2018 Friday Session C 1400
Friday, November 2, 2018 | Session C, Terrace Lounge | 2pm
Learnability in Romance: How Indirect Input Helps Children Acquire the Contrast Between Null and Overt Subjects
H. Forsythe, C. Schmitt, D. Greeson
Background: In consistent null subject languages (Holmberg 2005, Barbosa 2009,2011a,b), the choice between an overt or null pronoun depends on the discourse status of the antecedent. This choice can therefore guide the processing and interpretation of ambiguous pronominal subjects in languages like Spanish (Alonso-Ovalle etal.2002, Carminati 2002, Keating etal.2016). In (1), for example, both pronouns are grammatical, but the null variant is more likely to refer to the preceding subject (Juan, a “same-reference” interpretation), while the overt variant more likely triggers a non-subject reading (Pedro or another person, a “switch-reference” interpretation).
Here, we ask how children learn to use the null/overt distinction to determine pronoun reference.
(1) Juan le pega a Pedro y después {ø/él} se va. P(ø=Juan) > P(él=Juan) Juan hits Pedro and then {ø/he} leaves.
The learning problem is not trivial. In order to use this cue to help identify antecedents, children must track the association between null/overt pronominal subjects and same-/switch- reference contexts, respectively. However, to do that, children must first know how to identify same- and switch-reference contexts—that is, they must already know what the antecedent is.
We hypothesize that children solve this problem by first examining the distribution of null and overt 1st-and 2nd-person pronouns, which have predetermined antecedents (and which do not alternate with DPs). We present two studies: (i) a corpus study showing that the null/overt contrast is in principle acquirable from the input distribution of 1st-and 2nd-person subject pronouns alone, and that children replicate this contrast in their own production; and (ii) a pronoun resolution experiment showing that children can use the contrast to interpret grammatically ambiguous 3rd person pronouns by around the same age.
Corpus Methods: 5,591 spontaneous utterances by 2 mother-child dyads were extracted from the Schmitt corpus of Mexico-City Spanish. We extracted all animate, pronominal subjects preceded by at least one other clause in the same utterance, coding for form (null, overt) and reference (same, switch). Results (Fig.1): Mothers produced more overt subjects in switch- reference (19.6%) than same-reference contexts (6.7%). However, the difference was only significant within the 1st&2nd-person, not the 3rd-person. Children mirror the adult distribution.
Experiment Methods (Fig.2): Using forced-choice picture selection, we tested interpretations of ambiguous null and overt subjects with 40 adults and 76 children from Mexico City. Results (Fig.3): Adults associated null subjects to same-reference interpretations significantly more often than overt subjects (F1(1,38)=11.6,p<0.002, F2(1,5)=17.5,p<0.009).
Children over 4;6 showed the same difference (F1(1,28)=10.0, p<0.005, F2(1,1)=178.7,p<0.05), though they produced fewer same-reference responses overall. Children 4;6 and under were not sensitive to the contrast.
Conclusion: Beginning around 4½, we observe children using the null/overt contrast to help interpret grammatically ambiguous 3rd person pronouns—much earlier than previously reported (Shin and Cairns 2012). Given the structure of the input, however, it is unlikely that they could have acquired this knowledge by tracking the form and reference of 3rd-person subject pronouns alone. We therefore argue that children track 1st-and-2nd-person pronouns, a claim that is supported by their adult-like production of these pronouns at around the same age.