A User-Based Approach to Spanish-Speaking L2 Acquisition of Chinese Applicative Operation Nana Y-H Huang Cambridge University yhh28@cam.ac.uk directionality (e.g., vender ‘sell’ and alquilar 1 Introduction: Low Applicative Opera- ‘rent’) and verbs of motion (e.g., lanzar ‘throw’ tions and pataer ‘kick’), the applied argument would be ambiguous between a goal and a source. Recent studies of argument structure distin- Cuervo provides such an example as (3). guishes non-core (applied) arguments from core arguments in the sense that non-core ones do not (3) Valeria le vendió el auto a su hermano. belong to the basic argument structure of verbs Valeria CL sold the car DAT her brother and that they enter argument structures through Applicative Operations (AO) introduced by func- 1. Valeria sold the/her car to her brother. tional heads such as Low Applicative-source 2. Valeria sold her brother’s car. (LA-source) or Low Applicative-goal (LA-goal) heads (Pylkkänen, 2000; 2002; 2008; Cuervo, 1.2 Applicative operations in Chinese 2003). Because languages make use of different applicative heads, in this study, I examine the In Chinese, AO is as productive; nevertheless, acquisition of Chinese AO by Spanish-speaking unlike Spanish, Chinese only allows LA-source L2 learners and propose a usage-based approach (see (4)) but not LA-goal (see (5)): for the results collected from a comprehension (4) Zhangsan tou-le Lili liang tai diannao. task and an acceptability judgment task. Zhangsan steal-PERF Lili two CL computer 1.1 Applicative Operations in Spanish ‘Zhangsan stole Lili of two computers.’ Cuervo (2003) reports that in Spanish a predicate which expresses the transfer of a theme to a goal, (5) *Zhangsan sheji-le Lili liang jian qunzi. such as verbs indicating creation (e.g. cocinar Zhangsan design-PERF Lili two CL skirt ‘cook/bake’, construer ‘build’, and etc.), allows LA-goal, where the applied argument is the da- ‘Zhangsan designed Lili two skirts.’ tive argument, as in (1). 1.3 Research Questions (1) Valeria le diseñó una pollera a Anna. This study examines Spanish L2ers’ acquisition Valeria CL designed a skirt DAT Anna of Chinese AO and considers the learnability Lit.: ‘Valeria designed Anna a skirt.’ problem posed by the superset-subset relation between Spanish and Chinese on this structure A Spanish applied argument can also appear in (i.e. Spanish allows both LA-goal and LA-source the environment of a transfer predicate with ‘re- while Chinese allows only LA-source). We pre- verse directionality’, such as robar ‘steal’, sacar dict learners to wrongly transfer LA-goal, which ‘take from’, and extraer ‘take out from’. In this is allowed in L1 Spanish, to L2 Chinese despite case the applied argument is understood as the the lack of positive evidence for the use of LA- possessive source of the theme object. goal in L2 input. Furthermore, due to lack of (2) Pablo le robó la bicicleta a Anna. negative evidence (from the fact that AO do not appear in pedagogical textbooks nor in class- Pablo CL stole the bike DAT Anna rooms designed for L2ers), L2 Chinese input Lit.: ‘Pablo stole Anna the bike.’ lacks information regarding ungrammaticality of The source argument appears in dative case LA-goal, which would be necessary for L2ers to which has the same morphosyntactic properties rule out incorrect hypotheses. That is, these of a recipient argument; therefore, it is predicted learners are expected to show overgeneralization that in the context of verbs with underspecified from early on till even at the advanced level. Copyright © by the paper’s authors. Copying permitted for private and academic purposes. In Vito Pirrelli, Claudia Marzi, Marcello Ferro (eds.): Word Structure and Word Usage. Proceedings of the NetWordS Final Conference, Pisa, March 30-April 1, 2015, published at http://ceur-ws.org 124 2 Methods was provided on the side which learners could choose if they were unsure of the response. See To test our prediction on L1 transfer effects we Appendix D. designed two tasks to probe different knowledge of L2 structures: one being implicit and mean- 2.2 Participants ing–focused; the other being explicit and form- 20 L2ers and 10 natives speakers (NS) of Chi- focused. nese serving as a control group participated in 2.1 Materials and Procedures this study. All NS were graduate students born and raised in Taiwan. Most