz, 2010; Housen and Kuiken, 2009; Kormos and Safár, 2009; O’Brien, Segalowitz,
Collentine and Freed, 2006). Nonetheless, researchers have felt particularly attracted by the
development or acquisition of various dimensions of language performance, namely
complexity, accuracy and fluency. Less of an interest has been show in the probable effects
of these cognitive factors on crosslinguistic influence (CLI) in written foreign language
performance (but see Odlin, 2009). The study presented here adds to this line of
investigation. It explores the relationship between working memory capacity (WMC),
attentional control (AC) and attention switching (AS) on CLI in third language acquisition.
More specifically, it examines the relationship between these cognitive measures and CLI in
the area of syntax, or to be more precise, in relation to VP headedness in L3 English.
The participants who took part in the study were Spanish/Catalan bilingual school-age
learners of L3 English (n= 94) with prior knowledge of L2 German. To assess WMC, a letter
span task was employed. On the other hand, attention was measured using parts A and B of
the trail making test, for the measurement of AC and AS, respectively. The elicitation
technique used in order to elicit written production from these learners was an immediate
recall task based on Charles Chaplin’s silent film ‘Modern Times’. Learners were asked to
listen twice to a ca. 2-minute narrative in English that explained the first part of Modern
Times. While they listened to the narrative, they were allowed to look at visual stimuli
consisting of 3 photographs extracted from the film. Afterwards, they were asked to write the
narrative heard to somebody who had never heard it before. The learners performed the
immediate recall task in class in presence of the researcher and their teacher. In addition to
this, learners were administered the Oxford Placement Test, a standardized and independent
comprehensive test aimed to control for proficiency in the target language and avoid a
confound effect between proficiency and the cognitive measures. For the administration of
the cognitive tests, learners were taken individually to a quiet room, so as to guarantee and
maximize the necessary silence conditions required to perform the letter span and trail
making tasks.
9
Evidence of CLI related to VP headedness was assembled in the written task, thereby
corroborating previous findings from studies with learners of comparable language
background but gathered by means of a narrative story-telling task (Sánchez, 2011a, b, in
press). The overall occurrences of CLI were correlated with the different cognitive measures
at different proficiency levels, and statistical analyses run on the data yielded low to moderate
negative correlations with all of them. These preliminary findings suggest a relationship
between cognitive factors and the occurrence of CLI, in that learners with lower working
memory and attention capacities transferred more than their higher. This is taken as evidence
that shortages in cognitive abilities may prevent learners from making an effective use of the
linguistic resources at their disposal, and also from inhibiting the activation of foreign
languages previously learned.
Language switching: The role of linguistic relatedness and language status
Michela Mosca, Julia Festman & Harald Clahsen (University of Potsdam, Germany)
One of the most remarkable skills of multilingual individuals is ‘language control’ (e.g.
Abutalebi & Green, 2007), the ability to fluently switch between different languages while
avoiding intrusions from other languages. Measuring language-switch costs represents a
familiar tool to investigate the processes involved in language control. In particular, switch
costs are taken to reflect inhibitory processes to suppress the non-targeted language and have
been argued to be determined by an individual’s proficiency in the respective languages; see
e.g. Green (1986, 1998).
In this project, we assess not only the impact of language proficiency on switch costs but also
that of linguistic relatedness (typologically more vs. less related) and of language status
(native vs. non-native). Our study is the first one that compares two groups of trilinguals on a
language switching task. Furthermore, relative to previous trilingual switch-cost experiments
(e.g. Linck et al. 2012), we used a relatively large set of experimental stimuli, few stimulus
repetitions, and a relatively low total number of trials. Experimental stimuli were also
carefully matched according to word frequency, word length, conceptual complexity,
semantic category and cognate status.
To measure switch costs, we employed a picture-naming task involving the switching
between three languages (English, Italian, German). Two groups of late unbalanced
trilinguals participated, differing with respect to the order in which these three languages
were acquired:
- Group 1: N=30, mean age: 28.8 - L1 Italian, L2 German, L3 English – L2 mean AoA: 16.6,
L3 mean AoA: 10.4
- Group 2: N=30, mean age: 29.5 - L1 German, L2 English, L3 Italian - L2 mean AoA: 9.9,
L3 mean AoA: 22.6
Lexical proficiency was assessed through a verbal fluency test; the scores were identical in
Groups 1 and 2 for the L1s (10/10) and similar for both the L2s (Group 1: 7.2/10, Group 2:
7.5/10) and the L3s (Group 1: 6.6/10, Group 2: 5.2/10).
10
The results of the picture-naming task revealed a significant contrast in switch costs between
Groups 1 and Group 2, despite comparable levels of lexical proficiency. As illustrated in
Figure 1, for Group 1 switch costs from the L1 into the L2 were similar to the switch costs
from the L2 into the L1, whereas costs were higher for switching from both L3 into L1 and
L3 into L2 than vice versa. Group 2 indicated higher costs for switching into the stronger
language for both L1-L2 and L1-L3, but not for L2-L3 in which case switch costs were
similar.
These results challenge the familiar inhibition model of language control and suggest that
trilingual language switching is not only affected by an individual’s lexical proficiency, but
also by whether switching involves linguistically more vs. less related and native vs. nonnative languages.
On pronouns that drop (out of German)
Ylva Falk (Stockholm University, Sweden)
The null-subject parameter (NSP) and its related properties is a well-studied syntactic feature
within the field of L3 learning (e.g., Rothman & Cabrelli Amaro 2010; Montrul et al 2010).
Many of these studies of this feature involve languages that are both typologically close and
thus formally similar. In the present study we will explore whether the L2 status factor (e.g.,
Bardel & Falk 2012) exerts such a strong impact that the NSP from a language that is neither
typlogically close nor similar L2 (Spanish/Italian) can be transferred into the L3 (German).
Data were recorded from two absolute beginners of L3 German (L1 Swedish, L2 English,
French, Spanish, Italian) during a period of five months and consist of semi-guided
spontaneous speech. The preliminary results show that both participants transferred the NSP
from their weakest and most recently learned L2 (Spanish/Italian) into German, cf.:
(1) Zuerst habe eine Lektion
First have a lesson
TL: Zuerst habe ich eine Lektion
(2) Und dann duscht
And then takes-a-shower
TL: Und dann duscht er/sie
In (1 & 2) the subject pronouns are dropped which is not allowed in German, or in the L1
Swedish. In order to quantify the cases of correct/incorrect cases an obligatory occasion
analysis was carried out.
The findings will be discussed in light of the predictions of the L2 status factor, which
hypothesizes that the syntax of an L2 is more likely to be transferred into an L3,
irrespectively of typology or similarity, due to the many cognitive features that L2 and L3
have in common.
11
References
Rothman, J. & Cabrelli Amaro, J. 2010. What variables condition syntactic transfer?: A look
at the L3 initial state. Second Language Research, 26(2), 189-218.
Montrul, S., Dias, R. & Santos, H. 2010. On some null subject parameter-related properties in
the L3 acquisition of Brazilian Portuguese. Estudos da Lingua(gem), 199-232
Bardel, C. & Falk, Y. 2012. The L2 status factor and the declarative/procedural distinction.
In: S. Flynn and J. Rothman (eds.) Third language acquisition in adulthood. Pp. 61-78.
Amsterdam: John Benjamins.
