ESD Test Suite Examples
Someone old (maybe it was Abrams) arrived.
A parenthetical, for our purposes, is a constituent set off from its context by appropriate punctuation (parentheses, dashes, semi-colons) and not integrated with that larger context via ‘ordinary’ syntactic and semantic means. Such constituents have been the focus of debate in the syntactic and pragmatic literature (see Blakemore 2006) regarding the nature of their connection to the sentence. Some authors (e.g., Haegeman 1988) maintain that they are extra-grammatical, while others (e.g., Potts 2002, 2003) argue that they are generated by the grammar but require special treatment in interpretation. For our purposes, given that they are part of the string, the grammar must assign an analysis to them somehow, including integrating them into the MRS. However, that integration is necessarily ‘loose’, reflecting their special status.
On our analysis, the parenthetical predication relates parenthetical clause (represented by the local top handle of the embedded clause) to the host, in this case represented by the intrinsic argument (be it an eventuality or an instance) of the constituent that the parenthetical attaches to. In terms of the scope tree, the parenthetical relation is attached by sharing the handle of the EP related to its syntactic attachment site.
h0:parenthetical[ARG1 i1, ARG2 h2] h0:[ARG0 i1]
Potts (2002, 2003) notes that certain types of parentheticals (e.g., as parentheticals as in They are, as you say, late.) are distinguished by the fact that they give rise to conventional implicatures, rather than providing part of the asserted content. This means that the portion of a structure that is parenthetical must be identifiable from the representation used for interpretation. Our parenthetical predicate, and the fact that this is the only connection between the parenthetical material and the rest of the MRS, should be sufficient for these purposes. Similarly, Blakemore (2006) notes that what she calls ‘discourse parentheticals’ (examples like our testsuite exmaple) are interpreted as separate from their host clauses but make use of the context the host clause provides. Again, the parenthetical predication should provide both the demarcation in the semantic representation and the connection required.
This is (another) case where there is no qeq between the handle-valued argument position (ARG2) of parenthetical and its argument. This represents the claim that it would not be meaningful for quantifiers to be able to scope in between these.
- Blakemore notes that in the literature on the pragmatics of parentheticals, other constructions, including apposition and non-restrictive relatives are also often considered under the same umbrella. Our analyses of these phenomena are not that close. Should they be?
- Move this one to quasi-semantic phenomena?
- Say something more about the choice of ARG1 as a way to anchor this?
The ARG1 can be x-type or e-type. Add an example to the testsuite?
A consultant (maybe it was Abrams) arrived.
- I'm not sure I'm on board with the lack of qeq here ... I'd like the quantifiers to be able to go top-most in the parenthetical, rather than having to ‘escape’ out to the matrix clause. Am I thinking about this wrong?
- The fingerprints as given above don't lead to quite as much highlighting as I'd like in the search interface. Alternative:
h0:parenthetical[ARG1 i1, ARG2 h2] h0:[ARG0 i1] h2:[ARG0 e]
Expert External Commentary
Blakemore, D. (2006). Divisions of labour: The analysis of parentheticals. Lingua 116(10):1670-1687.
Haegeman, L. (1988). Parenthetical adverbials: the radical orphanage approach. In: Chiba, S. (Ed.), Aspects of Modern English Linguistics. Kaitakushi, Tokyo, pp. 232–254.
Potts, C., (2002). The syntax and semantics of as-parentheticals. NLLT 20:623–689.
Potts, C., (2003). Conventional implicatures, a distinguished class of meanings. In: Ramchand, G., Reiss, C. (Eds.), The Oxford Handbook of Linguistic Interfaces. Oxford University Press, Oxford.