4.4 The Account and the Three Desiderata
4.4.1 Explaining Definite Variance and ‘There’ Acceptability
It has already been noted that Inherited predicts Definite Variance by proposing that definite and indefinite DPs differ in their propensity to be topical. However, given that the data used to support Definite Variance within the literature uniformly consists of out-of-context sen- tences where the target DP is clause-initial, the following, slightly weaker, proposal is suffi- cient to predict Definite Variance: for sentences in canonical form presented out-of-context, clause-initial definite DPs are highly likely to be construed as topics, whereas clause-initial indefinite DPs display a less robust likelihood to be construed as such. I will remain neu- tral about whether assessors of out-of-context sentences with clause-initial indefinite DPs are slightly more likely to understand the DP as topical than as non-topical, or are no more likely to understand the DP as topical than they are to consider any other topicable item
in the sentence as such. The prediction then emerges that, when an individual assesses a sentence such as (1a), she will be highly likely to imagine a CG that is required to contain an appropriate file card for the clause-initial DP, which results in presupposition failure and an oddness response if she furthermore envisages that the imaginary interlocutors share her belief that the restrictor is empty. On the other hand, when an individual assesses a sentence such as (1b), she is not significantly more likely to construe the clause-initial DP as topical than to construe it as non-topical, with presupposition failure and a sense of oddness arising only if she does construe the DP as topical.
This perspective on the link between a DP’s being clause-initial, definite and a likely topic receives substantial support from the literature. For instance, Comrie (1979) (p.19) claims that ‘subjects tend to be definite, animate, and topic’. Similarly, Aissen (2003) (p.2) holds that ‘[t]he prototypical or unmarked situation is for the transitive subject [...] to be definite, [and] to be topical’. Geurts (2007) (p.266) claims that definite DPs ‘are more likely to be interpreted as topical, especially when they occur in subject position’.33Dalrymple and Nikolaeva (2011) (pp.50-3) argue that ‘most topics are definite’, and that ‘specific indefinites may be interpreted as topics, although not as easily as definite noun phrases’. The simplest way of implementing the proposed link between likely topicality and clause-initial, definite DPs is by endorsing the view that definite DPs differ from indefinite ones insofar as the former include a lexically-encoded preference for topicality. While this preference may be overridden by means of prosody, the QUD or sentence structure, it will mean that assessors display a significant tendency to treat a clause-initial definite DP as topical when no such factors serve to indicate that some other DP should be construed as topical. This completes my explanation of Definite Variance.
Since I have not yet mentioned the attempted explanation of ‘There’ Acceptability given by approaches to the problem of contingently empty restrictors in general, and Inherited accounts in particular, my discussion of this matter requires more detail than the explana- tion of Definite Variance did. I will argue that assessors of out-of-context existential ‘there’- sentences are highly likely to understand the post-verbal DP and its restrictor as non-topical. This provides an explanation of ‘There’ Acceptability: given that the oddness response is attributed to the failure of an existence presupposition, and given that an existence presup- position is predicted to arise and fail only when some part of a DP is topical, it follows that DPs with empty restrictors will be highly unlikely to elicit the oddness response when they occur in the post-verbal position of existential ‘there’-sentences. My argument will begin by considering evidence supporting the view that occurrences of post-verbal DPs and their restrictors are non-topical in the unmarked or default case. I will observe that it does not follow that such items can never be construed as topical, and I will argue that an occur- rence of a coda-less ‘there’-sentence may contain a topical post-verbal DP only if it is issued as an answer to a question that has one of two forms. I will then suggest that prominent counter-examples that have been advanced against the hypothesis that post-verbal DPs are non-topicable may be explained as cases where an implicit QUD reflected by a question with an appropriate form is accommodated. I will conclude that an assessor of an out-of-context existential ‘there’-sentence is unlikely to consider the sort of occurrence that corresponds with the topicability of the post-verbal DP or its restrictor.
