Assessing the noun counting strategy postulated by the syntactic bootstrapping approach in the input to Spanish-learning children Cynthia Pamela Audisio and Alejandrina Cristia Observing the world-situation contingent to an occurrence of a verb does not

CRISTIA, ALEJANDRINA; AUDISIO, CYNTHIA PAMELA

Nijmegen

Congreso; Many Paths to Language (MPaL); 2017

Max Planck Institute for Psycholinguistics

Assessing the noun counting strategy postulated by thesyntactic bootstrapping approach in the input toSpanish-learning childrenCynthia Pamela Audisio and Alejandrina CristiaObserving the world-situation contingent to an occurrence of a verb does not suffice to learnits meaning. Instead, syntactic bootstrapping theories claim that young children?s interpreta-tion of the meaning of verbs are guided by their knowledge of the relationship between syntacticand semantic structures (Gleitman, 1990; Landau & Gleitman, 1985). That is, they infer whata verb means aided by an innate bias to map nouns in sentences onto participants in events in aone-to-one fashion (Yuan et al., 2012). For instance, transitive verbs occurring with two nounsare interpreted as events that involve some kind of relation between two participants. Lidz etal. (2003) and Lee & Naigles (2008) conducted a series of experiments to test this hypothesis onchildren learning languages that allow argument omission, Kannada and Mandarin respectively.The purpose of this study is to investigate whether the mapping bias postulated by syntacticbootstrapping theories would lead Spanish-learning infants to an accurate classification of theverbs in their input. Real-world utterances in children?s linguistic input tend to be morecomplex than made-up sentences included in experimental designs. In addition, Spanish is arather inflectional language that also allows argument omission whenever its referents can beretrieved from the extralinguistic situation or previous discourse (e.g. No toques). Furthermore,we aim to assess whether the noun counting strategy is equally successful across verb classes(intransitive, transitive and ditransitive verbs).Argentinean infants (N = 10 whose parents had ¿ 16 years education; N= 10 whose parentshad < 7 years of education) were followed up longitudinally (Rosemberg et al., 2015-2016). Their input was initially recordedat mean age 0; 14 using an unobtrusive recorder carried in a breast pocket for 4 hours, ofwhich the middle 2 have been fully transcribed. For this analysis, we excluded utteranceswith more than one verb. The remaining utterances were parsed following two strategiesregarding nominals: either all nominals in the utterances were tagged as such or exclusivelythose included in previous experimental studies (i.e. common and proper nouns and pronounsin the nominative, accusative and prepositional case). The noun counting strategy has beenexperimentally confirmed only in sentences including the latter. Both parsing strategies couldalso include or exclude verb inflections for person in the nominal count. In addition, every verbin the corpus was classified as intransitive, transitive or ditransitive by consulting a number ofonline databases (e.g. Rojo, 2001) based on Spanish corpora. Furthermore, we stated whetherthe verb in question had causative meaning by assessing whether it occured with an agent anda patient (as opposite to non causative verbs which occured with either an agent or a patient).Results showed striking differences in the proportion of utterances accurately classified (i.e.where the number of nouns and the verb type in the utterance matched), principally betweenstrategies with and without verb inflections. The fact that Spanish allows argument droppingaccounts -in part- for the results. Other phenomena such as lengthy phrases with several nom-inals -e.g. whole clauses functioning as verbs? arguments- might have influenced the resultsas well and stand as a challenge to inexperienced learners. Interestingly, we found differencesin the proportion of verbs correctly classified by noun counts across verb classes. This mightexplained why some verb classes are acquired by children earlier than others. Corpus studieslike the one presented here are valuable as they complement experimental studies with evidenceof real linguistic data which might reveal some limitations in the mechanisms experimentallyassessed.References[1] L. Gleitman, ?The structural sources of verb meanings,? Language acquisition, vol. 1, no. 1,pp. 3?55, 1990.[2] B. Landau and L. R. Gleitman, Language and experience: Evidence from the blind child.Harvard University Press, 1985.[3] J. N. Lee and L. R. Naigles, ?Mandarin learners use syntactic bootstrapping in verb acqui-sition,? Cognition, vol. 106, no. 2, pp. 1028?1037, 2008.[4] J. Lidz, H. Gleitman, and L. Gleitman, ?Understanding how input matters: Verb learningand the footprint of universal grammar,? Cognition, vol. 87, no. 3, pp. 151?178, 2003.[5] G. Rojo, ?a explotaci ́on de la base de datos sint ́acticos del espanol actual (bds),? inLing ̈u ́ıstica con corpus. Catorce aplicaciones sobre el espanol (J. De Kock, ed.), pp. 225?286,Univ. de Salamanca, 2001.[6] Rosemberg, C.R, Alam, F., Stein, A. Migdalek, M., Menti, A. y Ojea, G. (2015-2016). Language Environments of Young Argentinean Children. CONICET (DOI in progress).[7] S. Yuan, C. Fisher, and J. Snedeker, ?Counting the nouns: Simple structural cues to verbmeaning,? Child Development, vol. 83, no. 4, pp. 1382?1399, 2012.