L2ers were under- An Animation Matching Task (AMT) was used graduate students with the exception of 3 people to probe L2er’s implicit knowledge because it being Catholic priests. L2ers had learned Chi- called for a focus on meaning. The AMT includ- nese in Taiwan for at least 3 years and came ed 12 items (6 test sentences and 6 fillers). The 6 from different Spanish-speaking countries. Span- test sentences included verbs underspecified for ish was the native language for all L2ers. English directionality of transfer. The 6 fillers bore only was the second most proficient language. surface similarity and served to distract partici- Before the study, L2ers had completed a 40-item pants’ focus in different ways. 2 contained syn- Chinese proficiency cloze test developed by Yu- tactically unacceptable sentences; another 2 con- an (2014). Based on the scores, they were divid- tained sentences that matched both animations; ed into Advanced (AD) and Intermediate (IN) the other 2 contained sentences that matched nei- group. Table 1 summarizes the participants’ ther of the two animations. See Appendix A. background information and cloze test scores. On each trial, the L2ers first saw 2 animations on the computer screen. Next, they heard the tar- Group NS AD IN get sentence presented auditorily. Participants were required to match the sentence to the cor- Number of 10 10 10 rect animation. For example, participants (6) Zhansan reng-le Lisi yi jian waitao. Mean age 26.2 26.9 24.1 Zhangsan toss-PERF Lisi one CL coat (ranges in (22-28) (23-38) (20-36) Lit: ‘Zhangsan tossed Lisi one coat.’ brackets) The sentence was preceded by two animations: Duration NA 8.4 5.7 (a) Zhangsan tossed one coat to Lisi; (b) Zhang- (years) of san tossed one of Lisi’s coats away. Participants chose which animation was a better match for the formal sentence by ticking the answer on the answer instruction sheet. They were told at the beginning of the test Length NA 5.7 4.8 that if they found both animations matching the sentence, they could select both. If they found (years) of (3-11) (3-9) neither matching the sentence or if they could not residence in understand the sentence, they could choose Taiwan ‘don’t know’ option on the side and choose/state the reason. See Appendix B. Cloze test 39 35 29 Following the AMT was the Acceptability score (38-40) (33-37) (27-32) Judgment Task (AJT), which tapped participants’ explicit knowledge on forms. 2 different types of (ranges in verbs that induced opposite directionality of brackets) transfer (i.e., grammatical LA-source and un- grammatical LA-goal) were included, 3 items per Table 1: Participants’ Background Information type. In addition, with 6 control sentences and 6 fillers, the AJT contained 18 items in total, half 3 Results and discussion grammatical and half ungrammatical. Please see Table 2 presents the percentage of how often par- Appendix C. Rating scale ranged from very un- ticipants chose a certain animation in the AMT acceptable (1), unacceptable (2), acceptable (3), (for example, the (a) condition in example (6) to very acceptable (4). A ‘don’t know’ option above depicts a Goal condition). 125 It is, therefore, proposed that subjects’ experi- Group Source *Goal *Both Don’t know ence in L1 to resort to context in the face of am- biguity caused by verbs underspecified for direc- NS 100 0 0 0 tionality helps advanced L2ers overcome over- generalization. The sensitivity trained in L1 is AD 57 10 33 0 transferred to L2 learning and displayed in that IN 23 17 57 3 more attention is paid to the co-occurring applied argument in the face of ambiguous thematic role Table 2: Percentages of choice in the AMT assigned to applied argument. Advanced L2ers might have accumulated enough indirect statisti- A 2-sample z-test was performed separately to cal information (Reali and Christiansen, 2005) compare proportions between any 2 among the 3 tracked from co-occurrences of recurring se- groups. The results showed that any 2 groups quences of words before being able to overcome were significantly different from each other in overgeneralization. This finding suggests that the the choice for Source and for Both, but not sig- effects of L1 transfer result not only from the nificantly different in Goal. IN group as expected similarity and/or difference of linguistic facts showed overgeneralization in wrongly choosing between the native and the target language, but Both, while AD group seemed to be able to over- also from L2ers’ experience gained in their na- come overgeneralization and limit the construc- tive language. tion of Chinese AO to LA-source from the fact that the choice for Both was greatly decreased References and that for Source was greatly increased at the C. Cuervo. 