Patterns of cognate facilitation effect in trilingual speakers as an index of the structure
of the trilingual mental lexicon
Agnieszka Lijewska & Hanka Błaszkowska (Adam Mickewicz University, Poland)
The typical finding in experiments involving reaction times (RTs) to cognates (words sharing
form and meaning across languages) and language specific control words is that they take
shorter to respond to in comparison to control words. This discrepancy between RTs obtained
for cognates and controls (the cognate facilitation effect) is typically taken to indicate that
during lexical access words from all languages known to an individual are activated and/or
that representations of cognates are shared across languages (Costa, Santesteban, & Caño,
2005; Dijkstra, Miwa, Brummelhuis, Sappelli, & Baayen, 2010; Lemhöfer & Dijkstra, 2004;
Peeters, Dijkstra, & Grainger, 2013). The majority of studies reported in the psycholinguistic
and neurolinguistic literature have investigated the processing of cognates only in the
bilingual context. Only a few studies have looked into the cognate processing patterns in
trilingual participants (Lemhoefer, Dijkstra, & Michel, 2004; Szubko-Sitarek, 2011, 2014;
Poarch & van Hell, 2012; Lijewska & Chmiel, 2014). Interestingly enough, only two of the
trilingual studies have tested the processing of those cognates which are shared between two
foreign languages rather than between the native and a foreign one (Lijewska & Chmiel,
2014; Szubko-Sitarek, 2014). The results reported by Szubko-Sitarek (2014) revealed no
significant cognate facilitation effect for L3-L2 cognates in an L3 lexical decision task.
Szubko-Sitarek interpreted this as a lack of evidence for the influence of psychotypology (the
perception of cross-linguistic similarity) on the internal structure of the trilingual mental
lexicon and hinted at the possible influence of learning experience. Similarly, on the basis of
the results recorded in two translation tasks, Lijewska and Chmiel (2014) suggested that the
lack of cognate facilitation effect for L2-L3 cognates may indicate that it is not
psychotypology but learning experience which influences how words are linked in the
trilingual mental lexicon.
The present study aimed to investigate processing patterns of L1-L3 cognates (i.e. shared
between the native language and the second foreign language) and L2-L3 cognates (i.e.
shared across two foreign languages) in a group of trilingual speakers. Twenty-nine Polish
learners of German as L2 and English as L3 translated L1-L3 and L2-L3 cognates and
matched control words in two translation tasks. Task 1 involved L3-L1 translation (from
English into Polish) and Task 2 involved L3-L2 translation (from English into German). All
of the participants learned their L3 via their L1. Such a design enabled us to look into the
possible influences of psychotypology and/or learning experience on cross-language
connections in the trilingual mental lexicon. Preliminary results reveal cognate facilitation
12
effect in both translation tasks for both types of cognates but L3-L2 cognates show stronger
facilitation than L3-L1 ones. These results will be discussed in the light of the current
hypotheses and models concerning lexical processing and cross-language connections in the
mental lexicon of trilingual speakers.
References
Costa, A., Santesteban, M., & Caño, A. (2005). On the facilitatory effects of cognate words
in bilingual speech production. Brain and Language, 94(1), 94–103.
Dijkstra, T., Miwa, K., Brummelhuis, B., Sappelli, M., & Baayen, H. (2010). How crosslanguage similarity and task demands affect cognate recognition. Journal of Memory and
Language, 62(3), 284–301.
Lemhoefer, K., Dijkstra, T., & Michel, M. (2004). Three languages, one ECHO: Cognate
effects in trilingual word recognition. Language and Cognitive Processes, 19(5), 585–
611. doi:10.1080/01690960444000007
Lemhöfer, K., & Dijkstra, T. (2004). Recognizing cognates and interlingual homographs:
Effects of code similarity in language-specific and generalized lexical decision. Memory
& Cognition, 32(4), 533–550.
Lijewska, A., & Chmiel, A. (2014). Cognate facilitation in sentence context – translation
production by interpreting trainees and non-interpreting trilinguals. International Journal
of Multilingualism, 1–18. doi:10.1080/14790718.2014.959961
Peeters, D., Dijkstra, T., & Grainger, J. (2013). The representation and processing of identical
cognates by late bilinguals: RT and ERP effects. Journal of Memory and Language,
68(4), 315–332. doi:10.1016/j.jml.2012.12.003
Poarch, G. J., & van Hell, J. G. (2012). Cross-language activation in children’s speech
production: Evidence from second language learners, bilinguals, and trilinguals. Journal
of Experimental Child Psychology, 111(3), 419–438. doi:10.1016/j.jecp.2011.09.008
Szubko-Sitarek, W. (2011). Cognate facilitation effects in trilingual word recognition.
Studies in Second Language, 1(2), 189–208.
Szubko-Sitarek, W. (2014). Multilingual Lexical Recognition in the Mental Lexicon of Third
Language Users (2015 edition.). New York: Springer.
Reflexive binding in Norwegian L2 and L3
Guro Busterud (NTNU, Norway)
In this paper I present results from a study of L2/L3 acquisition of Norwegian reflexive
binding by adult advanced learners. Anaphors vary both in binding distance (1) and
orientation (2). English anaphors can be bound locally only (1a). However, some languages
allow for long distance binding (LDB). The Chinese anaphor ziji can be bound within the
root sentence (1b), while LDB of the Norwegian anaphor seg can cross non-finite sentences
only (1c,d). What counts as a possible antecedent also varies. Norwegian anaphors are
subject oriented (2a,b), while English anaphors can be bound both by subjects and objects
(2c).
47 participants from 3 different L1 backgrounds were included: English (local binding, object
orientation), Russian (non-finite LDB, subject orientation) and Chinese (finite LDB, object
13
orientation). 9 sentence types were tested using a truth-value judgment task and an
interpretation judgment task.
Results from this study show no transfer of L1’s binding system per se. Instead, the results
indicate that some aspects of L1’s binding system are subject to L1 transfer, while there is
little evidence for L2 transfer. It is also interesting to notice that none of the L2/L3 groups
achieve a system which is identical to the binding system of Norwegian L1 speakers.
The English L1 speakers differ statistically from the other groups, as they more readily allow
for object binding. These facts point towards negative L1 transfer (contrary to Cumulative
Enhancement Model’s (CEM) prediction that prior languages remain neutral or enhance L3
acquisition (Flynn et al. 2004). Acceptance of object antecedents is lower among the L1
Russian and L1 Chinese speakers, despite the fact that both groups had acquired English L2
(i.e. an object oriented language) before acquiring Norwegian. This indicates that there is no
L2 transfer of anaphoric orientation, contradicting the L2 transfer hypothesis (e.g. Bardel &
Falk 2007).
The results for binding distance reveal that there is no transfer of LDB per se. Contrary to
Yuan (1998) having LDB in L1 seems not to be a prerequisite for accepting LDB in an
L2/L3. Interestingly, the English group has the highest acceptance of LDB, despite the fact
that their L1 lacks LDB. Several studies reveals that L2 learners tend to accept LDB across
non-finite sentences (1c), but not across finite sentences (1d) (e.g. Finer & Broselow 1986,
Jiang 2009), a pattern called the tensed-infinitive asymmetry (TIA) (Yuan 1994). The
Norwegian binding system resembles this pattern, and it is therefore especially interesting to
look at the TIA in Norwegian L2/L3. Surprisingly, the TIA is not equally distributed among
the L2/L3 groups. The Russians strictly displayed the TIA. Few Chinese L1 and English L1
speakers display the TIA. I analyse these findings as evidence of L1 transfer of the
grammatical category +M FINITE. There is however no evidence of transfer of this category
from L2 to L3, which might be interpreted as contrary to CEM’s statement of language
acquisition being cumulative (Berkes & Flynn 2012). Deciding these data’s impact on
Typological Primacy Model (e.g. Rothman 2011) is difficult, since it is not straight forward
whether any of the languages are (psycho)typological close to Norwegian.