Many in the literature endorse the strong position that the post-verbal DP of existen-
tial ‘there’-sentences cannever be topical. Evidence for this strong view emerges from a consideration of occurrences of ‘there’-sentences where attempts are made at rendering the post-verbal DP the obligatory topic. For instance, ‘there’-sentences for which the DP has undergone left dislocation, such as (17), seem unacceptable. Furthermore, it often sounds more natural to respond to a QUD that renders a particular DP the likely topic of an answer with a canonical variant such as (18b), rather than a ‘there’-sentence like (18a):
17. ?As for (some / three / at least three / many) kings, there (are such individuals / is such a group) in the yard.
18. (What property do (some / three / at least three / many) kings have?) (a) ?There are (some / three / at least three / many) kings [in the YARD]F.
(b) (Some / Three / At least three / Many) kings are [in the YARD]F.
This evidence against the topicability of post-verbal DPs in existential ‘there’-sentences may be compared to the behaviour of material consisting of combinations of parts of post-verbal NPs and codas. The apparent acceptability of (19a) – (20d) might be taken to suggest that the NPs and codas of existential ‘there’-sentences are topicable, though this matter will be returned to later:
19. (a) As for kings, there are (some / three / at least three / many / few) of them in the yard.
(b) As for the yard, there are (some / three / at least three / many / few) kings in it. (c) In the yard, there are (some / three / at least three / many / few) kings (there). (d) As for kings in the yard, there are (some / three / at least three / many / few) of
them.
20. (a) (What property do kings have?)
There are [(SOME / THREE / at least THREE / MAny / FEW)]F1 kings [in the
YARD]F2
(b) (What property does the yard have?)
There are [(some / three / at least three / many / few) KINGs]F1 [in]F2 the yard.
(c) (What is in the yard?)
There are [(some / three / at least three / many / few) KINGs]F in the yard.
(d) (How many kings are in the yard?)
There are [(SOME / THREE / at least THREE / MAny / FEW)]F kings in the
yard.
Due to this sort of data, Reinhart (2004) (p.60) claims that ‘all studies agree’ that the post- verbal DP of an existential ‘there’-sentence can never be topical. While it is frequently held that occurrences of existential ‘there’-sentences are always all-focus (e.g. Kuno (1972), Kuroda (1972), Babby (1980), Lambrecht (1994)), some have argued that the codas of exis- tential ‘there’-sentences are obligatorily topical (e.g. Rando and Napoli (1978)) or optionally topical (e.g. Bentley et al. (2012), Cruschina (2012), Cruschina (2015)).
However, the literature also contains a weaker position, which holds that the unmarked option is to understand occurrences of post-verbal DPs as non-topical. One advocate of this weaker view is Leonetti (2008, 2016), who claims that occurrences of coda-less ‘there’- sentences are all-focus in the ‘unmarked case’, but that contextual factors may bring about different choices of IS, including the topicality of the post-verbal DP. Compelling evidence for this weaker position is provided in Rando and Napoli (1978), where it is argued that the two most common prosodic patterns for existential ‘there’-sentences both involve a falling pitch accent on the NP, as in the following examples:34
21. (a) There are some KINGs in the yard. (b) There are some KINGs in the YARD.
It is reasonable to classify these prosodic patterns as neutral prosody for existential ‘there’- sentences. Clearly, such prosody is incompatible with the topicality of the DP or its restric- tor. This suggests that the unmarked option is to understand occurrences of post-verbal DPs and restrictors as non-topical, but that deviations from neutral prosody may render these items topical. The weaker position observes that data such as (17), (18a) and (18b) simply show that attempting to force the topicality of post-verbal DPs often sounds un- natural, without demonstrating the impossibility of construing post-verbal DPs as topical in certain settings. For example, the challenge of coming up with an indefinite expression anaphoric with the DP in (17) (i.e. the definite ‘them’ will not suffice) might be used to ar- gue that left-dislocating a DP from a ‘there’ construction generates difficulties independent of its putative non-topicability. Similarly, (18a) is improved if it occurs relative to a slightly different question that continues to render the post-verbal DP of the answer a likely topic, such as ‘Where are there (some / three / at least three / many) kings?’.
