運用MacWhinney（1995）研發之「兒童語言資料交換系統」（Child Language Data Exchange System，簡稱CHILDES）內的Leonard 語料庫，本研究期望藉由語料的分析，了解以英語為母語的特定型語言障礙（Specific Language Impairment，簡稱SLI）兒童其句法發展的問題所在，並檢證van der Lely（2005）宣稱SLI 兒童具有句法異常而提出的「建構語法複雜性的缺陷」（Deficit in Computational Grammatical Complexity，簡稱CGC）假設。
本研究的結果顯示Leonard 語料庫中的SLI 兒童，儘管在時態標記（-ed）、一致標記（-s）和助動詞移位（auxiliary inversion）上經常犯錯，但在其他句法結構的生成上（例如：格位指定、A 移位、WH移位）卻顯現沒有問題。依據Chomsky（1998, 1999, 2001, 2005）的「極簡理論」（Minimalist Program，簡稱MP），本文提出的結論是：SLI 涉及的是PF 部份裡的問題，而非句法操作上的缺陷。因此，顯示「建構語法複雜性的缺陷」假設無法與本研究之結論相容。

Based on the Leonard corpus in the Child Language Data Exchange System
(CHILDES), this paper tries to determine the nature of grammatical errors made by English-speaking children with specific language impairment (SLI) and to evaluate a recent account of SLI, namely van der Lely’s (2005) Deficit in Computational Grammatical Complexity (CGC) hypothesis, which asserts that SLI can be characterized by a primary deficit in syntax. However, it is found that the SLI children in the Leonard corpus correctly perform syntactic operations such as Case assignment,A-movement and Wh-movement even though they frequently commit errors on tense marking, agreement marking and auxiliary inversion. Within the framework of Chomsky’s Minimalist Program (1998, 1999, 2001, 2005), this paper argues that SLI may involve a PF deficit rather than a syntactic deficit. Therefore, it is shown that the CGC hypothesis is not compatible with this finding.