Neurophysiological correlates of L1 attrition and L2 acquisition: A continuum based on proficiency Kristina Kasparian1,2 & Karsten Steinhauer1,2

(1) School of Communication Sciences & Disorders, McGill University, Canada; (2) Center for Research on Brain, Language and Music, Canada

INTRODUCTION STUDY 1: ITALIAN LEXICAL-SEMANTICS

PARTICIPANTS:

Neurobiological “critical period” for language is still controversial: - Claim that maturational constraints on brain plasticity1,2,3 impede “native-like” neuro-cognitive mechanisms of L2-processing4, whereas L1 is “privileged” in that it is hard-wired5 and stable. - Opposing view: Factors such as proficiency or exposure have a greater impact than age-of-acquisition (AoA) on language processing in the brain6.

Immigrants immersed in new language in adulthood allow us to explore this question, as they become highly-proficient in late-L2 while experiencing changes or attrition in L1.

Aims: - Effect of proficiency level on patterns of L1- and L2-processing

- Does proficiency modulate processing patterns, regardless of whether the language was learned as L1 or L2? - Does L1-attrition parallel L2 acquisition7, but in reverse?

- Neural correlates of L1-attrition (largely unexplored to date) - Is there evidence of attrition in online L1-comprehension

Four groups: (1) Attriters: First-generation immigrants of Italian (L1) + English

(advanced, dominant, late-acquired L2). Unanimously report L1-decline Age = 36 yrs; AoA-English = 28 yrs; LoR = 9 yrs; n = 24

(2) Late L2-learners: English (L1) + Italian (advanced, late-acquired L2) Age = 31.6; AoA-Italian = 20; n = 20

(3) Italian native-speakers in Italy (monolingual controls) Age = 30.6, n = 30

(4) English native speakers (monolingual controls) Age = 30.5, n = 30

- Several ERP (reading) experiments assessing syntax8,9 (not reported here) and lexical-semantics10 in Italian and English.

- Proficiency in each language is derived as a composite measure based on written C-test11,12, written error-correction test, and verbal semantic fluency13; participants divided into High/Low subgroups by median split.

- Task during ERP recording: Acceptability judgment (rating 1-5).

Study 1: Difficulties with lexical-semantics within L1? - Italian sentences with confusable words (cappello (hat) vs. cappella (chapel);

mento (chin) vs. menta (mint)) - Is detection of violation mediated by proficiency level, both in L1-

attriters and L2 learners? Study 2: Co-activation of L1 meaning when reading in L2?

- English sentences with interlingual ‘false-friend’ homographs (estate (property vs. summer) and interlingual cognates (cabin/cabina; idea/idea)

- Is there co-activation14-16 for both bilingual groups, or just L1 to L2? - Is there less/no L1-co-activation with increased attrition?

BACKGROUND & AIM:

References: [1] Penfield, W., & Roberts, L. (1959). Speech and Brain Mechanisms. New York: Athenaeum. [2] Lenneberg, E. (1967). Biological foundations of language. New York: Wiley. [3] Ullman, M. T. (2001). The neural basis of lexicon and grammar in first and second language: The declarative/procedural model. Bilingualism: Language and Cognition, 4(1), 105-122. [4] Weber-Fox, C. M., & Neville, H. J. (1996). Maturational constraints on functional specialization for language processing: ERP and behavioural evidence in bilingual speakers. Journal of Cognitive Neuroscience, 8(3), 231-256. [5] Marchman, V. (1992). Constraint on plasticity in a connectionist model of the English past tense. Journal of Cognitive Neuroscience, 4, 215-234. [6] Steinhauer, K., White, E. J., & Drury, J. E. (2009). Temporal dynamics of late second language acquisition: Evidence from event-related brain potentials. Second Language Research, 25(1), 13-41. [7] Tanner, D., Osterhout, L., et al. (in press). [8] Kasparian, K., Vespignani, F., & Steinhauer, K. (2013). Age-of-acquisition or proficiency? An ERP study of morphosyntactic processing in first language attrition. Talk presented at the International Symposium on Bilingualism (ISB9), Singapore, June 10-13, 2013. [9] Kasparian, K., Vespignani, F., & Steinhauer, K. (2013). When the second language takes over: ERP evidence of L1-attrition in morphosyntactic processing. Talk presented at the International Conference on Multilingualism, Montreal, Canada, October 24-25, 2013. [10] Kasparian, K., Vespignani, F., & Steinhauer, K. (2013). My Italian is not what it used to be: Investigating the neural correlates of L1 attrition and late L2 acquisition. Poster presented at the Workshop on Neurobilingualism, Groningen, The Netherlands, August 25-27, 2013. [11] Keijzer, M. (2007). Last in first out? An investigation of the regression hypothesis in Dutch emigrants in Anglophone Canada. PhD dissertation, Vrije Universiteit Amsterdam. [12] Kras, T. (2007). L2 acquisition of the lexicon-syntax interface and narrow syntax by child and adult Croatian learners of Italian. PhD dissertation, University of Cambridge. [13] Schmid, M. (2011). First language attrition. Cambridge University Press. [14] Thierry, G., & Wu, Y. J. (2007). Brain potentials reveal unconscious translation during foreign-language comprehension. PNAS, 104(30), 12530-12535. [15] Dijkstra, T., & Van Heuven, W. J. B. (2002). The architecture of the bilingual word recognition system: From identification to decision. Bilingualism: Language and Cognition, 5, 175–197.[16] De Bruijn, R.A., Dijkstra, T., Chwilla, D.J., & Schriefers, H.J. (2001). Language context effects on interlingual homograph recognition: Evidence from event-related potentials and response times in semantic priming. Bilingualism: Language and Cognition, 4, 155–168.