Results from this study reveal no cluster effects: there is no relationship between the
acceptance of LDB and rejection of object antecedents or vice versa (cf. White 1995, Yuan
1998). Acceptance of LDB to object antecedents is low, suggesting that the L2/L3 grammars
are in accordance with UG principles (cf. Thomas 1991, Hamilton 1998, Hawkins 2001).
(1) a. Maryi thought that Susanj criticized herself*i/j.
b. Zhangsani renwei Lisij zhidao Wangwuk xihuan zijii/j/k.
Zhangsan thinks Lisi knows Wangwu like SE
‘Zhangsani thinks that Lisij knows that Wangwuk likes SEi/j/k.’ (Cole, Hermon & Sung
1990: 1)
c. Peteri hørte Tomj le av segi/*j.
Peteri heard Tomj laugh at SEi/*j
d. Peteri hørte at Tomj lo av seg*i/*j.
Peteri heard that Tomj laughed at SE*i/*j
14
(2) a. Peteri fortalte Jonj om seg selvi/*j.
Peteri told Jonj about SELF*i/j
b. Mari forteller Susani at Ann hater segi.
Mari told Susani that Ann hates SEi
c. Maryi told Susanj a story about herselfi/j.
Literature
Bardel, C. & Falk, Y. (2007): «The role of the second language in third language acquisition:
the case of Germanic syntax». Second Language Research 23 (4): 459-484.
Berkes, É. & Flynn, S. (2012): «Further evidence in support of Cumulative-Enhancement
Model». In Cabrelli Amaro, J., Flynn, S. & Rothman, J. (red.) Third Language
Acquisition in Adulthood, 143-164. Amsterdam/ Philadelphia: John Benjamins.
Busterud, G. (2014): Anaforiske bindingskonstruksjoner i norsk som andrespråk.
Unpublished PhD thesis, NTNU.
Finer, D. & Broselow, E. (1986): «Second language acquisition of reflexive binding». In
Proceedings of the North Eastern Linguistics Society 16: 154-168.
Flynn, S., Foley, C. & Vinnitskaya, I. (2004): «The Cumulative-Enhancement Model for
Language Acquisition: Comparing Adults’ and Children’s Patterns of Development in
First, Second and Third Language Acquisition of Relative Clauses». International Journal
of Multilingualism 1 (1): 1-16.
Hamilton, R. (1998): «Undetermined binding of reflexives by adult Japanese-speaking
learners of English». Second Language Research 14: 292-320.
Hawkins, R. (2001): Second Language Syntax: A Generative Introduction. Cornwall:
Blackwell Publishing.
Jiang, L. (2009): «A referential/quantified asymmetry in second language acquisition of
English reflexives by Chinese-speaking learners». Second Language Research 25 (4):
469-491.
Rothman, J. (2011): «L3 syntactic transfer selectivity and typological determinacy: The
typological primacy model». Second Language Research 27: 7-27.
Thomas, M. (1991): «Do second Language Learners Have «Rogue» Grammars of
Anaphora?» In Eubank, L. (ed.) Point counterpoint: Universal Grammar in the second
language, 375-388. Amsterdam: John Benjamins.
White, L. (1995): «Input, Triggers, and Second Language Acquisition: Can Binding Be
Taught?» In Eckman, F.R., Highland, D., Lee, P. W., Mileham, J. & Weber, R. R. (ed.),
Second Language Acquisition Theory and Pedagogy, 63-78. Mawah, NJ: Lawrence
Erlbaum.
Yuan, B. (1994): «Second Language Acquisition of Reflexives Revisited». Language 70 (3):
539-545.
Yuan, B. (1998): «Interpretation of binding and orientation of the Chinese reflexive ziji by
English and Japanese speakers». Second Language Research 14 (4): 324-340.
15
Sensitivity to word order errors in L3 processing: Evidence from self-paced reading and
grammaticality judgments
Ana Stutter Garcia & Claudia Felser (Potsdam Research Institute for Multilingualism,
Germany)
The possible influence of multilingual speakers’ previously acquired languages on L3
acquisition has been the subject of some debate. The L2 status hypothesis claims that other
non-native languages are the main source of transfer in L3 acquisition [1], whilst others have
argued that none of the previously acquired languages is necessarily 'privileged' in this way.
This controversy apart, the extent to which previous findings from acquisition studies extend
to L3 processing has rarely been examined. The primary aim of this study is to examine
whether properties of learners' L2 affect processing and grammaticality judgments in their
L3.
We investigated the L2 status hypothesis by examining L3 German speakers’ sensitivity to
illicit word order patterns that either do or do not have a canonical counterpart in their L1 or
L2. Methods used include an online self-paced reading task with grammaticality judgments
(SPR) and an offline grammaticality judgment correction task (GJCT). Participants included
24 native speakers of German and two groups of high-intermediate learners of German: a
group of L1 Spanish/L2 English/L3 German speakers (n=27), and a group of L1 English/L2
Spanish/L3 German speakers (n=25).
In a total of three experiments, we tested structures that involved illicit object placement.
Experiment 1 examined illicit postverbal pronoun placement in main clauses, a surface
pattern that is grammatical in English but highly marked in Spanish (Table 1). Experiment 2
looked at illicit preverbal pronoun placement in subordinate clauses, a pattern that is the
canonical one in Spanish but which is ungrammatical in English (Table 2). Experiment 3
tested participants' sensitivity to object placement within the German 'verbal bracket', which
has no direct equivalent in either English or Spanish (Table 3).
Analyses were conducted on the accuracy measures of the SPR and the GJCT, while reading
times (RTs) were compared across conditions and groups in the critical and the spillover
regions. The L3 participants had more difficulty detecting ungrammatical sentences in the
SPR than in the offline GJCT across all three experiments, whereas the L1 German only
performed statistically worse in the SPR in Experiment 2. Between-group comparisons for
SPR judgment accuracy showed no significant differences between the two L3 groups in any
of the three experiments. Both L3 groups performed significantly worse than the L1 controls
across all three experiments, however. In the GJCT, the L3 participants' judgment accuracy
was significantly worse than the natives' in Experiment 1 only.
Reading times at the critical region and at the spillover region showed an effect of group in
all three experiments, with the L3 participants displaying significantly slower RTs than the
natives. An effect of condition was found in Experiment 2 and 3, with ungrammatical items
being read more slowly than the grammatical counterparts. No interactions emerged between
group and condition in any of the experiments.
Taken together, our results show a general learner effect, with L3 speakers showing more
reduced sensitivity to word order errors than native speakers especially in the online task, but
no evidence for L2 (or L1) influence.
16
Syntactic transfer from French and Basque to L3 Spanish: the null-subject parameter
Hugues Lacroix (Université de Montréal, Canada)
Spanish and Basque are both pro-drop languages – i.e. languages where the subject can be
implicit, as in examples (1) and (2). Despite their fundamental typological differences, these
two languages share a similar set of rules governing the pro-drop parameter. French, by
contrast, requires an explicit subject argument, as in (3), though it is typologically closer to
Spanish.
(1) pro Vivo en una casa.
pro Live in a house.
“I live in a house”
(2) pro Etxe batean bizi naiz
pro house one in live am.
“I live in a house”
(3) J’habite dans une maison.
I live in a house.
These differences represent a challenge for French-speaking learners of Spanish as a foreign
language, in which they tend to overproduce overt subject arguments (at least at intermediate
and low-proficiency level). In view of this, the question arises whether French speakers’
competence in another pro-drop language (in this case Basque) may positively influence their
acquisition of that parameter in Spanish (i.e., positive transfer), or whether French remains
their primary reference language, leading them to overgeneralize explicit subject arguments
into Spanish (i.e., negative transfer).
