PhD project:

The case of the non-native-like first language: Neurophysiological evidence of first-language attrition

Supervisor: Dr. Karsten Steinhauer, McGill University

While multilingual speakers worldwide agree that it is more challenging to learn a second-language (L2) in adulthood than in childhood, whether this is due to a neurobiological “critical-period” for language-learning is controversial.  According to many, there are maturational limits on the brain’s ability to change with experience; thus, while one’s first-language (L1) is hard-wired and stable from the early years of life, any L2 learned after a certain age must rely on different (non-native) processing systems in the brain than those used for the L1. In both behavioral and brain imaging research, it has been shown that bilingual speakers’ linguistic performance and processing in a language acquired at a later age typically falls short of being “native-like”, especially in complex and subtle aspects of grammar. However, because the L2 is usually the less-proficient language, studies have not always been able to determine whether late learners show differences from native speakers because they acquired the language late, or because they do not have sufficient command of the language. Therefore, it is still unresolved whether experiential factors such as proficiency level or exposure have a greater impact than age-of-learning on how language is processed in the brain. First-generation-immigrants who move to a new country in adulthood offer new light on this controversial question, as they become highly-proficient in the late-acquired but predominantly-used L2, while experiencing “attrition” (a decline in proficiency) in their native-L1, after years of limited exposure.

The main goal of this research project is to tease apart age-effects from proficiency-effects in the study of how first- and second-languages are treated in the brain. “Attriters” (those who experience attrition) bridge the gap between the study of L1- and L2-acquisition and allow us to approach the “critical period hypothesis” from a different perspective: are “attriters” still native-like in how their brain processes their first-language, despite their self-reports of gradually-increasing difficulties in that language since immigration? Are they native-like in the second-language they were immersed into in adulthood, and do they show interference effects from this second-language onto their first-language?

To study these questions, my PhD project comprises 6 ERP experiments and a number of behavioral tasks conducted in two languages (Italian and English) with 4 separate participant groups: (1) Italian-English attriters (immigrants who moved to Montreal from Italy in adulthood and are now highly-proficient and dominant in English, reporting difficulties/attrition in Italian); (2) English-Italian L2-learners (who acquired Italian in adulthood and are highly-proficient); (3) Italian monolingual native-speakers living in Italy; (4) English monolingual native-speakers in Montreal. Our ERP and behavioral experiments span a number of research domains, as they are driven by hypotheses from theoretical linguistics, psycholinguistics, sociolinguistics, and cognitive-neuroscience. Our methods also span different methodologies, as we use detailed questionnaires, self-report scales, behavioral proficiency measures, timed reading tasks, timed writing tasks, timed production tasks, a priming study, and finally event-related potentials, to look at both vocabulary and grammar. Importantly, we also bridge the gap between behavioral methods and neurocognitive-methods by conducting correlations and regressions between the two methods, in order to get a “big picture” understanding of the many factors at play in bilingual populations. Specifically, the language domains under investigation are: (1) number agreement; (2) word-order in relative clauses; (3) regular/irregular verbs; (4) semantically confusable words; and (4) false-friends, interlingual cognates and homographs. Together, these studies are among the very first worldwide to investigate the neurophysiological correlates of first-language attrition. In terms of L2 processing, these studies are also among the first to examine L2-to-L1 transfer using event-related brain potentials.

To find out about some of our findings, check the list of ongoing presentations and papers on the Publications page. If you have questions about my experimental design (groups, measures, experimental conditions), feel free to email me (kristina.kasparian[at]

Additional research projects:

Effects of L2-proficiency and L1-background on L2 sentence processing: An ERP study of adjective-noun word order in French and Mandarin late learners of English.

Collaborators: Dr. Karsten Steinhauer, Dr. John E. Drury, Nicholas Bourguignon (PhD Candidate)

This study examines: (i) effects on second language (L2) processing tied to transfer/interference between L1 and L2 grammars and (ii) influences of L2-proficiency on the neurocognitive mechanisms underlying online sentence processing in late/adult L2 learners. The processing of correct/incorrect adjective-noun order in English (e.g. the big/white vase; not: the vase *big/ *white) is investigated in an ERP sentence reading/judgment study.

We tested native-French and native-Mandarin late-learners of English as well as native English speaking controls. In English and Mandarin, adjectives are uniformly pre-nominal. In French, however, while some adjectives occur pre-nominally (e.g. le grand vase), most occur post-nominally (e.g. le vase blanc), creating a potential L1-L2 conf lict.

Native English speakers elicit a biphasic ERP pattern (N400+P600) for all violations. Both L2 groups demonstrate this (‘nativelike’)
ERP-profile (contra strong versions of the Critical Period Hypothesis). However, only the French-L1 group show an additional
effect, and only where we expected L1-L2 conflict. In the condition that constitutes a violation in French but is grammatical in English (e.g., the white vase = le *blanc vase), French-L1 learners show an N400 followed by a P600-type effect time-locked to the adjective. This suggests that L1 grammar is activated during L2 processing (extending similar findings for lexical processing by Thierry and Wu, 2007 to the domain of syntactic processing). Interestingly, this initial L1-activation is quickly overridden, as it does not impede native-like processing of the L2 English violation, neither in terms of ERP patterns nor in subjects’ end-of-sentence grammaticality judgments.

We are also investigating: (a) how L2 ERP-profiles may be predicted by proficiency level, regardless of L1-background, and more so by structure-specific than by general L2-proficiency (cloze-test), and (b) whether low L2-proficiency gives rise to more persistent L1 transfer effects (and less native-like L2 processing).

Proficiency and L1-L2 transfer effects in nominal morphology: ERP evidence from French and Mandarin learners of English

Collaborators: Dr. Karsten Steinhauer, Dr. John E. Drury, Nicholas Bourguignon (PhD Candidate)

This study examines the effects of L1-background (i.e., transfer), and  L2-proficiency in late L2 learners, focusing on Nominal Morphology (NM, i.e., articles and their interaction with plural/singular markers). Some languages (e.g., English/French) make use of this information, whereas languages devoid of NM (e.g., Mandarin) raise the controversial issue of whether the absence of such features in the L1 favors native-like L2 processing. Further, though some ERP evidence has revealed distinct processing mechanisms in L1 vs. late-L2, reliance on native-like mechanisms may increase with L2 proficiency.

ERP and behavioral (grammaticality judgment) data from French-L1 and Mandarin-L1 (N=18) late-learners of English were compared to native controls. Target sentences contained noun phrases article-noun number mismatches (e.g. “They paved a road/a *roads in the summer”) which was expected to elicit a LAN/P600 in native speakers. Of interest was whether this mismatch would yield different patterns in the L2-groups, and whether variations in ERP-responses related to L1-background, age of acquisition (AoA) and L2-proficiency would be observed.

As predicted for the native-speakers, the mismatching nouns (“a *roads”) elicit a left anterior negativity (LAN, 350-450 ms), followed by a P600. In contrast, French and Mandarin participants show an N400/P600 pattern and a subsequent anterior negativity. Interestingly, irrespective of L1 background, the P600 amplitude in L2 learners is correlated with proficiency, and more so for structure-specific measures (error rates) than for general L2 proficiency (cloze-test). In a stepwise regression, only the behavioral error rates (acceptability for incorrect sentences) survives as a significant predictor of size of P600 effect.

So far, our results do not clearly support L1-background/transfer effects. Our P600 findings are consistent with a proficiency-dependent continuum, rather than a categorical L1 vs. L2 distinction, and furthermore suggest that future research should more carefully examine structure-specific (rather than general) proficiency measures.