Acknowledgements: (1) Supervision: Dr. Karsten Steinhauer (McGill University) & Dr. Francesco Vespignani (Università degli Studi di Trento, Italy); (2) Valuable input: Dr. John Drury (Stonybrook University, USA), Dr. Francesca Postiglione (Università degli Studi di Trento, Italy), Dr. Monika Schmid (University of Essex, UK), Dr. Eleonora Rossi (Pennsylvania State University); (3) Research assistants: Kristina Maiorino, Dr. Paolo Zandomeneghi, Filippo Vicari, Lucia D’Arienzo, Linna Jin.

Correct (each word in pair occurs in proper context)

Per coprire la testa, il pescatore porta il cappello di lana. (To cover his head, the fisherman wears the hat of wool.)

Swap (words in pair are switched)

Per coprire la testa, il pescatore porta la cappella di lana. (To cover his head, the fisherman wears the chapel of wool.)

Mismatch (word from a different pair)

Per coprire la testa, il pescatore porta la menta di lana. (To cover his head, the fisherman wears the mint of wool.)

N400 (300-550ms): Controls (Fig. a) high-proficiency ‘attriters’ (Fig. b) and high-proficiency L2-learners (Fig. c) show a significant N400 in both violation conditions. Low-proficiency attriters (Fig. d) (i.e., more attrition) and low-proficiency L2-learners (Fig. e) show N400 only for mismatch condition, failing to automatically detect the swap.

Proficiency is the meaningful factor (cond x prof, p < 0.05) rather than group (cond x group, p = 0.35). High proficiency speakers elicit a significant N400 for swap vs. correct (p < 0.0001), whereas low proficiency do not (p = 0.64). High proficiency bilingual groups are not statistically different from controls (p = 0.62), contrary to low proficiency bilingual groups (p = 0.01). Interestingly, within each proficiency level, attriters and L2 learners do not differ significantly (p > 0.1).

OVERVIEW OF STUDIES:

STUDY 1: ENGLISH LEXICAL-SEMANTICS

RESULTS RESULTS

DISCUSSION DISCUSSION

P600 (650-1000ms): Controls (Fig. a) show a small but significant P600 in both violations. High-proficiency ‘attriters’ (Fig. b) and low-proficiency attriters (Fig. d) show a large, significant P600, but neither L2 group does (Fig. c and e); instead, low-proficiency L2-learners show prolonged negativity in mismatch condition.

Both proficiency and group membership have impact on P600 (cond x antpost x prof, p < 0.05; cond x group, p < 0.05), for swap and mismatch violations. In ‘swap’ vs. correct , controls and attriters elicit a significant P600 (though different in distribution), whereas L2 learners do not. Collapsing by proficiency, high proficiency speakers showed highly significant P600 effect (p < 0.0001), contrary to low proficiency speakers (p = 0.16).

EH (homograph in English-homograph context) The prisoner violated the parole after a month.

EC (cognate in English-homograph context) The prisoner violated the cabin after a month.

IH (homograph in Italian-homograph context) The poet rhymed the parole of the verse.