2003. Datives at Large. PhD Thesis. MIT. higher proficiency level. As for the AJT, Table 3 presents the mean C. J. Huang. 2007. Hanyu dongci de tiyuan jiegou yu scores with the standard deviation in the brackets qi jufa biaoxian (The thematic structures of verbs in Chinese and their syntactic projection). Yuyan of each group by verb types. Using an alpha level Kexue (Linguistic Sciences) 6(4): 3-21. of 0.05, paired t-tests showed that only NS ex- hibited significant difference in the responses to B. MacWhinney. 2004. A multiple process solution to 2 types of verbs, while L2 groups did not. the logical problem of language acquisition. Jour- nal of Child Language, 31, 833-914. Group Verb type L. Pylkkänen. 2000. What applicative heads apply to. Consumption Creation In M. Minnick, A. Williams, and E. Kaiser (eds.), NS 3.53(0.39) 1.36(0.24) working papers in Proceedings of the 24th Annual AD 3.22(0.54) 3.33(0.44) Penn Linguistics Colloquium 7(1). IN 3(0.34) 3.23(0.38) L. Pylkkänen. 2002. Introducing Arguments. PhD Thesis. MIT. Table 3: Mean scores for the AJT L. Pylkkänen. 2008. Introducing Arguments. Cam- In contrast with the result in Table 2, AD group bridge, MA: MIT Press. did not perform better in AJT than IN group in F. Reali and M. Christiansen. 2005. Uncovering the rejecting ungrammatical AO-Goal introduced by richness of the stimulus: Structure dependence and verbs of creation. The question is how we can indirect statistical evidence. Cognitive Science, 29, explain for AD group’s inconsistency in over- 1007-1028. coming overgeneralization. T. Q. Sun and Y. F. Li. 2010. Hanyu fei hexin Notice that the major difference between the 2 lunyuan yunzhun jiegou chu tan (Licensing non- tasks is whether the verb specifies directionality core arguments in Chinese), Zhongguo Yuwen of transfer. Verbs included in the AMT are the (Studies of the Chinese Language) 334: 21-33. verbs that do not favor a particular direction of B. Yuan. 2010. Domain-wide or variable-dependent transfer and therefore the introduced applied ar- vulnerability of the semantic-syntax interface in L2 gument is inherently ambiguous between Goal acquisition? Evidence from wh-words used as exis- and Source in the L1 Spanish. In other words, the tential polarity words in L2 Chinese grammars. verbs that trigger ambiguity in L1 Spanish are Second Language Research 26: 219-60. where subjects first overcome overgeneraliza- tion. 126 Appendix A: Test Sentences in the AMT Type of Verbs Item Question Target Sentence Number Test sentences na2 ‘take’ 1 小明拿了小華一本雜誌 ban1 ‘carry’ 6 小張搬了小李一張桌子 reng1 ‘toss’ 9 張三扔了李四一件外套 tou1 ‘steal’ 3 阿明偷了阿華兩瓶可樂 mai4 ‘buy’ 8 張三買了李四一支毛筆 ying2 ‘win’ 12 小張贏了小李一隻手錶 Fillers sha1 ‘kill’ 2 小明殺了小華兩頭小羊 (matches both) gei3 ‘give’ 7 老李給了老張一隻小鳥 (matches both) dao3 ‘collapse’ 10 老王倒了小李一棵小樹 (ungrammatical) gei3 ‘give’ 4 老李關了老張一隻小鳥 (ungrammatical) song4‘give’ 5 小華送了小李兩瓶可樂 (matches neither) jiao1 ‘teach’ 11 張三教了瑪莉兩題數學 (matches neither) Appendix B: Sample Answer Sheet of the AMT Question Which Animation do If you tick ‘I don’t know’, please tick or state the Number you choose? reason 1. A I don’t Neither of the two animations is correct. B know I do not understand the sentence that I heard. Other reason _______________ Appendix C: Test sentences in the AJT Type of Verbs Item Question Target Sentence Number Verbs of Consumption chi1 ‘eat’ 1 李四吃了張三兩個蛋糕 he1 ‘drink’ 8 小華喝了小明兩瓶紅酒 yong4 ‘use’ 17 小李用了小張一支鉛筆 Verbs of Creation kao3 ‘bake’ 6 *阿華烤了小明一個蛋糕 zhu3 ‘cook’ 12 *小李煮了老張一頓晚餐 zao4 ‘build’ 14 *張三造了老李一棟房子 Control Sentences chi1 ‘eat’ 2 李四吃了兩個蛋糕 he1 ‘drink’ 9 小華喝了兩瓶紅酒 yong4 ‘use’ 13 小李用了一支鉛筆 kao3 ‘bake’ 4 阿華烤了一個蛋糕 zhu3 ‘cook’ 11 小李煮了一頓晚餐 zao4 ‘build’ 16 張三造了一棟房子 Fillers gei3 ‘give’ 3 老李給了老張一隻小鳥 song4 ‘give’ 7 小華送了小李兩瓶可樂 jiao1 ‘teach’ 15 張三教了瑪莉兩題數學 gei3 ‘give’ 5 *老李給了隔壁老張 song4 ‘give’ 10 *小華送了鄰居小李 jiao1 ‘teach’ 18 *張三教了同學瑪莉 Appendix D: Sample Answer Sheet of the AJT Very Unacceptable Acceptable Very Unacceptable Acceptable 1. 阿明吃了我兩個蛋糕。 1 2 3 4 I don’t know 127