Eviden ence for F Fea eature e Reassem embly: Gen ender er a - PowerPoint PPT Presentation

Eviden ence for F Fea eature e Reassem embly: Gen ender er a and number f fea eatu tures i in L2 Russian Mila Tasseva-Kurktchieva Angelina Rubina tassevak@mailbox.sc.edu arubina@email.sc.edu AAAL 2019

Goals • To assess whether L2 learners of Russian can acquire both • A feature that is present in their L1 (English), namely [number], and • A feature which is not evident in their L1, namely [gender] • To assess whether extra-grammatical properties have influence on the acquisition of grammatical features AAAL 2019 Goals Feature reassembly Methods Results Conclusion 2

Feature Reassembly (Lardiere, 2009) • L2 acquisition is more than what we commonly call parameter resetting • Feature interpretability (Chomsky, 1995; White et al, 2004) • In terpretable features = head features • Un interpretable features = relational features (i.e., need checking and interpretation at Spell-Out)  Uninterpretable features, especially those not present in L1, will be un-acquireable (Hawkins and Chan, 1997)  But what if both languages have the same feature realized in syntax? AAAL 2019 Goals Feature reassembly Methods Results Conclusion 3

Feature Reassembly (Lardiere, 2009) • The feature(s) could be exactly the same  should not cause L2A problems 1) (in) definiteness in English and Spanish English Spanish Sg Pl Sg Pl indefinite a Ø el/la Ø definite the the les les • The feature(s) could have different compositionality  should cause (some) L2A problems 2) (in)definiteness in French and English French English Sg Pl Sg Pl indefinite un/une des a/an Ø definite le/la les the the AAAL 2019 Goals Feature reassembly Methods Results Conclusion 4

Feature Reassembly (extension) • Slabakova (2008), Cho & Slabakova (2014): difficulty continuum • Features which require mapping of L1 morphology to L2 morphology of equivalent compositionality are easiest to acquire ( [±definiteness] in English and Spanish) • Features which map L1 to L2 morphology but require compositional reassembly are more difficult ( [±definiteness] in English and French) • Features which map a morphologically covert property in one language that is set in the discourse onto a morphologically overt material in the other language are the most difficult ( [aspect] in English and Bulgarian) 3) Sue baked cakes for 10 years before she 5) Sue peče torti v prodəlʒenie na 10 godini. became a doctor. 4) Sue baked 3 cakes yesterday. 6) Sue iz peče 3 torti včera . AAAL 2019 Goals Feature reassembly Methods Results Conclusion 5

Implications and predictions • Gender and number agreement will cause different problems for native speakers of English acquiring L2 Russian • Acquisition of Russian [NUMBER] requires mapping of L1 morphology to L2 morphology plus feature reassembly • Acquisition of Russian [GENDER] requires L1 context to L2 morphology mapping Easier to acquire Harder to acquire Number agreement Gender agreement • There will be delays in the RTs on the more difficult [GENDER] feature • There will be additional frequency effect within each feature Goals Feature reassembly Methods Results Conclusion AAAL 2019 6

The languages and features English​ Russian​ Singular​ nice book​ krasiv- AJA FEM.SG knig- А FEM.SG nice phone​ krasiv- YI MASC.SG telefon- Ø MASC.SG nice mirror​ krasiv- О E NEUT.SG zerkal- О NEUT.SG Plural​ nice book- S krasiv- YE FEM.PL knig- I FEM.PL nice phone- S krasiv- YE MASC.PL telefon- Y MASC.PL nice mirror- S krasiv- YE NEUT.PL zerkal- A NEUT.PL Goals Feature reassembly Methods Results Conclusion AAAL 2019 7

The cline of difficulty (Slabakova 2008, 2009) Easier Harder to acquire to acquire L1 F morpheme L1 F context L1 F context L2 F morpheme, +reassembly L2 F morpheme , +reassembly L2 F context , +reassembly [gender] on Fem and [number] on Nouns [gender] on Masc Nouns Neut Nouns and Adj’s AAAL 2019 Goals Feature reassembly Methods Results Conclusion 8

Methodology • Participants: 21 second semester learners of Russian as L2 • 10 participants in SET 1 • 11 participants in SET 2 • Materials: • A pool of concrete nouns and adjective from 1 st semester Russian curriculum • Canonical [gender] and [number] affixes • Nominative case • Form of the NPs: Adj [gender, number, NOM] —N [gender, number, NOM] AAAL 2019 Goals Feature reassembly Methods Results Conclusion 9

Methodology • Tests: • SET 1: give the features on A  expect participants to retrieve them on N • Forced choice comprehension task • Fill in the blanks task • SET 2: give the features on the N  expect participants to retrieve them on A • Grammaticality judgement task • Forced choice production task • Background questionnaire AAAL 2019 Goals Feature reassembly Methods Results Conclusion 10