The present study attempts to address this question with a semi-controlled oral production
task carried out in the Northern Basque Country with an experimental group of L1 French,
L2 Basque and L3 Spanish speakers (n=9, group 1) and another of L1 Basque, L2 French and
L3 Spanish speakers (n=14, group 2). Regarding the transfer of the null-subject parameter,
we considered the three following hypotheses, according to three different models:
1. L2 status factor (Falk & Bardel, 2011): participants will transfer from their L2, so there
will be a significant difference between the two groups (i.e., group 2 will produce less
explicit overt subject).
2. Cumulative Enhancement Model (Flynn, Foley, & Vinnitskaya, 2004): participants will
transfer from their pro-drop language (Basque), no significant difference between the two
groups will be noted and they will behave almost like native speakers in regard to that
parameter. As they both have a pro-drop language in their previously acquired languages,
they will be able to transfer the null-subject parameter on a per-parameter basis.
3. Typological Primacy Model (Rothman, 2011): participants will transfer from the language
that is typologically closer to Spanish – French in that case, a non-pro-drop language –, no
significant difference will be noted between the two groups and they will behave like any
French-speaking learners of Spanish.
According to the preliminary analysis of our data, the order of acquisition does not seem to
influence significantly the transfer as both groups behave in a similar way. On that account,
the “L2 status factor” does not seem to apply to our experimental groups. Further
comparisons to the control group (native Spanish speakers) and to another experimental
17
group (L1 French, L2 Spanish) will allow us to determine if the two Basque/French/Spanish
experimental groups score better than the French/Spanish group. If that is the case, it would
support our second hypothesis, else it would provide evidence for our third hypothesis.
References
Cenoz, J. (2001). The effect of linguistic distance, L2 status and age on cross-linguistic
influence in L3 acquisition. En J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Crosslinguistic Influence in Third Language Acquisition: Psycholinguistic Perspectives (pp. 820). Clevedon, UK: Multilingual Matters.
Falk, Y., & Bardel, C. (2011). Object pronouns in German L3 syntax: Evidence for the L2
status factor. Second Language Research, 27(1), 59–82.
Flynn, S., Foley, C., & Vinnitskaya, I. (2004). The Cumulative-Enhancement Model for
Language Acquisition: Comparing Adults’ and Children’s Patterns of Development in
First, Second and Third Language Acquisition of Relative Clauses. International Journal
of Multilingualism, 1(1), 3-16. doi:10.1080/14790710408668175
Rothman, J. (2011). L3 syntactic transfer selectivity and typological determinacy: The
typological primacy model. Second Language Research, 27(1), 107–127.
The acquisition of anything by Catalan/Spanish bilingual and Spanish monolingual
learners of English: Evidence for the Typological Primacy Model of L3 acquisition
Eloi Puig-Mayenco (Universitat Pompeu Fabra) & Heather Marsden (Univ. of York, UK)
This study tests two influential models of transfer of previously acquired linguistic
knowledge in L3/Ln acquisition—Cumulative Enhancement (CEM, Flynn et al. 2004) and
Typological Primacy (TPM, Rothman 2011)—through an investigation of knowledge of the
polarity item anything. We compare L1-Spanish learners of L2 English with L1-Catalan-L2Spanish learners of L3 English.
The distribution of English any(thing) is complex, and includes licensing by a nonveridical
operator in questions and conditionals (1–2), and by an anti-veridical operator such as
negation (3) (Giannakidou 2011). However, if any is not c-commanded by a licensing
operator, the result is ungrammatical (4). In Catalan and Spanish two types of quantifier
correspond to English anything, depending on the context: positive indefinites, alguna cosa
(Catalan), algo (Spanish) ‘something/anything’; and negative indefinites, res (Catalan), nada
(Spanish) ‘nothing/anything’. Focusing on the negative indefinites (Vallduví 1994), Catalan
res, like anything, can be licensed by both non- and anti-veridical operators (5−7); but unlike
anything, can optionally precede negation (8). Spanish nada, on the other hand, is only
licensed by anti-veridical operators, thus nada is not grammatical in affirmative questions (9)
or conditionals (10). Moreover, nada must be c-commanded by negation (11–12).
Extrapolating from Gil & Marsden (2013), we assume that transfer of the properties of
res/nada may influence the learners’ behaviour on English anything. If so, then the different
L3/Ln models make predictions for L1-Catalan-L2-Spanish bilinguals on the English
sentence types in (1–4) as follows:
18
• CEM: Facilitative transfer will occur from both Catalan and Spanish, resulting in
nativeEnglish-like judgements on all types.
• TPM: Transfer will occur either solely from Catalan or solely from Spanish. If from
Catalan, learners will be target-like on (1–3), but will also allow the ungrammatical (4). If
from Spanish, learners will be target-like on (3–4), but will not allow (1–2).
The L1-Catalan-L2-Spanish participants were divided into two groups based on onset of
Spanish acquisition before age 7 (“Early”) or from age 7 (“Late”). Participants completed a
web-based Acceptability Judgement Task with 4 tokens each of the types in (1–4) mixed with
28 other sentences. “Beginner” and “Advanced” learners were identified through scores on
the online Cambridge English test.
Rates of accuracy are shown in Table 1. It is notable among the beginner groups that, while
scores are generally >68%, the L1-Catalan-L2-Spanish speakers had low accuracy on the
ungrammatical *any+neg construction (4), suggesting transfer from Catalan. Also, the L1Spanish speakers had low accuracy on conditionals (2) whereas the L3 groups did not,
providing evidence against transfer from Spanish. Together these findings support the TPM
rather than the CEM prediction, with Catalan being the language that the L3 learners transfer
from. We conclude by addressing outstanding questions of: why the L1-Spanish beginners
had high accuracy on questions (a teaching effect); why the advanced L3 groups were
somewhat less successful on *any+neg than the advanced L2 group on conditionals (an input
effect); and why Catalan and not Spanish is the source of transfer in the L3 learners (an effect
of phonology and vocabulary).
19
References
Flynn, S., C. Foley & I. Vinnitskaya. 2004. The Cumulative-Enhancement Model for
language acquisition: Comparing adults’ and children’s patterns in first, second and third
language acquisition of relative clauses. International Journal of Multilingualism 1: 3-16.
Giannakidou, A. 2011. Positive polarity items and negative polarity items: variation,
licensing, and compositionality. In Maienborn, C., K. von Heusinger & P. Portner. (eds.)
Semantics: An International Handbook of Natural Language Meaning (Second Edition).
Berlin: Mouton de Gruyter. 1660-1713.
Gil, K. & H. Marsden. 2013. Existential quantifiers in second language acquisition: A feature
reassembly account. Linguistic Approaches to Bilingualism 3: 117–149.
Rothman, J. 2011. L3 syntactic transfer selectivity and typological determinacy: The
typological primacy model. Second Language Research, 27: 107–127.
Vallduví, E. 1994. Polarity items, n-words and minimizers in Catalan and Spanish. Probus, 6:
263– 294.
The role of language dominance in German-Turkish heritage speakers of L3 English
Anika Lloyd-Smith (University of Hamburg, Germany), Henrik Gyllstad (Lund University,
Sweden) & Tanja Kupisch (University of Konstanz, Germany)
The present study in concerned with global accent in German-Turkish early bilinguals,
heritage speakers (HS) of Turkish, who acquired English as their chronologically third
language (L3). The goal is to shed light on the relevance of language dominance in L3/Ln
language acquisition.
