Without 'Focus'

Abstract

It is widely accepted that a notion of ‘focus’, more or less as conceived of in Jackendoff (1972), must be incorporated into our theory of grammar, as a means of accounting for certain observed correlations between prosodic facts and semantic/pragmatic facts. In this paper, we put forth the somewhat radical idea that the time has come to give up this customary view, and eliminate ‘focus’ from our theory of grammar. We argue that such a move is both economical and fruitful. Research over the years has revealed that the correlations between prosody, ‘focus’, and the alleged semantic/pragmatic effects of focus are much less clear and systematic than we may have initially hoped. First we argue that this state of affairs detracts significantly from the utility of our notion of ‘focus’, to the point of calling into question the very motivation for including it in the grammar. Then we look at some of the central data, and show how they might be analyzed without recourse to a notion of ‘focus’. We concentrate on (i) the effect of pitch accent placement on discourse congruence, and (ii) the choice of ‘associate’ for the so-called ‘focus sensitive’ adverb only. We argue that our focus-free approach to the data improves empirical coverage, and begins to reveal patterns that have previously been obscured by preconceptions about ‘focus’. Without ‘Focus’ 2 1. ‘FOCUS’ AND ‘FOCUS EFFECTS’ Following Chomsky (2007) and Jackendoff (1972), generative linguists have incorporated into theories of grammar a notion of ‘focus’: to be a ‘focus’ is a feature of syntactic constituents which gets interpreted both phonologically and semantically/ pragmatically by rules of the grammar. This notion of ‘focus’ is intended, of course, to serve as a means of accounting for certain correlations that we observe between prosodic facts on the one hand, and semantic/pragmatic facts on the other. We think it is fair to say that when “Jackendoffian” ‘focus’ was incorporated into the grammar, it was hoped, at least initially, that there would be one uniform and well-defined notion of ‘focus’, with some clear semantic/pragmatic criterion for identifying the focused constituent, and highly systematic prosodic marking of the focused constituent; and that this notion of ‘focus’ would be useful for stating generalizations and making the right predictions regarding a variety of phenomena which involve a correlation between prosodic facts and semantic/pragmatic facts (those phenomena that we are accustomed to call ‘focus effects’). That is, we think that at least initially, a really appealing theory of ‘focus’ and ‘focus effects’ was envisaged, one which would supply an affirmative answer to the following three questions. • Is there a semantic/pragmatic ‘definition’ of ‘focus’? That is, is there such a semantic/pragmatic criterion which (i) given any utterance α and its context, would enable us to identify the focus (or foci) of α; and (ii) would be independent of any further semantic/pragmatic generalizations that we would like to capture using the notion ‘focus’? • Is there a prosodic ‘definition’ of ‘focus’? That is, is there some highly systematic prosodic marking of ‘focus’? I.e., is there a prosodic criterion which, given any utterance α (including its syntactic structure and prosody), would determine what the possible focal structures of α are? • Can we have a prosodic ‘definition’ and a semantic/pragmatic ‘definition’ of focus which ‘define’ the same ‘focus’? That is, if a certain semantic/pragmatic phenomenon involves a Vol. 6: Formal Semantics and Pragmatics: Discourse, Context, and Models 3 Nirit Kadmon & Aldo Sevi constituent which seems to be prosodically marked, and if we want to hypothesize that that constituent is a ‘focus’, then will our hypothesis be compatible with both the prosodic and the semantic/pragmatic ‘definitions’ of focus? Undoubtedly, no semantic/pragmatic criterion for identifying foci could ever be fool-proof (sometimes we might not be able to tell whether the conditions that it specifies hold or not). Also, it has been obvious since very early on that it is not the case that prosody completely disambiguates focal structure. But if it could be possible to come up with both a prosodic and a semantic/pragmatic ‘definition’ of focus which together give us enough information about what the focus of each utterance might be, then we might still have a contentful theory of ‘focus’ which could be checked against the facts. At least when we restrict our attention to a limited set of data, it might seem that an appealing theory of ‘focus’ can be achieved. Consider examples (1) and (2). (1) A: Who did you introduce to Sue? B: I only introduced Bill to Sue. L∗ H∗L L% (2) A: Who did you introduce to Sue? # B: I only introduced Bill to Sue. L∗ H∗ L L% We observe the following facts: prosody: In each of the B sentences, there are just two words carrying a pitch accent: only, and one of the names. The name carries the nuclear accent. discourse congruence: B is a felicitous response to A in (1), but not in (2). The accented name in (1B) corresponds to the wh-word in (1A), whereas the accented name in (2B) does not correspond to the wh-word in (2A). www.thebalticyearbook.org Without ‘Focus’ 4 semantic effect: In example (1), the domain of quantification of only is unambiguously fixed, so that the only possible interpretation is ‘the only individual I introduced to Sue was Bill’. We are accustomed to refer to this interpretation by saying that only ‘associates with’ Bill. In this particular example, we have just noted that only obligatorily associates with Bill. These examples are usually analyzed based on the assumption that in the B sentence, the accented name is a ‘focus’—that is, the NP Bill/Sue carries a syntactic feature F, the H∗ pitch accent is the prosodic manifestation of this feature, and the F feature is semantically/pragmatically interpreted. We may assume, following Rooth (1985), that the F feature semantically induces a set of alternatives, also called a ‘focus semantic value’. We may assume the focus semantic value of example (3) is as in (4). (3) I introduced [Bill]F to Sue. (4) the set of propositions which say of some individual d that the speaker introduced d to Sue—where d is an atomic individual or a plural individual e.g., something like this: {I introduced Bill to Sue, I introduced Tom to Sue, I introduced Jack to Sue, I introduced Bill and Tom to Sue, I introduced Bill and Jack to Sue, I introduced Bill, Tom and Jack to Sue,. . . } This is then used to explain the facts, as follows. We want to predict that the focus in the answer must correspond to the wh position in the question. Following Roberts (1996), and Beaver & Clark (2008), this can be done by positing the following constraint. (5) The Current Question Under Discussion Constraint on Fo-