Back to list

Detail of contribution

Auteur: Andrew WEIR

Pronoun drop and article drop: a unified analysis

Abstract/Résumé: I discuss the case of object drop in English 'reduced written register', such as recipes (Haegeman 1987, Massam 1989, Massam and Roberge 1989) and diaries. (1) a. Put batter in oven. Bake _ for ten minutes. b. Received credit card bill in mail today; will shred _ later. Object drop differs from subject drop in these registers (Haegeman 1997, 2007, a.o.); notably, dropped subjects can be of any person, while dropped objects can be third person only. (2) a. _ Will go to gym later. (1p, 'I' or 'we') b. [comments on a student's essay] _ Don't need to go into so much detail here. (2p, 'you') c. Saw Bill today. _ Looked really mad. (3p, 'he') (3) a. Will shred _ later. (3p, 'it') b. Very angry at Bill. ?Might fire _. (3p, 'him' or 'her') c. Don't talk to me. *Will fire _. (*2p 'you') d. Don't want to talk to my boss. *Would fire _. (*1p 'me') Haegeman (1987) proposes an analysis in which null objects are traces bound by null topic operators, but I show that null objects in these registers do not show characteristics of traces. For example, they are not island-sensitive, pace Haegeman (1987); e.g. 'Wait for ten minutes after removing _ from oven', a gap in an adjunct clause, is licit (cp. *'The chicken, wait for ten minutes after removing t from the oven'). Rather, I propose that the ability to drop objects is dependent on article drop in these registers, as shown in (4). (4) a. Take two peppers. Chop _ peppers finely. [recipe] b. _ Man accused of robbing _ store [headlines, Stowell 1991] c. Saw _ very strange man today [diaries] Rather than being simply an unpronounced version of 'a' or 'the', I analyze the absence of the article in these cases as the presence of a phonologically null determiner with the semantics of a choice function. Note that this null article cannot receive a generic interpretation, unlike 'a'. (5) In my day, _ gentleman wouldn't do that. (no generic reading 'A gentleman wouldn't do that') I adopt Tomioka (2003)'s analysis for Japanese null pronouns, in which the absence of a determiner, combined with NP ellipsis, allows a constituent with pronominal-like semantics to go wholly unpronounced. I argue that a similar process is at work in English reduced written register. This analysis allows us to explain the person restriction: if a null object underlyingly contains an elided NP and silent determiner, e.g. [bake [ ] for ten minutes], then we expect such null objects to only have the third person referents which NPs can. If this analysis is correct, then it allows us to account for apparently significant grammatical variation between registers of English (the licitness of null objects) by appealing to a relatively minor lexical difference (the availability of a null determiner).