So far, the prevalent question in L3/n acquisition has been the source of transfer into L3. In
phonology, both L1 and L2 can be transfer sources (Gut 2011, Wrembel 2012). L2 transfer is
predicted by the L2 Status Factor Model (Bardel & Falk 2007), while the Typological
Primacy Model claims that perceived similarities between languages determine multi-lingual
transfer (Rothman 2010, Rothman & Cabrelli Amaro 2010). Based on the existing evidence,
it seems impossible to fully support or discard one of these models. Possibly, transfer sources
20
also depend on the speakers’ proficiency level (Hammarberg & Hammarberg). Second, the
learners in most studies had acquired L1 during early childhood and L2 at age 10 or later.
This means their L1 proficiency exceeded their L2 proficiency, and L1 grammatical
development had been completed before L2 acquisition began. We know little about cases in
which L1 and 2L2 were both acquired early, and where L2 became dominant over time, as is
the case for HS.
This study seeks to explore the role of language dominance in L3 acquisition along two
dimensions: (i) relative exposure/use of the L1 and (ii) relative proficiency in the L1.
Interestingly, education oriented-research suggested that the extent to which bilinguals
exhibit positive cognitive learning effects is related to their L1 and L2 proficiency: Positive
cognitive learning effects arise if learners have reached a certain level of proficiency in both
background languages (Cummins 1976). To our knowledge, this alleged proficiency level has
never been made explicit.
We analysed global accent our speakers’ three languages, as well as their relative language
dominance in German and Turkish. The HS were exposed to Turkish from birth, and two
German between 0-9 years; English was acquired later as a foreign language. We carried our
three separate accent rating studies. In the first two, the HS were rated by L1 German
speakers (n=15) when speaking German and by L1 Turkish speakers (n=15) in Turkish.
Degree of accentedness was independent of AoO in German, but related to frequency of
Turkish exposure/use. The HSs’ speech was deemed either (i) native in German/foreign in
Turkish (Group G), or (ii) native in Turkish/foreign in German (Group T), or (iii) native in
both languages (Group B). In the third experiment the HSs’ English was rated by native
English teachers in Germanys (n=20), whose task was (i) to judge the degree of accentedness
and (ii) classify their accent as German, Turkish, English or “other”. Results showed that
Group B to be considered most native-like and group T most foreign-sounding. Group G
tended to be classified as German-sounding, Group B were most heterogeneous, Group T was
mostly classified as “other”. Thus, although all HSs attended German schools and were
exposed to German-accented English, only some were identified as speakers of German,
which excludes the possibility that transfer is systematically conditioned by typological
proximity. At the same time, there was no systematic transfer from L1 Turkish either.
Instead, there is evidence that the degree of language balance between L1 and L2 is crucial
for phonological transfer into L3 English.
Transfer from nowhere? The unknown factor in L3 learning
Christina Lindqvist (Uppsala University) & Ylva Falk (Stockholm University)
The aim of this study is to explore the role of the recency factor for the source of lexical
transfer in L3 learning. This will be investigated in Swedish native speakers’ oral production
during their first encounter with Dutch L3.
Recency of use is one of the four factors that are most likely to determine transfer source, and
which are commonly discussed in the L3 literature (De Angelis 2007). Apart from recency,
L2 status, (psycho)typology and proficiency are often brought up. However, the recency
factor has rarely been explored in detail in previous research on lexical transfer in L3, in
21
contrast to in particular the L2 status factor and the (psycho)typology factor. These two
factors have received a lot of attention in earlier L3 studies (Williams & Hammarberg 1998;
Cenoz et al. 2001; Ringbom 2007). As for the recency factor, Williams & Hammarberg
(1998) suggested that recency, in interplay with the other three factors, influenced the
activation of German L2 in the oral production of Swedish L3. To our knowledge, no study
has singled out the role of the recency factor in L3 learning. Moreover, the very concept of
recency seems to have different denotations. For example, Shannon (1991) views the concept
from a learning perspective, claiming that the most recent language equals the last learned
background language. On the other hand, Williams & Hammarberg (1998) adopt the
perspective of language use, defining the most recent language as the one last used. In our
research we operationalized the recency factor as “language not in use”, i.e. number of years
since the last encounter with the language.
The study includes 41 native speakers of Swedish. They had knowledge of English L2 and at
least one Romance L2. A specific requirement for participation was that they should have no
previous knowledge of Dutch or German. The participants were recorded in pairs during their
first encounter with Dutch. They performed various simple tasks, which were led by a Dutchspeaking person. In the analysis, all instances of lexical transfer were identified and coded for
language source. We had the following research question: Does recency play a role for the
activation of the background languages? Our hypothesis was that the most recently used L2
will be the main source of lexical transfer. In terms of recency, English was the most recently
used L2 for all participants. About half of them also reported on highly recent use of French
and/or Spanish L2.
The results clearly reject our hypothesis. First of all, lexical transfer of English L2 is
extremely rare in the data. Second, the participants with recent use of French and/or Spanish
L2 do not transfer from these languages to the extent that was expected. Third, and most
strikingly, the language that is most frequently transferred is German. However, the
participants had claimed not having any knowledge of this language. The results show that,
all in all, about two thirds of all transferred items belong to German. Possible explanations to
these unexpected results will be discussed.
References
De Angelis, G. 2007. Third or additional language acquisition. Clevedon: Multilingual
Matters.
Cenoz, J., Hufeisen, B. & Jessner, U. (eds.). 2001. Cross-linguistic influence in third
language acquisition: psycholinguistic perspectives. Clevedon: Multilingual Matters.
Ringbom, H. 2007. Cross-linguistic similarity in foreign language learning. Clevedon:
Multilingual Matters. Shannon, B. 1991. Faulty language selection in polyglots.
Language and cognitive processes 6: 339-350.
Williams, S. & Hammarberg, B. 1998. Language switches in L3 production. Implications for
a polyglot speaking model. Applied linguistics 19: 3: 295-333.
22
What accounts for phonetic performance in a third language? A correlational study
Magdalena Wrembel (Adam Mickiewicz University, Poland)
Several scholars have stressed that the acquisition of third language (L3) phonology is a
particularly understudied domain which deserves special attention in scientific investigations
(cf. Hammarberg 1997, Cabrelli Amaro 2013, Wrembel 2015). To this end, a series of
parallel studies was conducted by the present author aimed at gaining a more comprehensive
understanding of L3 phonological acquisition.
This contribution will present the results of three selected measures of phonetic performance
in L3 including ratings of perceived global foreign accent, acoustic measurements of voice
onset time (VOT) and codification of metaphonological awareness. The conducted analysis
involved two selected study groups with different language repertoires (i.e. L1 Polish, L2
English, L3 French vs. L1 Polish, L2 English, L3 German). The multilingual participants
included 60 adult Polish learners with an advanced knowledge of their L2 English (B2/C1)
and elementary to intermediate proficiency in their respective third languages (A1-B1), with
30 participants per each language grouping.
In the first part of the study L3 pronunciation performance was assessed by means of
perceptual ratings involving three components: foreign accentedness, comprehensibility and
pronunciation accuracy judgements. The conducted online survey included 20 raters who
evaluated samples of read L3 speech on 7-point Likert scales. Secondly, acoustic
measurements of VOT of stressed onset plosives elicited through target words read in carrier
phrases in participants’ L1, L2 and L3 were performed. The major objective was to explore
sources of cross-linguistic influence (CLI) in the acquisition of L3 VOT patterns in
comparison to VOT values in the remaining languages of the participants.
Finally, metaphonological awareness was investigated through the application of stimulated
recall verbal protocols, in which the participants were to modify and comment on their L3
pronunciation after listening to excerpts of their text reading recording. The protocols were
audio-recorded, transcribed and coded for the purpose of the objectivisation using a specially
designed coding system (Wrembel 2013, 2014). A composite measure of metaphonological
awareness was calculated on the basis of the observed instances of self-repair of L3
pronunciation, performed phonetic analysis and metacognitive comments on cross-linguistic
interactions.
