Testing a computational model of causative overgeneralizations: Child judgment and production data from English, Hebrew, Hindi, Japanese and K'iche'.

Ben Ambridge ,Stewart M Mccauley ,Inbal Arnon ,Shira Zicherman ,Tomoko Tatsumi ,Dani Bekman ,Rukmini Bhaya Nair ,Mario Marroquín Pelíz ,Dipti Misra Sharma ,Margarita Julajuj Mendoza ,Sindy Fabiola Can Pixabaj ,Ramya Maitreyee ,Clifton Pye ,Soumitra Samanta ,Ayuno Kawakami ,Seth Campbell ,Ruth A Berman ,Amir Efrati ,Laura Doherty ,Bhuvana Narasimhan ,Pedro Mateo Pedro ,Amy Bidgood ,Kumiko Fukumura

doi:10.12688/openreseurope.13008.2

Abstract

How do language learners avoid the production of verb argument structure overgeneralization errors ( *The clown laughed the man c.f. The clown made the man laugh), while retaining the ability to apply such generalizations productively when appropriate? This question has long been seen as one that is both particularly central to acquisition research and particularly challenging. Focussing on causative overgeneralization errors of this type, a previous study reported a computational model that learns, on the basis of corpus data and human-derived verb-semantic-feature ratings, to predict adults' by-verb preferences for less- versus more-transparent causative forms (e.g., * The clown laughed the man vs The clown made the man laugh) across English, Hebrew, Hindi, Japanese and K'iche Mayan. Here, we tested the ability of this model (and an expanded version with multiple hidden layers) to explain binary grammaticality judgment data from children aged 4;0-5;0, and elicited-production data from children aged 4;0-5;0 and 5;6-6;6 ( N=48 per language). In general, the model successfully simulated both children's judgment and production data, with correlations of r=0.5-0.6 and r=0.75-0.85, respectively, and also generalized to unseen verbs. Importantly, learners of all five languages showed some evidence of making the types of overgeneralization errors - in both judgments and production - previously observed in naturalistic studies of English (e.g., *I'm dancing it). Together with previous findings, the present study demonstrates that a simple learning model can explain (a) adults' continuous judgment data, (b) children's binary judgment data and (c) children's production data (with no training of these datasets), and therefore constitutes a plausible mechanistic account of the acquisition of verbs' argument structure restrictions.

Highlights

The question of how language learners come to avoid verb argument structure overgeneralization errors such as *The clown laughed the man – in some cases after a protracted period of producing them – has been described as a “learnability paradox” (Pinker, 1989: 415); “one of the most...difficult challenges for all students of language acquisition” (Bowerman, 1988: 73)
Causative forms (e.g., *The clown laughed the man vs The clown made the man laugh) across English, Hebrew, Hindi, Japanese and K’iche Mayan
Binary grammaticality judgments (4;0-5;0) Data from the binary judgment task show that, with the apparent exception of Japanese, children aged 4;0-5;0 are capable of providing meaningful grammatical acceptability judgments for sentences containing more- and less-transparent causative verb forms, though they show some evidence of judgments that correspond to overgeneralization errors

Summary

Introduction

The question of how language learners come to avoid verb argument structure overgeneralization errors such as *The clown laughed the man – in some cases after a protracted period of producing them – has been described as a “learnability paradox” (Pinker, 1989: 415); “one of the most...difficult challenges for all students of language acquisition” (Bowerman, 1988: 73). The problem is this: On the one hand, children need to be able to use verbs in argument structure constructions in which they have not witnessed them; this type of productivity is the hallmark of human language. Children need to be able to constrain this generalization process in order to avoid producing ungrammatical utterances such as *The clown laughed the man. These types of errors, in which English-speaking children incorrectly mark causation using the transitive causative for verbs that prefer the periphrastic causative (e.g., The clown made the man laugh) are the focus of the present study; along with equivalent errors in Hebrew, Hindi, Japanese and K’iche Mayan. Similar errors have been observed in naturalistic data for Japanese (Nakaishi, 2016; see the experimental study of Fukuda & Fukuda, 2001), though they have not, to our knowledge, been investigated for any of the other languages included here

Objectives

Methods

Results

Conclusion