«But as a multi-level objection marker: addressing issues raised by but-questions by Swee Khee Brenda Seah Advisor: Prof. Laurence Horn Submitted to ...»
But as a multi-level objection marker:
addressing issues raised by but-questions
Swee Khee Brenda Seah
Advisor: Prof. Laurence Horn
Submitted to the faculty of the Department of Linguistics in partial
fulfillment of the requirements for the degree of Bachelor of Arts
April 16th, 2014
Previous research on the English adversative marker but has focused on two aspects: the
type of meaning it encodes, and the substance of that meaning. But has been analyzed variously as contributing a conventional implicature (Grice 1967) or a secondary at-issue entailment (Potts
2005) on the one hand, and as behaving as an utterance modifier (Bach 1999) or marker of procedural meaning (Blakemore 2000) on the other. But’s function can be divided into PA and SN functions (Anscombre & Ducrot 1977); the latter includes correction, while the former encompasses uses such as denial of expectation, argumentation, and formal contrast. Attempts at a unified analysis of but usually focus on accounting solely for PA-type but by reducing multiple usages to an underlying function, e.g. argumentation (e.g. Anscombre & Ducrot 1983) or semantic opposition (e.g. Umbach 2005), or proposing that but marks another rhetorical relation, such as objection (Zeevat 2012). One deficiency in much of the literature is the preponderance of data where but connects declarative sentences, to the exclusion of instances of but prefacing non-declaratives, and but used to begin a speaker’s turn in a discourse (turn-initial but). The latter uses of but must be accounted for in a unified analysis of but.
This paper describes but used to preface questions (“but-questions”) and resolves two challenges to a unified analysis of but. First, but-questions are not as affected as but-assertions would be if but is removed, giving the impression of there being a “functional” but as opposed to a “rhetorical” but. Second, there is an opposition between but that contributes propositional content in some but-assertions and but that does not, especially in but-questions. The solution lies in extending Zeevat’s (2012) intuition that but is an objection (and, derivatively, adversative) marker with Lang’s (2000) claim that but can mark adversativity on multiple levels, where the but-clause contrasts with some assumption retrievable from the context.
(1) We will be watching the numbers all night (A), but what else should we be looking for in tonight’s results (B)?
But in (1) seems to ease topic development. Here, but marks the contrast between the assumption that the speaker continues on the same topi
This solves the first problem. The second problem is solved by attributing the production of propositional content not to but per se, but to inferential processes required to make the utterance make sense. This approach allows but to be treated as basically an utterance modifier.
By describing but-questions and addressing issues they raise for a unified analysis of but, this paper contributes towards a more complete understanding of the function of but, and also allows for a unified analysis of but in the future. (499 words) Seah 3 Acknowledgements I would like to thank my advisor, Larry Horn, for his help and patience throughout the process of writing the senior essay. His encyclopedic knowledge of his field(s) gave me some solid ground to stand on as I waded through the marsh of literature, while his insightful comments aided me in crafting my analysis. I am also very grateful to our instructor for the year-long senior essay course, Raffaella Zanuttini, whose enthusiasm, encouragement and guidance saw our class through from the initial conception of our ideas to the final versions of our presentations and essays. I will very much miss our senior essay class – Diallo, Maria, Pat and Patrick – and I thank them for their companionship throughout this journey and for making it an unlonely one.
Finally, my brother Brandon and my fiancé John definitely deserve some appreciation for supporting me as deadlines loomed – and for providing comic relief by making my senior essay topic the but of a few jokes.
1 Introduction Two aspects of the adversative marker but in English have been studied: the type of meaning it contributes to an utterance, and the substance of that meaning. In terms of the type of meaning but contributes to an utterance, it has been analyzed as contributing a conventional implicature (Grice 1975; Horn 2013 as “F-implicature”) or secondary at-issue entailment (Bach 1999 (“The myth of conventional implicature”), Potts 2005). But has also been analyzed as acting as a marker of procedural meaning (as opposed to indicative meaning), acting to comment on the utterance or manage the discourse at a higher level, as an utterance modifier (Bach 1999) or discourse marker (Blakemore 2000).
Those who discuss how to categorize but’s contribution often assume that there is a fundamental meaning of but that is obvious, for example, that it presents a contrast between the two conjuncts (Frege 1892, Grice 1975). However, many have distinguished among various uses of but and attempted a unified analysis of these uses. The most often cited uses of but include formal contrast, denial of expectation, argumentative, and correction. Analyses have been proposed in which most of these uses of but are derived from a common function of but, e.g.
presenting semantic opposition between conjuncts (Umbach 2005, Jasinskaja and Zeevat 2008, 2009), functioning in argumentation (Anscombre and Ducrot 1977, 1983, Merin 1999, Winterstein 2012), or marking objection (Zeevat 2012).