I will accept this weaker position. Nevertheless, I wish to identify the circumstances in which the post-verbal DP of an occurrence of an existential ‘there’-sentenceisa possible topic, in order to confirm their markedness and explain some prominent counter-examples to the strong view that post-verbal DPs are never topical. My claim is as follows, where ‘β’ and ‘β0’ represent optional post-verbal material that is possibly NP-internal or NP-external, and where ‘Wh’ represents a ‘wh’-phrase appropriate for ‘β0’:
Topicable Post-Verbal DPs: For an occurrence of an existential ‘there’-sentence with the form ‘There be Det N (β)(β0)’, ‘Det N’ is a potential topic when the occurrence is congruent with a QUD reflected by either: (i) ‘Be there Det N (β)(β0)?’, or (ii) ‘Wh be (there) Det N (β)?’.
For example, ‘some kings’ is a reasonable candidate for the topic of the responses in (22a) – (23b):
22. (a) (Are there some kings?) There [ARE]F some kings.
(b) (Are there some kings living in New York?)
34Rando and Napoli (1978) claim that the most common prosodic pattern forcontextualised‘there’-sentences
There [ARE]F some kings living in New York.
23. (a) (Where are there some kings?)
There are some kings [living in New YORK]F.
(b) (Where are some kings living?)
There are some kings living [in New YORK]F.
It is important to note that the post-verbal DPs are possible but not obligatory topics of the responses to (22a) – (23b). That is, there is always some alternative choice of topic compatible with the QUD, such as the restrictor (e.g. ‘kings’) or an additional DP (e.g. ‘New York’). However, the crucial point is that the prosody of the responses to (22a) – (23b) is marked, by virtue of diverging from neutral prosody for existential ‘there’-sentences.
The idea that the post-verbal DP of an occurrence of a coda-less ‘there’-sentences is un- derstood as topical only relative to an appropriate QUD explains prominent examples that have been advanced in the literature in order to oppose the view that post-verbal DPs are never topical. For instance, McNally (2011) (p.1834.) provides the following purported counter-example to the strong view, where the left dislocation of ‘a solution’ in the second sentence of the conjunction renders it the obligatory topic of that sentence:
24. They told us there was a solution, and indeed a solution, there was.
It is clear that the first sentence of the conjunction plays a crucial role in licensing this ap- parent counter-example. That is, infelicity emerges when the second conjunct opens a dis- course (as in (25a)), and when an alternative first conjunct that includes the DP is chosen (as in (25b)):
25. (a) ?(Indeed) a solution, there was.
(b) ?(We discovered that / We didn’t care if) there was a solution, and indeed a solution, there was.
A consideration of the type of sentences that license the immediate occurrence of McNally’s second conjunct indicates that they are ones that suggest a search for a solution without resolving the matter of whether or not a solution turned out to exist. Such sentences are likely to induce individuals to accommodate a QUD expressed by a sentence of the form ‘Was there a solution?’, to be answered by subsequent utterances. It follows that the first conjunct of McNally’s purported counter-example to the hypothesis that post-verbal DPs are non-topicable invites the accommodation of an unanswered polar QUD pertaining to the existence of a solution, with the DP ‘a solution’ therefore being licensed as the topic of the subsequent answer.
Similarly, Borschev and Partee (2002) present examples of negative existential sentences in Russian for which they take the DP to be the theme. They define ‘theme’ as ‘what is being talked about in a sentence, which is presupposed to be familiar to the hearer’, which accords with my notion of a ‘sentence topic’. Their key example is the following:
(I looked-for kefirACC.M.SG.) KefirGEN.M.SGin storeN EGwasN.SG.
