ACTA LINGUISTICA TOM. 37 (A MTA NYELVTUDOMÁNYI KÖZLEMÉNYEI, 1987)

1987 / 1-4. sz. - ROBINSON, P. J.: Metaphors for the description of acquisition data: form constituency "trees" to dependency "frames"

Acta Linguistica Academiae Scientiarum Hungaricae, Tomus 37, (1-4), PP• 31-62 (1987) METAPHORS FOR THE DESCRIPTION OF ACQUISITION DATA: FROM CONSTITUENCY "TREES" TO DEPENDENCY "FRAMES"* By P.J. ROBINSON 1.1. Lexical Grammar and Language Learning Since Levenstons's claim that studies of lexical acquisiton had been "discriminated against" (1979) there has been a renewel of interest in the These metaphors constitute the bases for separate descriptive paradigms; those of constituency theory and dependency theory respectively. There are points of opposition between these paradigms which I identify in the main body of the paper and these give each paradigm its separate status. Below constituency theory, in the first paradigm, are the basic 'tree' metaphor and the associated concepts of hierarchy and level. Above constituency theory are various models which are based, wholly or in part, on the constituency view of language structure. Similarly with the dependency paradigm; below it is the basic metaphor of the 'frame' and the related concepts of linearity and range. Above the theory are models which derive from it. These points are all expanded in the main text. Previously models based on constituency theory have been in the ascendant in studies of acquisiton data. The constituency paradigm has therefore established leadership, or 'hegemony', in the field of describing data. Such descriptions have, of course, been constrained by the underlying metaphor, the modular distinctions between levels, and the related concept of hierarchy. I try to outline the consequences such constraints have had for hypotheses about 'parsing', but the main consequences has been the decoupling of lexis and syntax in studies of Ll and L2 acquisition data. However I attempt in my account of the process of 'fission' to show the continuity between acquiring an initial 'unit' and then storing syntactic evidence of its potential 'range' of distribution in the lexicon. Distinctions between levels and the notion of hierarchy have been motivated to a large extent by the desire to achieve 'modularity' and 'economy' of description in the models concerned. But these do not appear to be as motivating to current investigators as they were previously. For example, modularity is now 'dispreferred' by those like Jackendoff (1977) and Hudson (1984) who prefer to try and characterize the 'interaction' between levels of description in terms of a 'network'. Similarly evidence from acquisition and text analysis points to the regular recurrence of many similar and partly analyzed patterns or chunks, which are stored holistically. Peters has suggested these are the building blocks needed by the learner for subsequent analysis and Pawley and Syder have pointed to the fact that such 'phraseological' patterns continue to be useful to adults. Thus the motivations underlying the constituency paradigm are being questioned, both by those who wish to demonstrate the interrelatedness of levels of description, and of 'everything else' (Hudson 1986) in cognitive structure, and by performance evidence of considerable redundancy in storage. The metaphor of the 'frame', and the related concepts of linearity and range are increasingly important in recent lexicalist theories like LFG and 'Lexicase' (Starosta 1987). Thus 'hegemony' appears to have shifted to the dependency paradigm. Akadémiai Kiadó, Budapest

Next