This essay focuses on one gap in the literature on but: the use of but in informationseeking questions. I have chosen to focus on this for two reasons. First, almost all examples of the usage of but in the literature focus on declarative sentences, where but interacts with the asserted content. This is problematic as but, like other adversative connectors, can be used to conjoin different speech acts (Lang 2000), and the conjoined clauses may not have asserted content (e.g.
it is an interrogative clause). Existing analyses of but usually employ terminology that works for declaratives, but not for other speech acts. The intuition underlying a full account of but must be extensible to cases where but conjoins different speech acts. Therefore, I have chosen to look at questions beginning with but (“but-questions”) in particular, as a representative case of but conjoining different speech acts.
Second, many accounts of but treat it as a conjunction only. While but very often conjoins two clauses (“connective” but), but can also be used to begin a sentence without Seah 6 anything preceding it by the same speaker (“turn-initial” but). Turn-initial but has been addressed in a few discussions (Bach 1999, Zeevat 2012), but it would be worthwhile to examine it in more detail. But in questions may be connective or turn-initial, and turn-initial use of but in questions is fairly common.
In this essay, I aim to do the following: provide a preliminary taxonomy of but-questions and a description of but’s function in the different types of but-questions, and discuss the implications of my data for determining the type of meaning but contributes to an utterance and the function it has in discourse. I adopt the view that but is ultimately an utterance modifier that marks objection on multiple levels. By addressing some apparent problems that stand in the way of a unified analysis of but, I contribute to a more complete survey of the function of but and allow for a unified analysis of but-questions and but-assertions.
The rest of this paper will be structured as follows: Section 2 discusses in more detail the extant literature on but; Section 3 recapitulates the need to study but-questions and describes the process I used to search for examples of but-questions in the Corpus of Contemporary American English (COCA); Section 4 provides a description of but-questions and the issues they raise for describing the function of but; Section 5 discusses these issues; and Section 6 concludes the essay.
An appendix containing extended contexts for the but-questions I cite follows the bibliography.
2 Background: Treatments of but Many have attempted to pin down the meaning of but. Larger theoretical questions are often encountered along the way, most notably regarding the proper division of content in an utterance and the way discourse works. Two questions in particular can be asked about the contribution of but: first, what kind of contribution is it within the utterance, and second, what is the semantic substance of the contribution? Section 2.1 addresses the first question by reviewing the notion of conventional implicature, while section 2.2 addresses the second question by summarizing recent research into but’s meaning.
distinguishes between the “sense of the clause” and the contribution of the conjunction, saying that the conjunction “actually has no sense and does not change the sense of the clause but only illuminates it in a peculiar fashion”. His primary example here is although, but mentions in a footnote that but and yet work in a similar way. Here we see that the contribution of a conjunction such as but is distinct from the contribution of the clause which hosts it. Yet, that contribution is hard to pin down. In his earlier work, Begriffsschrift (1879), Frege cannot find an expression for the difference between and and but in his formal approach to logic; “a speaker uses ‘but’ when he wants to hint that what follows is different from what might at first be supposed” (as quoted by Horn (2013)).
Grice characterizes the notion of “implicature” in his William James Lectures, delivered in 1967 at Harvard University.1 He, like Frege, distinguishes between what is said (the truthconditional content) and what is implicated (non-truth-conditional content), while both fall under what is meant by the speaker. Grice is primarily concerned with conversational implicature, which arises from the interaction of the conversational maxims with the context of the utterance, and mentions very briefly the notion of conventional implicature, which arises from the conventional meaning of a word or expression. The now classic example he gives is the sentence “She was poor but she was honest”, which conveys the assertion “She was poor and she was honest”, and also the implicature that there is some contrast between her honesty and poverty. The implicature is non-cancellable, as you cannot deny the contrast right after saying the sentence. The implicature can be traced to the conventional meaning of the word but (as replacing but with and would avoid the implicature, making the implicature detachable).
The category of conventional implicature (henceforth CI) has been subject to much scrutiny. Karttunen and Peters (1979) affirm the existence of the category of conventional implicature. They assimilate most presuppositions to CIs by means of the logic they propose for the latter, which probably has led to the impression for many that CIs can be described like presuppositions. They also add that CIs can be backgrounded, that is, entailed by the shared knowledge in the common ground. They cannot be challenged in a direct way. In their paper, they do not discuss but, choosing instead to focus on words such as too, either, also, even and only.
1 For reference I looked up the compilation “Studies in the Way of Words”, published in 1989. Seah 8 On the other hand, Bach (1999) regards CIs as a mix of different phenomena, and argues that many alleged CIs are actually part of what is said, and they seem to be CIs because of the illusion that the propositional content associated with the CI-trigger seems irrelevant to the truth conditions of the utterance. He argues for a multidimensional semantics, where a sentence can express more than one proposition. He thereby proposes that such alleged CIs are a secondary aspect of what is said. Other alleged CIs may be utterance modifiers, which neither contribute to what is said, nor generate CIs, but rather perform “second-order speech acts”. Interestingly, he classifies but under both types: a given instance of but may either express a secondary aspect of what is said, or otherwise perform a second-order speech act. Blakemore (2000) likewise denies the existence of CIs, albeit under a different framework, that of Relevance Theory.
Potts (2005) revisits Grice’s original discussion of CIs in order to extract a description of the properties of CIs. In doing so, he introduces a new set of data that fits this description.