The findings point to intricate patterns of correlations between the participants’ composite
measures of metaphonological awareness, rated foreign accentedness and VOT values. On
the whole, the participants with higher levels of awareness were perceived as less accented,
more intelligible and accurate in terms of their L3 pronunciation performance. The L3 VOT
values were found to be intermediate between L1, L2 and target values, pointing to a
combined native and non-native language influence. Moreover, significant moderate to
strong correlations between L3 phonetic performance and external factors were found for
several variables including years of formal training in L3, age of onset and L3 proficiency.
The analysis aimed at comparing L3 performance in different language pairings and
providing further evidence for the existing models developed for third language acquisition
(e.g. Flynn et al. 2004, Bardel & Falk 2007, Rothman 2011) in an attempt to tease apart the
influence of typology from L2 status effect in L3 phonological acquisition.
23
References
Bardel, C. & Falk, Y. 2007. The role of the L2 in L3 acquisition: the case of Germanic
syntax. Second Language Research 23(4): 459-484.
Cabrelli Amaro, J. (2012). L3 Phonology: An understudied domain. In: J. Cabrelli Amaro, S.
Flynn, and J. Rothman (eds.), Third Language Acquisition in Adulthood.
Philadelphia/Amsterdam: John Benjamins, 33-60.
Flynn, S., Foley, C. & Vinnitskaya, I. 2004. The Cumulative-Enhancement Model for
Language Acquisition: Comparing Adults’ and Children’s patterns of Development in L1,
L2 and L3 acquisition of Relative clauses. The International Journal of Multilingualism
1(1): 3-16.
Hammarberg, B. 1997. Conditions on transfer in phonology. In Second Language Speech:
Structure and Process, A. James and J. Leather (eds), 161-180. Berlin/New York: Mouton
de Gruyter.
Rothman, J. 2011. L3 Syntactic Transfer Selectivity and Typological Determinacy: The
Typological Primacy Model. Second Language Research 27(1): 107-127.
Wrembel, M. (2013). Metalinguistic awareness in third language phonological acquisition. In
K. Roehr & G. A. Gánem-Gutiérrez (eds.), The Metalinguistic Dimension in Instructed
Second Language Learning. London: Bloomsbury, 119-143.
Wrembel, M. (2014). Metaphonological awareness in multilinguals; a case of L3 Polish,
Language
Awareness,
Taylor
and
Francis.
http://www.tandfonline.com/doi/full/10.1080/09658416.2014.890209
Wrembel, M. (2015, to appear). Cross-linguistic influence in second vs. third language
acquisition of phonology. In U. Gut, R. Fuchs & E.-M. Wunder (eds.), Universal or
diverse paths to English phonology? Berlin: Mouton de Gruyter.
Exploring the Mental Lexicon of the Multilingual: Testing vocabulary knowledge and
cognate recognition in the L2 and L3
Csaba Szabo (The Open University)
Recent empirical findings in the field of Multilingualism have shown that the mental lexicon
of a language learner does not consist of separate or individual entities, but rather of an
intertwined system where languages can interact with each other. Although general language
competence can differ for different languages, when acquiring a new language, as a
multilingual language user, the learner will undoubtedly rely on previous language
knowledge and learning competence (see e.g. Szubko-Sitarek, 2015; Daryai-Hansen et. al,
2014; Falk and Bardel, 2010). Accordingly, multilingual language learners have been
considered differently to second language learners in a growing number of studies, however
studies on the variation in learners lexical knowledge both in the L2 and L3 and the effect of
cognates on the target languages has been relatively scarce (see e.g. Cenoz, 2013).
In this view, the reported study offers a unique contribution to the discussion of the impact of
prior lexical knowledge on additional language learning in case of the Hungarian minority in
Romania studying English as an L3 in that it (1) is the only known study in this cultural
context that empirically investigates knowledge both in the L2 and L3 and (2) it employs a
new frequency-based vocabulary testing-tool (Romanian Vocabulary Levels Test; Szabo,
2014) to investigate the effect of cognate facilitation and cross-linguistic transfer in the case
24
of the L2 and L3. In particular, the primary aim of this study is to bring to light an alternative
conceptualisation of the Vocabulary Levels Test (VLT; Nation, 1983) widely used in SLA
and, by deploying it to investigate the role and influence of Romanian as a Romance
language and an L2 on the acquisition of English. The justification for this alternative
approach is evaluated through a quantitative study where the results indicated that the new
tool is not just valid and reliable, but can meaningfully foster the comparison of lexical
proficiency in two different languages.
The participants (N = 40) have been divided into a study and control group and the results
elicited by the two tests (the VLT and RomVLT) are complemented by a self-rating
questionnaire. The data shows a significant correlation (Pearson’s r = .792, p < .001) between
students’ L2 and L3 overall lexical proficiency and cognate knowledge (r = .861, p < .001).
However, it is also indicated, that the explicit instruction on cognates between L2 and L3 had
no effect on the study group in case of adult students. Taking into account the
Interdependence Hypothesis (Cummins, 1978), the study suggests that investigating
multilinguals prior lexical knowledge is essential, especially as proximate languages (in this
case Romanian) will have an impact on the target interlanguage (English) and learners can
make use of this either intrinsically or through explicit instruction. Finally, the data supports
the hypothesis that the order of language acquisition and the context of acquisition in the case
of historical bilinguals are not necessarily direct indicators of language proficiency.
References
Cenoz, J. (2013). The influence of bilingualism on third language acquisition: Focus on
multilingualism. Language Teaching, 46, 71-86.
Cummins, J. (1978). Metalinguistic development of children in bilingual education programs:
Data from Irish and Canadian Ukrainian-English programs. In Paradis, Michel (ed.),
Aspects of bilingualism, 127-138 Columbia, SC: Hornbeam.
Daryai-Hansen, P., Gerber, B., Lörincz, I., Haller, M., Ivanova, O., Krumm H. & Reich H.
(2014). Pluralistic approaches to languages in the curriculum: the case of Frenchspeaking Switzerland, Spain and Austria. International Journal of Multilingualism,
DOI: 10.1080/14790718.2014.948877.
Falk, Y. & Bardel, C. (2010). The study of the role of the background languages in third
language acquisition. The state of the art. International Review of Applied Linguistics in
Language Teaching. IRAL 48(2-3): 185–220.
Nation, P. (1983). Testing and teaching vocabulary. Guidelines, 5, 12-25.
Szabo, Cs. (2014, in print). Exploring Linguistic Variation through the Romanian Frequency
List and the Romanian Vocabulary Levels Test. Current Issues in Linguistic Variation:
The 14th international conference of the Department of Linguistics. University of
Bucharest: Bucharest University Press.
Szubko-Sitarek, W. (2015). Multilingual Lexical Recognition in the Mental Lexicon of Third
Language Users. In: Miroslaw Pawlake, ed. Second Language Learning and Teaching.
Springer. Berlin, Heidelberg, DOI: 10.1007/978-3-642-32194-8.
25
The Acquisition of L3/Ln Morphosyntax in Norwegian Learners of Turkish
Emel Turker -Van der Heiden & Gözde Mercan (University of Oslo, MultiLing, Norway)
This study aims to investigate the development of certain morphosyntactic forms in
Norwegian university students learning Turkish as a foreign language. For this particular
population, Turkish is a L3 or Ln, as native speakers of Norwegian learn at least one
language (English) at school. To the best of our knowledge, this is the first research project to
investigate the acquisition of L3/Ln Turkish in this population.
