Have a personal or library account? Click to login
A Note on Weak vs. Strong Generation in Human Language Cover

A Note on Weak vs. Strong Generation in Human Language

By: Naoki Fukui  
Open Access
|Dec 2015

Abstract

This paper argues that various important results of formal language theory (e.g., the so-called Chomsky Hierarchy) may in fact be illusory as far as the human language faculty is concerned, as has been repeatedly emphasized by Chomsky himself. The paper takes up nested dependencies and cross-serial dependencies, the two important dependencies that typically show up in the discussion of the central classes of grammars and languages, and specifically shows that the fact that nested dependencies abound in human language while cross-serial dependencies are rather limited in human language can be naturally explained if we shift our attention from dependencies defined on terminal strings to abstract structures behind them. The paper then shows that nested dependencies are readily obtained by Merge, applying phase-by-phase, whereas cross-serial dependencies are available only as a result of copying Merge, which requires a constituency of the relevant strings. These results strongly suggest that dependencies are possible in human language only to the extent that they are the results from the structures that can be generated by Merge, leading to the conclusion that it is Merge-generability that determines various dependencies in human language, and that dependencies defined on the terminal strings are indeed illusory. A possible brain science experiment to demonstrate this point is also suggested.

DOI: https://doi.org/10.1515/scl-2015-0004 | Journal eISSN: 2470-8275 | Journal ISSN: 1017-1274
Language: English
Page range: 59 - 68
Submitted on: Dec 5, 2014
|
Accepted on: Dec 5, 2014
|
Published on: Dec 30, 2015
In partnership with: Paradigm Publishing Services
Publication frequency: 2 issues per year

© 2015 Naoki Fukui, published by The Chinese University of Hong Kong, T.T. Ng Chinese Language Research Centre
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.