1.2 Analyses of ‘There’-Sentences
2.1.1 Information Structure Overview
In this subsection, I will introduce the notion of information structure, along with the di- chotomies of topic-comment, background-focus and theme-rheme that are commonly dis- cussed in work on information structure.
Information structure(IS) is a term that originates with Halliday (1967), though similar ideas may be found in the ‘information packaging’ of Chafe (1976) and the ‘functional sen-
tence perspective’ of the Prague school (e.g. Sgall et al. (1986)). The most precise characteri- sation of IS is employed in Chafe (1976), where it is defined as follows:
Information Structure (IS) definition: A system of categorising aspects of the occur- rence of a sentence in order to analyse thepackaging of informationin light of theimme- diate communicative needs, goals and interests of interlocutors.
Given that IS is defined relative to interlocutors’ communicative needs, a background the- ory of communication must be described in order to clarify this definition. I shall assume that communication may be modelled in terms of thecommon ground(‘CG’) developed in Stalnaker (1978), which is defined as a set of propositions (and those sets that may be con- structed from propositions) associated with a conversation at a given time, consisting of the information mutually accepted by all interlocutors for the purposes of that conversation. CGs are primarily altered byassertions, as the proffered content of an assertion is analysed as a proposed change in the propositions contained in the CG, and an unchallenged as- sertion will update the CG in the proposed manner. Presuppositions are then analysed as requirements for the input CG, and a second means of altering the CG is via theaccommoda- tion of presuppositions. That is, if an occurrence of a sentence triggers a presupposition that the input CG fails to meet, the CG may be subsequently updated to reflect the requirements imposed upon the input CG, provided the requirements are suitably uncontroversial. For example, an occurrence of ‘Zaid knows that every king is bald’ asserts that Zaid knows that every king is bald, adding this information to the CG if unchallenged; yet it is normally thought to also presuppose that every king is bald, thereby requiring the CG to already in- clude this information. If the CG does not include the information that every king is bald, then discourse participants will either deem the occurrence infelicitous or accommodate the presupposition by adding the required information.
In order to encode interlocutors’ communicative needs, goals and interests within the CG, the notion of a question under discussion developed in Ginzburg (1995a), Ginzburg (1995b) and Roberts (1996) shall be used. It is standard to follow Hamblin (1973) in treating the ordinary meaning of a question as a set of propositions. The meaning of a question is then thequestion under discussion (‘QUD’) relative to a given context if the interlocutors enter it into the common ground, along with their intention to answer it. Selecting a QUD is a way of modelling interlocutors’ communicative needs due to this encoding of a set of propositions in the CG where certain information is taken for granted and certain other information is marked as absent yet desired. A proposition is apartialanswer to a QUD if it contextually entails the truth or falsity of at least one element of the set of propositions comprising the QUD, and it is acompleteanswer to the QUD if it contextually entails the truth or falsity of every element of the set of propositions. If a natural language sentence thatreflectsthe QUD (that is, expresses the same set of propositions as the QUD) is explicitly stated, then there is anexplicit QUD operative at that context, whereas otherwise the QUD isimplicit.
Finally, IS is said to pertain toimmediatecommunicative needs because there are certain other factors that might affect the way in which information is packaged that fall outside the remit of IS, such as long-term background knowledge or politeness considerations.1 This completes the clarification of the definition of IS.
There are three dichotomies that are frequently found in discussions on IS:topic-comment, background-focusandtheme-rheme. The topic-comment dichotomy is emphasised by Straw- son (1964), Chafe (1976), Gundel (1974), Kuno (1972) and Reinhart (1981), and centres around a distinction between the expression that indicates what a sentence is about (topic) and the parts of a sentence asserting something about this ‘aboutness item’ (comment). The background-focus dichotomy is emphasised by Rooth (1985), Krifka (2007) and B ¨uring (2016), and is described in terms of constituents that indicate the presence of alternatives that are relevant to the interpretation of the sentence (focus) and those parts of a sentence that in- voke no such alternatives (background). The theme-rheme dichotomy is emphasised by the Prague school, including Halliday (1967), Firbas (1969), Benesova et al. (1973) and Daneˇs (1974), and is characterised in terms of a division between the parts of a sentence that con- sist of old information relating to prior discourse (the theme) versus the parts that contain new information about the theme (the rheme).
For most sentences, there are multiple possible ways of carving them up according to these dichotomies, though factors such as prosody, communicative needs and syntactic structure will frequently place some restrictions on these possibilities (as shall be discussed in §(2.3.1)). In almost all cases, differences in information structure fail to affect the truth conditions of occurrences of sentences; instead, they affect the management of propositions within the CG.2
For example, we may mark the following occurrences of sentences according to reason- able choices of topic (‘T’), comment (‘C’), background (‘B’), focus (‘F’), theme (‘Th’) and rheme (‘Rh’):
[Zaid]T [burned everything in his yard]C.
[Zaid]B[burned everything in his yard]F.
[Zaid]T h [burned everything in his yard]Rh.
It should be apparent that there is considerable overlap between the properties each di- chotomy purports to capture, to the extent that some have taken ‘topic’, ‘background’ and ‘theme’ to denote the same feature of sentences, a feature which they contrast with the as- pect of a sentence referred to by ‘comment’, ‘focus’ and ‘rheme’. However, Ebert (2009) concurs with advice given in Moln´ar (1993) that we resist such a conflation, on the grounds that each of the three dichotomies capture a slightly different notion. Moln´ar (1993) argued that the topic-comment dichotomy relates to the factual aspect of an utterance, whereas the theme-rheme distinction depends on discourse-givenness from the hearer’s point of view, and the background-focus distinction relates to relevance from the speaker’s point of view. I will follow this recommendation to refrain from an overly hasty conflation of the cate- gories, since the connections between the different properties of occcurrences of sentences then becomes an empirical matter to be investigated rather than an a priori matter.
My concern in this thesis will predominantly centre on the idea of asentence topic, since this notion turns out to illuminate a range of issues related to DPs. An initial difficulty of relying on the topic-comment dichotomy emerges from the observation in Reinhart (1981)
2A potential exception arises with respect to focus-sensitive particles such as ‘only’, since some have argued
that different choices of focus have a semantic effect for such cases. See Krifka (2007) for a discussion of this phenomenon, as well as an overview of the notion of focus.
(p.56) that ‘[a]lthough the [linguistic] role of the relationtopic of is widely acknowledged, there is no accepted definition for it and not even full agreement on the intuitions of what counts as topic’; Ebert (2009) (p.19) reacted to this claim by writing that ‘[a]s far as I can sur- mise, the situation is no different 25 years later’. Despite the absence of a broad consensus regarding the nature of sentence topics, I intend to endorse the traditional notion of a topic as identifying what a sentence is about, with an occurrence of a sentence’s having at most one constituent as its topic. I will furthermore accept the position described by Ebert (2009), where ‘topiccan be seen as an entirely conceptual notion, independent of any syntactic, mor- phological or intonational markings. ... However, certain syntactic structures, morphologic markers and intonational means have been argued to indicate the topical status of a con- stituent. Hence topicality istestablevia such means’.
The remainder of this chapter will consist of a development of this rough characterisa- tion.