A grammar that uses (recursive) structure building operations is adequate if and only if it recognizes and generates nothing but interpretable, well-formed sentences that are part of the language that the grammar aims at characterizing. If structure building operations overgenerate by recognizing and generating sentences that are considered ungrammatical by native speakers of the language under analysis, some relevant constraint should be added to the grammatical description. This necessity was clearly stated in Ross (1967), and its urgency was revealed by operations like Move α before (Lasnik and Saito 1992), then Merge, Move and Agree more recently (Chomsky 1995-2012): these operations are very general and simply too powerful. On the one hand, they might highlight important universal principles of human language structure, but, on the other, they require too many restrictions to meet, ultimately, the upper bound of the relevant linguistic description. It is a matter of fact that these operations are often defined in such a way to be practically insufficient to circumscribe many relevant empirical facts. Furthermore, common constraints (e.g. binary merge, locality or derivation by phase...) that have been proposed to limit the generative power of grammars are not explicitly related to notions of complexity or simplicity in any explicit computational sense. For instance, neither time or space complexity measures are fully taken into consideration nor a formalization of the problem space is provided (cf. Barton et al. 1987); in the end, what might seem a simpler structure building operation lead to rather complex restrictions in order to attain to empirical (descriptive) adequacy. In this special issue, nine authors investigated the empirical fit of “simpler” grammatical theories, discussing how certain restrictions on structure building procedures, or different formulations of them, lead to theories that are more descriptively and explanatory adequate. This is done in this volume by mainly focusing on how linear constraints restrict, or are restricted by, the hierarchical structure that is built incrementally.
Special Issue on Directionality of Phrase Structure Building
CHESI C
2013-01-01
Abstract
A grammar that uses (recursive) structure building operations is adequate if and only if it recognizes and generates nothing but interpretable, well-formed sentences that are part of the language that the grammar aims at characterizing. If structure building operations overgenerate by recognizing and generating sentences that are considered ungrammatical by native speakers of the language under analysis, some relevant constraint should be added to the grammatical description. This necessity was clearly stated in Ross (1967), and its urgency was revealed by operations like Move α before (Lasnik and Saito 1992), then Merge, Move and Agree more recently (Chomsky 1995-2012): these operations are very general and simply too powerful. On the one hand, they might highlight important universal principles of human language structure, but, on the other, they require too many restrictions to meet, ultimately, the upper bound of the relevant linguistic description. It is a matter of fact that these operations are often defined in such a way to be practically insufficient to circumscribe many relevant empirical facts. Furthermore, common constraints (e.g. binary merge, locality or derivation by phase...) that have been proposed to limit the generative power of grammars are not explicitly related to notions of complexity or simplicity in any explicit computational sense. For instance, neither time or space complexity measures are fully taken into consideration nor a formalization of the problem space is provided (cf. Barton et al. 1987); in the end, what might seem a simpler structure building operation lead to rather complex restrictions in order to attain to empirical (descriptive) adequacy. In this special issue, nine authors investigated the empirical fit of “simpler” grammatical theories, discussing how certain restrictions on structure building procedures, or different formulations of them, lead to theories that are more descriptively and explanatory adequate. This is done in this volume by mainly focusing on how linear constraints restrict, or are restricted by, the hierarchical structure that is built incrementally.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
