Argument/adjunct asymmetry in the acquisition of inversion in wh-questions by English-speaking children and Korean learners of English: Frequency account vs. structural account.

摘要

This dissertation discusses a possible argument/adjunct asymmetry in the acquisition of wh-questions by English-speaking children and Korean learners of English. In doing so, it tests and compares two approaches to the study of language acquisition: the structure-based generative approach and the lexical-based input frequency approach.; One of the most frequent errors in the acquisition of English wh-questions involves a failure to perform subject-auxiliary inversion (e.g., *why he is laughing?). Stromswold (1990) and De Villiers (1991) propose that inversion is acquired earlier in argument wh-questions (e.g., who and what) than in adjunct wh-questions (e.g., how and why), attributing the asymmetry to the structural difference between the two in the G/B framework. By contrast, Rowland & Pine (2000, 2003) reject the idea that children learn general movement rules and that there is an argument/adjunct asymmetry. Instead, they argue that children learn specific wh-word + auxiliary combinations that are sensitive to the frequency of individual pattern in input. By comparing L1 and L2 learners' input and acquisition, this dissertation aims to evaluate the two approaches and to investigate the possible argument/adjunct asymmetry.; An input study was done by analyzing the corpora of six caregivers' speech from the CHILDES database for L1. The L2 analysis involved 23 textbooks, 18 television situation comedy scripts, and 4 movie scripts. It was found that both L1 and L2 learners receive similar input in terms of the relative order of frequency of inverted wh-questions (i.e., what > why > how > who).; In addition, two experimental studies were conducted. The first was a picture-aided elicited production task involving 17 English-speaking children. The second was a grammaticality judgment task involving 41 Korean learners of English for L2. Both studies found that both L1 and L2 learners did better at inversion with argument wh-questions than with adjunct wh-questions (i.e., what, who > why, how).; A comparison of the input and acquisition studies reveals that both L1 and L2 learners show an argument/adjunct asymmetry in the acquisition of inversion in wh-questions that cannot be traced to the input frequency, which supports the structure-based generative approach.