SET 1: Forced choice comprehension task  30 target trials  6 conditions, 5 trials per condition  20 fillers  4 practice trials Number Gender Target competitor competitor (MascSg) (FemPl) (FemSg) У меня есть желтый ___ из Италии. ‘I have a yellow MASCSG ____ from Italy. ’ AAAL 2019 Goals Feature reassembly Methods Results Conclusion 11

SET 2: GJT • 6 feature conditions: MascSg, MascPl, FemSg, FemPl, NeutSg, NeutPl • 2 grammaticality conditions • Grammatical: У меня в шкафу зеленая рубашка. 7) U menja v škafu zeljon aja rubašk a to me in wardrobe green FEMSG shirt FEMSG ‘In my wardrobe there is a green shirt.’ • Ungrammatical: Вот зеленая дерево на картинке. 8) Vot zeljonaja derev o na kartinke here’s green FEMSG tree NEUTSG in picture ‘Here is a green tree in the picture.’ AAAL 2019 Goals Feature reassembly Methods Results Conclusion 12

Results: SET 1, Forced choice comprehension SET1: Accuracy: correct, gender competitor, number competitor 100% 80% 60% 47% SET 1: RTs of correct, gender 40% 31% competitor, number competitor 21% 20% selections 0% 14000 13000 correct gender competitor number competitor 12000 11000 10000 correct gender number competitor competitor AAAL 2019 Goals Feature reassembly Methods Results Conclusion 13

Results: SET 1, Forced choice comprehension SET 1: Accuracy by condition singular plural 100% 80% SET 1: RTs by condition 54% 54% 60% 52% 52% 50% 14000 13250 38% 40% 13000 12131 20% 12986 12000 11237 0% 11000 Masc Fem Neut 11398 10694 10000 Masc Fem Neut singular 13250 12131 10694 plural 12986 11398 11237 AAAL 2019 Goals Feature reassembly Methods Results Conclusion 14

Results: SET 1, Forced choice comprehension SET 1: selection of gender vs. number competitor 100% 90% 80% 70% 54% 52% 60% 48% SET 1: RTs on correct, gender competitor, 50% 38% 40% number competitor selections 26% 26% 24% 30% 18% 14% 15000 20% 10% 14040 14000 0% 13702 Masc Fem Neut 13250 13000 correct 38% 54% 52% 12389 12131 12000 11895 gender competitor 14% 26% 24% 11815 11093 number competitor 48% 18% 26% 11000 10694 10000 correct gender competitor number competitor AAAL 2019 Goals Feature reassembly Methods Results Conclusion 15

Results: SET 2, Grammaticality judgement SET 2: correct responses singular vs. plural by gender singular plural 100% SET 2: RTs singular vs. plural by gender 80% singular plural 65% 65% 59% 58% 55% 11948 14000 60% 52% 10886 10458 12000 40% 10000 11442 10896 20% 8000 9073 6000 0% 4000 Masc Fem Neut 2000 0 Masc Fem Neut AAAL 2019 Goals Feature reassembly Methods Results Conclusion 16

Results: SET 2, Grammaticality judgement SET 2: grammatical vs. ungrammatical by gender 100% SET 2: RTs of grammatical vs. ungrammatical 80% by gender 60% grammatical ungrammatical 40% 12142 14000 12534 10761 20% 12000 10000 0% Masc Fem Neut 10603 10731 8483 8000 grammatical 61% 67% 68% 6000 ungrammatical 56% 53% 48% 4000 2000 0 Masc Fem Neut AAAL 2019 Goals Feature reassembly Methods Results Conclusion 17

Results: SET 2, Grammaticality judgement SET 2: Accuracy grammatical vs. ungrammatical by number grammatical ungrammatical SET 2: RTs grammatical vs. 100% 80% ungrammatical by number 80% 67% grammatical ungrammatical 60% 15000 12117 11063 65% 40% 60% 10000 20% 10283 9494 0% 5000 singular plural 0 singular plural AAAL 2019 Goals Feature reassembly Methods Results Conclusion 18

Results: SET 2, Grammaticality judgement SET 2: grammatical vs. ungrammatical by SET 2: RTs grammatical vs. ungrammatical condition by condition grammatical ungrammatical 100% 12602 13384 15000 36% 48% 11973 10751 55% 80% 61% 58% 10430 58% 9748 60% 10000 40% 7901 10263 67% 82% 10479 10432 10162 64% 70% 58% 52% 10067 20% 5000 0% MascSg MascPl FemSg FemPl NeutSg NeutPl 0 grammatical ungrammatical MascSg MascPl FemSg FemPl NeutSg NeutPl AAAL 2019 Goals Feature reassembly Methods Results Conclusion 19

Conclusions • What we see in the two comprehension tasks supports the Feature Reassembly hypothesis: • They produce better results and are faster on the [number] than the [gender] feature across the board • The gender feature on the head N is easier to comprehend than on the agreeing Adj • Masc is produces the worst results in both singular and plural AAAL 2019 Goals Feature reassembly Methods Results Conclusion 20