The short answer: Implications for direct compositionality (and vice versa)

Short answers in Scottish Gaelic and their theoretical implications

“…They can therefore be reasonably considered responses rather than "answers" in a strictly defined sense, since only the short answers are so restricted to use in question-answer pairs. Jacobson (2013) argues that there is additional empirical evidence to indicate that simple fragments are indeed different from larger responses like those in (61b)-(61c), since at least the full replies seem to allow for readings that are not available with the simple fragments. Specifically, with a particular intonation the full replies allow for what she calls a "best-I-can-do" reading, with which the XP corresponding to the whP does not inherit the presuppositions of the whP; thus a best-I-can-do interpretation for (61c) is one where Kodos is not presupposed to be an alien, an interpretation which is not possible with (61a).…”

“…Given that there are various empirical arguments in favour of some sort of grammatical relation between the ellipsis remnant and some incarnation of the antecedent clause, such as case-matching and the P-stranding generalization (see especially Ross 1969, Merchant 2001, Merchant 2004, it seems certain that the fully pragmatic WYSIWYG approach cannot be maintained. 17 Therefore I will focus upon the grammatical WYSIWYG approach, in particular the one proposed by Ginzburg and Sag (2000) and Jacobson (2013). The problem posed by VAs for the WYSIWYG is not obvious on the surface, so we need to delve into the details.…”

“…In order to capture the fact that the single phrase in a fragment is interpreted as a propositional response to the question, Ginzburg and Sag (2000) and Jacobson (2013) posit the existence of a grammatical construction called the Question-Answer Pair (Qu-Ans). The syntax of Qu-Ans is that it takes a Question of category X and combines it with an answer of the same category (a bare XP) to return a larger syntactic object of category Qu-Ans, and the matching of categories ensures that a number of key connectivity effects are accounted for.…”

“…For others (e.g. Groenendijk and Stokhof 1984, Ginzburg and Sag 2000, Jacobson 2013), the answer is interpreted by the use of a grammatical construction specific to the interpretation of question-answer pairs, called the Qu-Ans schema, which takes the short answer category and combines it with the question to provide a proposition. These strands of the WYSIWYG approach are united in proposing that the grammatical information encoded by B's answer is that encoded by the NP Kodos, but the latter strand, what we may call grammatical WSIWYG, goes further in proposing that grammatical information encoded by the antecedent questions is also relevant for the interpretation of the response, since both question and answer are required for the use of the Qu-Ans schema.…”

“…Building substantially on Morgan (1973), Merchant (2004) provides numerous empirical arguments for silent structure and movement in the derivation of fragment answers, showing that fragments display formidentity, connectivity and movement restriction effects, many of which are also seen with similar ellipsis constructions like sluicing (Ross 1969, Merchant 2001). Many of these arguments (or their sluicing equivalents) have been challenged by proponents of the grammatical DI theory of fragments (Sag and Nykiel 2011, Nykiel 2013, Jacobson 2013; see also Griffiths and Lipták to appear), and these authors have argued that they do not in fact provide evidence for silent structure. Whether or not these challenges are substantial enough to undermine the SS theory is an open issue at the moment, 1 but at the very least it should be acknowledged that the argument for the right analysis of fragment answers is there to be won.…”