Deletion, Deaccenting and Presupposition
C. Tancredi, 1992
In this dissertation, I examine the effects of deaccenting – the removal of phonological accent from a constituent – on interpretation. In general, deaccenting of an element is possible only if that element is salient in the discourse context. Salience alone, however, is not a sufficient condition for deaccenting. The discourse context in (1) makes salient both the verb hit and the NP John, and it is consequently possible for these elements to be deaccented in (1a,b). However, in (1c), it is not possible to deaccent both of these elements simultaneously. (Focus is indicated by CAPITALIZATION, deaccenting by small italics.)
(1) Mary hit John. Then,
a. BILL hit SUE.
b. BILL KICKED Mary.
c. #BILL hit Mary.
Deaccenting of both the verb hit and its diect object Mary in (1c) requires that the discourse context make salient a hitting of Mary, but this condition is not met in (1).
To account for the facts illustrated above, I propose that deaccenting plays a role in identifying the focus-related topic of a sentence, where it is a necessary condition for a sentence to be felicitous in a given context that the focus-related topic of that sentence be instantiated in the context. The focus-related topic of a sentence is generated (roughly) by replacing all focused constituents by variables and combining the resulting structure so as to end up with the smallest structure within which all properties of the remaining lexical elements are satisfied. The resulting structure will be instantiated in a context if there is another element in the context with which it is non-distinct, where variables count as non-distinct from other elements of the same semantic type. By this process, the sentences in (1) will have the focus-related topic in (2).
(2) a. x hit y
b. Mary
c. x hit Mary
The representations in (2a,b) are instantiated in the context consisting of the sentence Mary hit John in (1) above, and hence (1a,b) are felicitous. (2c), however, is not instantiated in this context, and hence is infelicitous.
The analysis sketched above provides an explanation for the felicity or infelicity of a sentence in a larger discourse context based upone the focus structure of the sentence and the composition of the context. While the analysis is of some interest by itself, even more importatnt is the use to which the analysis can be put in accounting for certain phenomena typically associated with VP deletion. Since Sag (1976) and Williams (1977), it has standardly been assumed that restrictions on pronominal interpretations in VP deletion contexts are to be explained in terms of the mechanism that assigns an interpretation to an empty VP. However, we find that identicaly restrictions appear in contexts in which a VP has been deaccented but not deleted. Thus, parallel to the sentence in (3a) which has only a strict and a sloppy reading available for the (deleted) pronoun, we find an identical restriction for the deaccented pronoun in (3b).
(3) a. Johni said hei is a genius because Bill did.
b. Johni said hei is a genius because Bill said he’s intelligent.
No analysis of VP deletion in the current literature is capable of accounting for this parallelism. If we assume that a phonologically deleted VP is represented in the LF representation of a sentence as a deaccented VP, however, it becomes possible to account for this parallelism in a principled fashion by assimilating the deletion cases to the deaccenting ones.
Thesis supervisor: James Higginbotham
Title: Professor of Linguistics and Philosophy
Table of Contents
1 VP deletion: the problem 8
1.1 Introduction 8
1.2 Previous analyses of VP ellipsis 12
1.2.1 Sag: a deletion-based analysis of VP deletion 12
1.2.2 Williams: a reconstruction-based analysis of VP deletion 17
1.2.3 Dalrymple, Shieber and Pereira: VP deletion as
pronominal interpretation of an empty VP 19
1.2.4 Summary of the three analyses 24
1.3 VP deaccenting 24
1.3.1 Further evidence 31
2 Deaccenting and presupposition 35
2.1 Introduction 35
2.2 Focus-related presupposition as topic 38
2.2.1 The standard analysis 38
2.2.2 Revisions to the standard analysis 41
2.2.3 Toward a theory of focus-related topic 43
2.3 Problems and extensions 46
2.3.1 Instantiation 48
2.3.2 Context incrementation 51
2.3.3 Multiple foci 58
2.4 Phonological accent and semantic focus 61
2.4.1 Semantic focus identification 61
2.4.1.1 The natural response test 62
2.4.1.2 Association with “only” 71
2.4.1.3 Refinements and caveats 74
2.4.2 Focus-related topics revisited 76
2.4.3 Standard presuppositions and focus-related topics 84
2.4.4 The necessity of deaccenting 86
2.5 As for x 89
3 VP deletion: the solution 91
3.1 Introduction 91
3.2 General form of the solution 92
3.2.1 Simple ambiguity resolution 93
3.2.2 Pronoun-induced ambiguities 98
3.3 Specifics of VP ellipsis 111
3.3.1 Copy based theories of VP ellipsis 112
3.3.2 Deletion based theories of VP ellipsis 119
3.3.3 Alternate deletion-based analysis 127
3.4 Pronominal interpretation revisited 131
3.4.1 Eliminative puzzles 132
3.4.2 Pseudo-sloppy identity 141