This paper focuses on the development of L3/Ln Turkish morphosyntax from the perspective
of Processability theory (Pienemann, 1998; 2005), a psycholinguistic account of the
developmental sequences in language learning, aiming to explain the order of emergence of
morphosyntactic phenomena.
The present data comprise of written final exam papers of Norwegian students learning
Turkish at the University of Oslo, across four semesters.
In particular, the genitive-possessive (GEN-POSS) agreement morphology is examined in
nominal groups and verbal constructions (noun clauses and relative clauses). To get a
complete picture of the acquisition process, we not only analyze the errors, but also consider
the native-like uses of the structures by the learners.
For example, in Turkish possessive NPs, the first one of the two consecutive NPs expresses
the possessor (1a). It modifies the second NP and is marked with the GEN case -(n)In. The
second NP, which is the head, carries an a POSS marker -(s)I(n), agreeing with the possessor
in person and number. (Kornfilt, 1997; Göksel & Kerslake, 2005; 2011).
(1a) Türkiye-nin (en ilginç) şehr-i
Turkey -GEN (most interesting) city-POSS.3SG
‘Turkey’s (most interesting) city / the (most interesting) city of Turkey’
The phrase (1b) produced by a second-semester learner with the same intended meaning as
the native form (1a) illustrates an omission error, as the POSS agreement is missing.
(1b) Türkiye-nin(GEN) (en ilginç) şehir-Ø
On the other hand, relative clauses are adjective clauses which modify a NP. In Turkish if the
relativized NP is the (direct/oblique) object or an adverbial modifier of the verb in the relative
clause, then the verb of the relative clauses carries the participle suffix -DIK (for past and
ongoing situations). The subject of the clause is in GEN case and the participle has the POSS
agreement marker (2a). (Göksel & Kerslake, 2005)
(2a) [Orhan Pamuk-un büyü -düğ -ü mahalle
Orhan Pamuk-GEN grow up-OPART-POSS.3SG neighbourhood
‘the neighbourhood in which/where Orhan Pamuk grew up’
The clause (2b), also from a fourth-semester learner demonstrates the erroneous addition of
extra suffixes. In addition to the target participle -DIK and the GEN-POSS agreement
morphemes as used in (2a) for the same intended meaning, this learner also added an ACC
suffix to the verb of the clause and an extra POSS agreement marker to the relativized noun,
which leads to ungrammaticality.
26
(2b) [Orhan
si(POSS.3SG)
Pamuk-un(GEN)
büyü-düğ(OPART)-ü(POSS.3SG)-nü(ACC)]
mahalle-
In this paper we will discuss the possible reasons why even the most advanced group in our
population who are very competent in other complex aspects of the language, make such
mistakes from the Processability theory perspective, in an effort to shed some light on the
processing of L3/Ln Turkish morphosyntax.
References
Göksel, A. & Kerslake, C. 2005. Turkish, a comprehensive grammar. London: Routledge.
Göksel, A. & Kerslake, C. 2011. Turkish, an essential grammar. London: Routledge.
Kornfilt, J. 1997. Turkish Grammar. London: Routledge.
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.
Persistence of non-facilitative transfer in L3 development
Felipe Amaro, Jennifer Cabrelli Amaro (University of Illinois at Chicago) and Jason
Rothman (University of Reading / University of Tromsø)
The Typological Proximity Model (Rothman, 2010, 2011, 2013a, 2013b) claims that
structural similarity across languages determines the source language (L1 or L2) for
morphosyntactic transfer at the initial stages of third language (L3) acquisition. One
consequence of the model is that non-facilitative transfer can obtain, defined as
morphosyntactic transfer of a domain from the typologically similar language when the other
language would provide a target-like structure. While substantial literature supports
typologically selective transfer (see Rothman, 2013a, for an overview), few studies have
investigated developmental consequences of non-facilitative transfer. The present study
examines changes to the feature configuration of embedded T in L3 Brazilian Portuguese
(BP) among L1 English/L2 Spanish bilinguals after initial non-facilitative transfer from
Spanish has obtained. Specifically, we investigate if and when these learners converge on the
BP embedded T.
In Spanish, the dative clitic of the raising verb parecer ‘to seem/appear’ cannot co-occur with
raising across an experiencer (RExp) as in *Pedro me parece saber todo. Conversely, as in
English, RExp is acceptable in BP: O Pedro me parece saber tudo ‘Pedro appears to me to
know everything’. This contrast is due to different featural specifications of embedded T
(Ausín & Depiante, 2000): In English and BP, embedded T is defective and assigns no Case.
However, embedded T in Spanish is not defective, and assigns case to Pedro. Movement of
Pedro across the experiencer me ‘to me’ therefore constitutes a violation in Spanish only.
Recent research of transfer of embedded T to L3 BP by English/Spanish bilinguals (CabrelliAmaro & Rothman, in press) shows initial stages transfer of the Spanish configuration,
whether Spanish is the L1 or L2.
27
Using Cabrelli-Amaro and Rothman’s data as a point of comparison, we tested 15 L1
English/L2 advanced Spanish/L3 advanced BP learners and 14 BP native controls to
determine whether learners reconfigure the feature specification in L3 development such that
BP parecer with an experiencer subcategorizes for a defective T. A four-point scalar
grammaticality acceptability task was administered in Spanish and BP. Acceptance of
sentences with parecer was tested in three conditions: with an embedded TP complement
projecting an experiencer (TPExp, example 1), with a vP complement and without an overt
experiencer (vPnoExp, example 2), and with an embedded AP complement (APExp, example
3). TPExp is ungrammatical in Spanish and grammatical in BP and English, while APExp
and vPnoExp are grammatical in all three languages.
Comparing learner and control BP data (Figure 1), a Repeated Measures ANOVA revealed
no significant Group*Condition interaction. Therefore, learners pattern with BP controls and
do not block RExp. Presuming the learners had a Spanish feature configuration of embedded
T at the L3 initial stages as shown in Cabrelli-Amaro and Rothman (in press), we contend
this advanced group has successfully reconfigured the feature specification and retreated
from initial nonfacilitative transfer. Interestingly, the Spanish data (Figures 2 and 3) suggest
regressive BP influence on the Spanish embedded T configuration. In light of these data, we
address the role of L3 àL2 transfer as a correlate of L3 proficiency gains.
Examples
(1) a. BP: Rodrigo me parece [TP saber inglês]
b. Spanish: *Rodrigo me parece [TP saber inglés]
c. English: Rodrigo seems to me [TP to know English]
(2) a. BP: Cristina parece [vP precisar de algo]
b. Spanish: Cristina parece [vP necesitar algo]
c. English: Cristina seems [vP to need something]
(3) a. BP: João me parece [AP cansado]
b. Spanish: Juan me parece [AP cansado]
c. English: John seems to me to be [AP tired]
Figures
28
Selected references
Cabrelli-Amaro, J., & Rothman, J. (in press). The relationship between L3 transfer and
structural similarity across development: In H. Peukert (Ed.), Transfer effects in
multilingual language development. Amsterdam: John Benjamins.
Rothman, J. (2013a). Cognitive Economy, Non-redundancy and typological primacy in L3
acquisition: Evidence from initial stages of L3 Romance. In S. Baauw, F. Dirjkoningen &
M. Pinto (Eds.), Romance languages and linguistic theory 2011 (pp. 217-248).
Amsterdam: John Benjamins.