‘(I was looking for kefir.) There wasn’t kefir in the store’.
The rules governing word order and prosody in Russian have been repeatedly argued to de- pend on theme-rheme structure (e.g. Sgall et al. (1986)), and the relevant principles classify ‘kefira’ as part of the theme. Borschev and Partee (2002) therefore conclude that the rheme of (26) is ‘ne bylo’ (‘N EGwas’), with ‘v magazine’ (‘in the store’) joining ‘kefira’ as part of the theme.
It is possible to challenge this view of the IS of (26). As observed by Borschev and Partee, Babby (1980) argues that word order is no longer an indicator of theme in Russian negative existential sentences, as such sentences are always all-rheme. However, if the topicality of the bare DP is accepted, then this is again explained by my observation about the circum- stances that license topical occurrences of post-verbal DPs. That is, the context is stipulated to include the information that the real or imaginary speaker was searching for kefir, with- out resolving whether or not she found kefir. Provided an assessor understands the store to be a salient searching location, it follows that the assessor is induced to accommodate the implicit QUD expressed by a sentence of the form ‘Was there kefir in the store?’. This licenses the topicality of the subsequent occurrence of ‘kefir in the store’.
In sum, I have argued that the unmarked option is to understand occurrences of the post-verbal DPs and restrictors of existential ‘there’-sentences as non-topical. The most sig- nificant evidence supporting this position is the identification of neutral prosody for exis- tential ‘there’-sentences provided in Rando and Napoli (1978). In light of my acceptance of the view of Fodor (2002), which holds that assessors assign to out-of-context sentences ‘the most natural (default) prosodic contour for the construction’ (see §(2.3.1)), it follows that an assessor of an out-of-context existential ‘there’-sentence would be highly likely to consider an occurrence that has neutral prosody and assign IS in a manner consistent with such prosody. Of course, the post-verbal DP of an occurrence of a ‘there’-sentence may be understood as topical when it is issued relative to an explicit QUD reflected by one of the two types of question detailed above, or clues that induce the accommodation of an implicit QUD reflected by one of the two types of question; however, out-of-context presentations of ‘there’-sentences will lack such features. In light of the topic-sensitive, presuppositional approach to the problem of contingently empty restrictors, an explanation of ‘There’ Accept- ability immediately follows: DPs with empty restrictors will be highly unlikely to elicit the oddness response when they occur in the post-verbal position of out-of-context existential ‘there’-sentences.
I conclude that my topic-sensitive, presuppositional account meets all three desiderata for a solution to the problem of contingently empty restrictors. Firstly, it attains descrip- tive adequacy: it predicts assessors’ responses to out-of-context presentations of sentences with perceived contingently empty restrictors, in addition to predicting assessors’ responses when such sentences are issued relative to an indication of their prosody or the QUD. Sec- ondly, it attains explanatory adequacy: Definite Variance and ‘There’ Acceptability are ex- plained by my claims that assessors of out-of-context sentences will be highly likely to con- strue clause-initial definite DPs as topical, will possibly construe clause-initial indefinite DPs as topical, and will be highly unlikely to construe the post-verbal DPs or restrictors of existential ‘there’-sentences as topical. While the explanation of Definite Variance is not par-
ticularly ‘deep’ – it is based on the hypothesis that definite DPs include a lexically-encoded preference to be understood as topical that emerges when they occur as clause-initial – the explanation of ‘There’ Acceptability derives from the independently motivated views that assessors of out-of-context sentences assign a choice of IS compatible with neutral prosody, and that neutral prosody for existential ‘there’-sentences involves a pitch accent on the post- verbal NP. Finally, my account attains generality: it relies upon independently motivated views pertaining to a system of file cards, the existence presuppositions associated with topics, and the likely IS of out-of-context and contextualised occurrences of sentences.