IC (cognate in Italian-homograph context) The poet rhymed the cabin of the verse.

CH (homograph in cognate context) The hikers rented the parole in the mountains.

CC (cognate in cognate context) The hikers rented the cabin in the mountains.

(E) (I)

Controls (Fig. a): Violation conditions and correct conditions pattern together. As expected, IC and IH are perceived as violations and elicit a large N400 as well as a small but significant P600.

Attriters (Fig. b) show facilitation in CC compared to EH (reduced N400 and less positive in P600 window).Only IC violation elicits a large N400, not IH (= correct in Italian). In addition to a P600 for IH (shared with controls), attriters elicit a P600 for EH (= incorrect in Italian), unlike controls.

(b) ALL ATTRITERS

L2 learners (Fig. c) resemble English controls in N400 window. They do not show early facilitation for CC but, like attriters, CC is least positive in P600 window. Although IH elicits a native-like N400 unlike in attriters, it also elicits a late positivity, along with EH(as in attriters) and IC.

(c) ALL L2 LEARNERS

= EH (, ) = IH (, ) = IC (, ) = CC (, )

(a) ITALIAN CONTROLS

N400

P600

(b) HIGH L1 PROFICIENCY ATTRITERS

N400

P600

(e) LOWER L2 PROFICIENCY LATE-LEARNERS

N400

(c) HIGH L2 PROFICIENCY LATE-LEARNERS

N400

-Study 1 examined Italian lexical-semantic processing during reading in L1 (for attriters) and in L2 (for L2 learners), compared to native-speakers of Italian still residing in Italy. - Results provide the first ERP evidence of ‘non-native-like’ online lexical-semantic processing in attriters, particularly in those with lower Italian (L1) proficiency scores (i.e., more attrition), and especially for confusable words (‘swap’ condition). - This finding favors the view of ongoing neuroplasticity in adulthood even in one’s L1. - Larger P600 effects in attriters may reflect more elaborated processing (“second thought”, monitoring). - Processing patterns were mediated by proficiency level in both bilingual groups, regardless of AoA - We are currently investigating effects of frequency and cloze probability, and correlation between ERP patterns and behavioral data

Cross-linguistic influence correlates with proficiency:

(1) EH - CC: Larger N400 effect (370-470 ms) in bilinguals with higher Italian proficiency (r = - 0.365, p < 0.01). Difference in N400 amplitude also correlated with individual proficiency measures: C-test (r = -0.316, p < 0.05), error-test (r = - 0.40, p < 0.01) and semantic fluency (r = - 0.291, p < 0.05). Relative proficiency level (Italian – English) also correlated with N400 (r = -0.380, p < 0.01).

(2) IH – EH: Italian semantic verbal fluency correlated with N400 effect at 370-470 ms (r = 0.303, p < 0.05) and 450-550 ms (r = 0.276, p < 0.05). Relative proficiency (Italian-English) correlated positively with the N400 at 370-470 ms (r = 0.275, p < 0.05). Positive correlations suggest that a greater Italian proficiency level was associated with a smaller (less negative) N400 effect for IH, and vice versa.

-Study 2 examined English lexical-semantic processing when reading in L2 (for attriters) and in L1 (for L2 learners), compared to native-speakers of English with no/minimal knowledge of other languages. - Results showed evidence of co-activation of two lexical systems while reading sentences in one language. - This co-activation was strongest in attriters (L1 to L2) but also present in English-Italian bilinguals (L2 to L1), who were not identical to monolingual English controls. - Stronger attrition (i.e., less Italian proficiency) was associated with less cross-linguistic influence in early N400 - Taken together, these 2 studies suggest that advancing L1-attrition and L2-acquisition have something in common – proficiency is key in determining neural correlates of language-processing, in both L1 and L2. - Evidence in favor of L1-attrition and proficiency-mediated patterns of language processing challenges the view of a neurobiological critical period.

(d) LOWER L1 PROFICIENCY ATTRITERS

N400

P600

(a) ENGLISH CONTROLS N400

P600

Knowledge of Italian, either as L1 (attriters) or as L2 (L2-learners) has an impact on how cognates are processed in English compared to monolingual controls, even beyond N400 time window (unexpected finding). In the case of L2-learners, this is evidence of transfer from L2 to L1.

Difference waves for EH – CC

= CONTROLS = ATTRITERS = L2 LEARNERS

N400

P600