Norwegian-Russian bilinguals learning English: The Linguistic Proximity Model
Marit Westergaard (University of Tromsø, Roksolana Mykhaylyk (Harvard University),
Natalia Mitrofanova (University of Tromsø) and Yulia Rodina (University of Oslo)
Several hypotheses have been developed with regard to cross-linguistic influence in
multilingual (Ln) acquisition: e.g., the Cumulative Enhancement Model (Flynn et al. 2004),
the Typological Primacy Model (Rothman 2011), and the L2 status factor (Bardel & Falk
2007). Our study contributes to this line of research, suggesting a new hypothesis, the
Linguistic Proximity Model (LPM), and providing new data on Ln acquisition of two English
structures: 1) Adv-V word order and 2) Subject-Auxiliary inversion (residual Verb-Second,
V2). According to Westergaard (2003), L1 Norwegian children often transfer V-Adv word
order in declaratives to English and need time to ‘unlearn’ V2, cf. (1). Russian native
speakers, on the other hand, prefer Adv-V word order (Kallestinova & Slabakova 2008), but
might have difficulties with V2 in questions since in Russian this structure is possible only
with a distinct prosody, cf. (2).
We address two main questions in our study: 1) Will bilingual Norwegian-Russian
adolescents have an advantage in learning English compared to monolingual children?, and
2) What is a stronger facilitating factor in Ln learning: general typological similarity or
linguistic structural similarity? According to the LPM, transfer in Ln acquisition occurs when
a certain linguistic property receives strong supporting input from the involved languages,
regardless of the order of acquisition (L1 or L2) or their general typological grouping. This
predicts that Russian syntactic properties will help children learn English Adv-V word order
and overcome Norwegian V2 influence.
In order to verify these predictions, we tested three groups of 12-13-year-old English
learners: L1 Norwegian (N=33), L1 Russian (N=25), and 2L1 Norwegian-Russian (N=12),
matched for proficiency with a modified BPVS-II (1997) test. A Grammaticality Judgment
Task was presented in the form of a pre-recorded test with 24 pseudo-randomized items and
24 fillers; see (1)-(2) and Table 1.
The overall results (Figure 1) show that the bilingual children have the same rate of
correct answers in both conditions (75%). A mixed modeling statistical analysis, comparing
bilinguals to Norwegian L1s, indicates that there are significant effects of the condition
(p<0.000), language group (p=0.027) and their interaction (p=0.000), but a comparison of the
bilinguals to the Russian L1s does not show the same results. The sentence grammaticality,
on the other hand, has a significant effect on the rate of correct answers for all three groups of
learners (p=0.000). For instance, the bilingual children evaluate the ungrammatical V-Adv
structure more correctly than L1 Norwegian children (88% vs. 67%; see Figure 3).
29
The data analyzed thus far suggest that while L1 Norwegian children over-accept
ungrammatical sentences in English with Norwegian word order (V-Adv), the bilingual
children notice these errors more often due to the facilitating influence of Russian. The role
of Norwegian in the bilingual data is less obvious, however, despite the Norwegian-dominant
language environment. To conclude, the ‘unlearning’ of structural properties such as V2 can
be better explained by the LPM rather than with other models, since Russian is not the L2 for
our bilingual participants, nor does it belong to the same typological group as English and
Norwegian. (500 words)
(1) English:
Emma
often
eats
sweets.
ENG = RUS
Russian: Emma
chasto jest
konfety.
Norwegian:
Emma
spiser ofte
konfekt.
(2) English:
What will
Norwegian:
Hva
Russian: Chto
the little girl
read?
vil
den lille jenta
lese?
malen’kaja devochka budet chitat’?
ENG = NORW
Table 1. Stimuli: Two conditions (N=24)
Structure
Condition 1 - Adverbs:
Declaratives with an adverb (N)
Condition 2 - Questions:
Questions with an auxiliary (N)
Grammatical
Ungrammatical
Wh Aux S V? (6)
*Wh S Aux V? (6)
S Adv V O (6)
*S V Adv O (6)
100 80 82 60 75 75 83 73 60 40 C1: Adverbs C2: Questions 20 0 L1Norw Norw-­‐Rus L1Rus Figure 1. Correct judgments of structures in two conditions - with adverbs (as in (1)) and
questions (as in (2)), %
100 80 83 60 40 63 83 78 67 53 L1Norw Norw-­‐Rus L1Rus 20 0 C1: Adverbs C2: Questions Figure 2. Correct judgments of grammatical sentences, %
30
100 80 60 88 83 67 80 72 79 L1Norw 40 Norw-­‐Rus 20 L1Rus 0 C1: Adverbs C2: Questions Figure 3. Correct judgments of ungrammatical sentences, %
Selected references
British Picture Vocabulary Scale II (BPVS II). 1997. Windsor, Berks: NFER- Nelson.
Kallestinova, E. & Slabakova, R. 2008. Does the verb move in Russian? In Proceedings of
FASL 16, 199-214. Michigan Slavic Publications.
Westergaard, M. 2003. Unlearning V2. Transfer, markedness, and the importance of input
cues in the acquisition of word order in English by Norwegian children. EUROSLA
Yearbook 3, 77–101.
The acquisition of Path in L4 German and the effect of previous languages
Veronique Scheirs (Université Catholique de Louvain, Belgium)
In this study, I investigate the acquisition of German motion expressions by adult
multilingual learners and the effects of previously learnt languages in the early stages of
acquisition (up to level A2 in the CEFR).
Spatial expressions and their acquisition are a major topic in the literature in various
frameworks (see among others Hickmann & Robert 2006, Asbury et al. 2008, Cinque &
Rizzi 2010 for an overview). It is generally accepted that Romance and Germanic differ in
their characteristic expression of motion. Using the terminology coined by Talmy’s (1975,
1985, 2000) influential typology of motion events, Romance generally conflates Path in the
verb, as in the French example in (1). Germanic on the other hand, encodes Path by means of
‘satellites’, (e.g. pre- or postpositions, verbal particles or adverbs), while the verb root
encodes manner of motion as illustrated for German in (2).
Difficulties and transfer in the L2-acquisition of a typologically different language have been
attested in a number of studies (Becker & Carol 1997, Inagaki 2001, Bernini et al. 2006,
Hohenstein et al. 2006, Goschler 2009, Larrañaga et al. 2011), however to my knowledge, the
issue has not yet been addressed in the context of subsequent L3 or L4 acquisition.
The participants of the study (n=46) were all monolingual native speakers of French, and had
learnt English at school. Two groups could be distinguished as to the L3 they had learnt prior
to German. While the first group (n=25) had studied the typologically close Dutch, the
second group (n=21) had studied another romance language (Italian or Spanish). The data
was collected using an elicited production narrative task, giving rise to a large number of
motion expressions.
The results show that while all learners seem to eventually acquire the target language’s Pathpattern, a number of differences between the two groups can be noted, pointing to the
31
influence of one of the non-native languages on the acquisition of German. Especially the
first group, which had started learning Dutch prior to German, showed numerous quite
obvious instances of transfer from Dutch, as was found in an analysis of their use of verbs,
prepositions and verbal particles. Besides positive transfer, with increased use of lexical
elements (mainly prepositions and one manner verb) or syntactic patterns that are almost
identical in Dutch and German, negative transfer was also observed. Indeed the Dutch group
frequently fails to realize overt case morphology and instead marks the difference between
the locative and directional reading of case changing prepositions with different word orderas is correct in Dutch. These observations of not only positive but also negative transfer lead
to the conclusion, that in the initial stage of the acquisition of German, the learners’ structure
of Dutch is transferred fully in the sense of Schwartz & Sprouse (1996).
The source of transfer is, however, less clear in the Romance group. The similarities between
English and the Romance languages, especially as they might be emphasized in the learner’s
interlanguage, make it impossible to distinguish and isolate one clear source.
32