Sorace, Antonella

2018-01-31T11:24:54Z

2018-01-31T11:24:54Z

1993

Tnis study has both a theoretical and a methodological dimension. Theoretically, it is concerned with variation and indeterminacy in linguistic acceptability judgments. Methodologically, it involves the application of a rigorous procedure for the elicitation of judgment data that is sensitive to informants' variable or indeterminate intuitions.

The theoretical focus is on the linguistic intuitions of native and non-native speakers of Italian about a number of grammatical phenomena related to the choice between the auxiliaries ESSERE ('be') and AVERE ('have') with non-transitive (unaccusative and unergative) verbs. It is argued that a purely syntactic account of unaccusativity is insufficient to capture the variation exhibited by these verbs. In particular, it is claimed that the unmarked selection of ESSERE with unaccusatives and of AVERE with unergatives in the present perfect tense is sensitive not only to a hierarchy of syntactic configurations (as assumed by the Government-Binding version of the Unaccusativity Hypothesis) but also to lexical hierarchies that subdivide the range of unaccusative and unergative verbs along semantic dimensions. Such hierarchies distinguish 'core', or prototypical, types of verbs from peripheral ones , and are consistent with the historical evolution of auxiliaries in Romance. However, auxiliary selection in syntactically marked 'restructuring' constructions, induced by certain Raising and Control verbs, is not sensitive to these semantic dimensions. It was predicted that the interaction between syntactic and semantic constraints would give rise to systematic variability in native speakers' linguistic intuitions, manifested in consistent and determinate acceptability judgments on core types of verbs, and variable or indeterminate judgments on peripheral types of verbs. It was also predicted that non-natives would differ from natives in terms of the extent to which indeterminate judgments penetrated from the periphery to the core.

Methodologically, this study represents the first application of magnitude estimation techniques to the elicitation of linguistic acceptability judgments. Magnitude estimation makes it possible to measure variability in acceptability judgments directly, which has the advantage of producing interval scales that can then be properly analysed by parametric statistics. Other ranking elicitation procedures produce only ordinal measurements. A systematic comparison between the judgments obtained by means of magnitude estimation and those obtained by means of a cardsorting ranking procedure indicates that both native and non-native speakers are able to judge acceptability via magnitude estimation with at least as much delicacy as via card-sorting. In some cases, magnitude estimation produces finer-grained distinctions of unacceptability, and reveals differences between native and near-natives which are not reproduced in the card-sorting task.

A series of experiments was conducted addressing the three issues of (a) variability in native intuitions, (b) progressive development of non-native knowledge, and (c) ultimate attainment at near-native competence levels. Acceptability judgments were collected from Italian native speakers and English-speaking learners of Italian at four proficiency levels (beginner, intermediate, advanced, near-native). A group of French near-native speakers of Italian was also tested for the purpose of comparison with the English near-natives.

The results show that (a) the judgments of native Italians are sensitive to different lexical-semantic hierarchies of unaccusative and unergative verbs: judgments on the basic syntactic reflexes of the unergative/unaccusative distinction (auxiliary selection and ne-cliticization) exhibit more or less determinacy depending on the semantic characterization of individual verbs: however, native speakers discriminate categorically between possible and impossible, obligatory and optional auxiliary change under restructuring, irrespective of the semantics of the inducing verb; (b) nonnative judgments reflect a difference in learnability between lexical-semantic and purely syntactic distinctions. Lexical-semantic hierarchies affect the development and ultimate shape of non-native grammars, in that interlanguage representations for core lexical classes are constructed earlier than those for peripheral classes, with non-native acceptability values gradually approximating the native values. Peripheral restructuring constructions, however, never become determinate in the interlanguage grammars of English learners, which are incomplete in this respect even at the nearnative level. In contrast, French near-native speakers of Italian show evidence of having constructed determinate, but divergent representations of these syntactic phenomena. It is argued that such differences in ultimate attainment reflect differences in the overall representations of unaccusativity in French and English.

http://hdl.handle.net/1842/26955

The University of Edinburgh

Annexe Thesis Digitisation Project 2017 Block 15

Already catalogued

Lexical conditions on syntactic knowledge: auxiliary selection in native and non-native grammars of Italian

Thesis or Dissertation

Doctoral

PhD Doctor of Philosophy