Menu
Home
Log in / Register
 
Home arrow Language & Literature arrow Theory and Data in Cognitive Linguistics

Gries

The chapter by Gries addresses a number of methodological questions which have come up in the cognitive literature. In particular, it addresses Bybee's (2010) criticism of collostructional analysis and discusses the roles that frequencies, conditional probabilities, and association measures play in the study of co-occurrence phenomena in usage-based linguistics. Here the basic question is how to induce generalisations from a corpus, and the research strategy is to use modern statistical techniques against a background notion of association measures as a way of exploring language structure from corpus data.

Gries's chapter sets out to define and exemplify the method of collostructional analysis and to provide a range of references for its successful application. It also clarifies a number of misunderstandings about the method, particularly concerning the choice of association measures, and the robustness of the measures given different corpus sizes and observed frequencies. It then exemplifies the explanatory and predictive power of collostructions from corpus data and experimental validation and outlines a research programme, or perspective, that views distributional and co-occurrence data as important in cognitive theoretical research, and motivates it both theoretically and empirically.

Given his theory's statistical underpinnings, and strong commitment to naturally occurring data, Gries then relates co-occurrence data in constructions to Zipfian distributions and entropy, and proposes a new quantitative definition of "construction".

Barðdal et al

Barðdal et alls chapter has a historical focus as well, but focuses on the area of reconstruction. In particular, the authors set out to reconstruct the dative subject construction in a proto-stage of Indo-European using evidence from Old Norse-Icelandic, Ancient Greek, Latin, Old Russian, and Old Lithuanian. The novelty of the chapter lies in the nature of the object of reconstruction. Within the field of reconstruction (see e.g. Fox 1995), the focus has traditionally been on structure, especially phonological structure. Reconstructing the syntax and semantics of lexical items and grammatical morphemes has been more controversial and less common. But Barðdal et al. go further than that by studying a complex construction, defined as a form-meaning pairing.

In more traditional models such as Generative Grammar the research question addressed by the authors would not even arise, as constructions are not recognised. As a result, related work tends to propose a very abstract meaning for the dative case. By contrast, here, as well as in earlier work by Barðdal (e.g. 2009), constructions are not only recognised but, following Croft (2001), they are seen as the very building blocks of grammar. Thus, dative case markers are seen as an integral part of the various constructions they occur in, and the task of reconstructing the dative becomes the task of reconstructing constructions featuring this case marker.

Barðdal et al. investigate semantic differences and similarities among the data from the five different branches of Indo-European included in their sample, and on this basis, reconstruct a possible proto-stage. Due to the exclusion of a number of other branches, they do not make any general claims about proto-Indo European as a whole, but rather suggest that their reconstruction may be applicable to a West-Indo-European stage.

An important tool in Barðdal et alls reconstruction is the semantic map model (see e.g. Croft 2001, Haspelmath 2003). Unlike the generative quest for Universal Grammar, the semantic map approach begins with the suggestion that what languages share is conceptual, and that language universals are to be discovered in constraints on the mapping between the conceptual plane and linguistic structure.

 
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Mathematics
Political science
Philosophy
Psychology
Religion
Sociology
Travel