+ Intervention effects and the acquisition of Chinese relative clauses Shenai Hu Anna Gavarró Mirta Vernice Maria Teresa Guasti Feb 14th, 2014, Trento + Outline Background and research questions Experiment 1: picture-sentence matching task Experiment 2: character-sentence matching task Discussion + The acquisition of relative clauses (1) il gatto che colpisce il cane Subject RC ‘the cat that hits the dog’ (2) il cane che il gatto colpisce Object RC ‘the dog that the cat hits’ Subject relative clauses (RCs) elicit better performance than object RCs in children speaking a variety of headinitial languages with postnominal RCs (e.g., Catalan, English, French, Hebrew and Italian) + The Relativized Minimality (RM) approach Friedmann, Belletti & Rizzi (2009) claim that the difficulty of object RCs lies in the structural similarity between the A’-moved element (relative head) and the intervening subject. (3) [DP il canei [CP che il gatto [VP colpisce __i ]]] (4) D NP …… D NP …… <D NP> [+R, +NP] [+NP] [+R, +NP] + Chinese is a language which combines SVO order and prenominal relative clauses (Dryer, 2005, 2008) (5) 打小狗的小猫 Subject RC [ _i da xiaogou de ] xiaomaoi hit dog DE cat ‘the cat that hits the dog’ (6) 小猫打的小狗 Object RC [xiaomao da _i de ] xiaogoui cat hit DE dog ‘the dog that the cat hits’ (7) 小猫打小狗。 A declarative sentence Xiaomao da xiaogou. cat hit dog ‘The cat hits the dog.’ + Predictions of the Relativized Minimality (RM) approach (8) subject RC (9) object RC Cheng (1986), Aoun & Li (2003) + Contradictory results of previous studies on Chinese RCs Studies Participants Methods Results Chang,1984 N=48 age 6,7,9,11 act-out task no preference Lee, 1992 N=61 age 4-8 act-out task subject preference Chiu,1996 N=65 age 3-6 act-out task subject preference (older children) object preference (younger ones) Cao, Goodluck & Shan, 2005 N=34 age 4-6 act-out task no preference + Research questions Does the subject/object asymmetry hold for the acquisition of Chinese RCs as predicted by RM? Does this asymmetry emerge during development and how? Does the head-final status of the RC affect its comprehension? + Experiment 1: Picture-sentence matching task Participants 80 children aged from 3;0 to 6;11 10 adults Materials and procedure (10) Subject RC Zhichu [_ da xiaogou de] xiaomao. point to hit dog DE cat ‘Point to the cat that hits the dog.’ (11) Object RC Zhichu [qingwa da _ de] laoshu. point to frog hit DE mouse ‘Point to the mouse that the frog hits.’ 1 2 + Results of Experiment 1 100% 80% 60% 40% 20% 0% subject RCs object RCs 3 73,1% 70,6% 4 80,6% 80% 5 90% 92,5% 6 96,9% 100% + Results of Experiment 1 (cont.) Adults: 100% accurate responses No significant difference between subject RCs and object RCs (χ 2(1) = 0.06, p > .05). Age (categorical variable) yielded a significant effect (χ 2(3) = 48.43, p < .001). + Discussion (1) (12) da xiaogou de xiaomao V O S hit dog DE cat Subject RC (13) xiaomao da de xiaogou S V O cat hit DE dog Object RC (14) Xiaomao da S V cat hit xiaogou. O dog 1 2 + Discussion (2) Results are an artifact of the task. Arnon (2005; 2010) and Adani (2009) propose a character-sentence matching task as an alternative to the picture-sentence matching task. + Experiment 2: Character-sentence matching task Participants 120 children aged from 3;0 to 8;11 20 adults Materials and procedure (15) Subject RC Na yi-ge shi [_ da xiaogou de] xiaomao? which one-CL is hit dog DE cat ‘Which one is the cat that hits the dog?’ (16) Object RC Na yi-ge shi [qingwa da _ de] laoshu? which one-CL is frog hit DE mouse ‘Which one is the mouse that the frog hits?’ + Results of Experiment 2 100% 80% 60% 40% 20% 0% 3 Subject RCs 47,8% Object RCs 24,8% 4 61,3% 23,1% 5 72,5% 20,6% 6 76,3% 45% 7 99,4% 45,6% 8 100% 95,6% + Results of Experiment 2 (cont.) Adults: 100% accurate responses A significant difference between subject RCs and object RCs (χ 2(1) = 50.41, p < .001; Wald Z = 16.88, p < .001) A significant effect of Age (e.g., between age three and age six (χ 2(5) = 145.78, p < .001; Wald Z = 4.08, p < .001) Error Types Which one is the cat that hits the dog? Correct Embedded Error Other Error Reversal Error Embedded error: wrong character in the right picture Reversal Error: right character in the wrong picture Other Error: wrong character in the wrong picture + Analysis of errors in subject RCs (1) 100% 80% 60% 40% 20% 0% 3 Correct 47,8% Embedded Error 19% Reversal Error 15,3% Other Error 17,8% 4 61,3% 15% 10% 13,7% 5 72,5% 8,7% 7,5% 11,3% 6 76% 11,8% 3% 8,8% 7 99,4% 0,6% 0% 0% 8 100% 0% 0% 0% + Analysis of errors in subject RCs (2) Subject RCs in Chinese were also difficult to comprehend up to six years of age and elicited a variety of errors. Children did not show a tendency to make a specific mistake (all p > .21). + Analysis of errors in object RCs (1) 100% 80% 60% 40% 20% 0% 3 Correct 24,8% Embedded Error 37% Reversal Error 22,9% Other Error 15,3% 4 23,1% 47,5% 23,1% 6,3% 5 20,6% 73,1% 3,8% 2,5% 6 45% 53,1% 1,3% 0,6% 7 45,6% 53,1% 1,3% 0% 8 95,6% 4,4% 0% 0% + Analysis of errors in object RCs (2) The pattern of errors in Chinese children diverges from that found in languages with postnominal RCs. The ‘Agent’ interpretation error (Embedded Error) is the most common in Chinese, instead of the ‘reversal’ error. + Discussion (1) (17) The subject/object asymmetry is found in Chinese as in languages with postnominal RCs Subject RCs are comprehended at an earlier age than object RCs. This finding is in line with RM approach’s predictions. Structural intervention + Discussion (2): subject RCs Up to six years of age children show problems All error types are equally distributed The errors reflect different sources of difficulty (Arnon, 2005) Linear intervention is also taxing for Chinese children (18) the cat [that _ hits the dog] NP V NP (19) [ _ da xiaogou de] xiaomao V NP NP + Discussion (3): object RCs The pattern of error differs from that of other languages: the ‘agent’ interpretation error is the most common in Chinese. Children choose the character which is the correct one for the simple declarative sentence. (20) Na yi-ge shi [xiaomao da _ de] xiaogou? which one-CL is cat hit DE dog wh-one-CL V NP V NP ‘Which one is the dog that the cat hits? (21) Na yi-ge shi xiaomao? which one-CL is cat Which one is the cat? + Conclusions The subject/object asymmetry holds for the acquisition of Chinese RCs, consistently with acquisition studies in other languages. The results of the picture-sentence matching task (Experiment 1) are biased, due to methodological issues. The character-sentence matching task (Experiment 2) is a reliable task. To sum up, structural intervention plays a more important role than linear intervention in Chinese RC comprehension. + GRAZIE! o All the participants in our experiments o Flavia Adani, Fabrizio Arosio, Carlo Cecchetto, HU Shuangling, ZHANG Meilian, and, for the materials, Candice Coyer, Anamaria Bentea & Stephanie Durrleman. o Project FFI2011-29440-C03-03