Quotation via Dialogical Interaction

Everett (2012) claims that Pira h̃ a lacks indirect quotation.

We limit our discussion and treatment here to ‘pure’, ‘direct’ and ‘indirect’ quotation, putting aside the recently much discussed ‘mixed’ and ‘scare’ quotation. We believe the account we develop can be extended to these, but that remains a future project.

The locus classicus for this view is Quine: ‘Take quotation marks: applied to any sort of expression, what they produce is a singular term (naming, as it happens, the expression inside)’ (Quine 1960). Geurts and Maier (2005) say: ‘If the quotation is a ‘pure’ one, it denotes a linguistic expression and its semantic type may differ from that of the quoted expression. For example, in [the example ‘Monosyllabic’ is not monosyllabic.] quotation maps the semantic type of the adjective ‘monosyllabic’ to that of singular terms, and the resulting expression denotes the word ‘monosyllabic’.

See also the following from the economist Brad DeLong:

It is conventional–at least in the circles in which I travel–to clean-up transcripts. Raw human speech looks silly in print. You eliminate the “ummms” and the “you knows” and the false starts. You collapse into one grammatical sentence those times when people start a sentence, go back and start it again, and then go back and start it again and finally finish it.

The convention is to edit the transcription of what they said into something different–into what they would have written if they had been not speaking but writing, sentence by sentence and paragraph by paragraph, and editing as they go.

But this seems to go rather beyond that, so much so as to convey a substantially distorted impression of how the symposium actually went \(\ldots \)

http://​delong.​typepad.​com/​sdj/​2013/​12/​thursday-idiocy-weblogging.​html.

Note that \(r\) may contain more objects with labels not mentioned in the type.

The convention is that a field

is used as a convenient notation for

where \(T_a\) is the type \(T\) restricted to a singleton type whose only witness is \(a\). This is why the apparent asymmetry between the ‘phon’ and ‘c-params’ fields on the one hand, and the ‘cat’ and ‘cont’ fields is merely notational.

\(Moves\) is the component of the dialogue gameboard where grounded utterances are represented.

Of course there is a bit of a chicken/egg issue here, given that direct quotation is one of the mechanisms that gets language off the ground.

In order to accommodate quotations of non-linguistic events, we would simply need to modify our characterization of the type of the head daughter as being grammatical to a somewhat wider class of types. Such types would certainly have a field for \(phon\) and might also have a field for \(cont\).

An anonymous reviewer for the Journal of Logic, Language, and Information asks whether by invoking similarity measures we are not rendering our proposal to be unfalsifiable. S/he writes ‘Suppose that John gave a lecture, and somebody asks Peter about the lecture, how it was. Peter: John said: ‘blablabla’. Can ’blablabla’ stand for the content of John’s lecture? If not, how is it blocked in the model?’ On our account the truth of Peter’s report requires an utterance of ‘blablabla’ to be similar to the sequence of utterances that made up John’s lecture, relative to a contextually given similarity measure. A number of similarity measures we have mentioned above would not accommodate such similarity, e.g., one based on identity of content and a fortiori on identity of form. In fact, a measure of similarity that would make Peter’s report true would need to be very coarse grained, essentially one that made all utterances of a given language similar, as for instance when one hears a lecture in a language one does not understand or when one wishes to convey that, for all intents and purposes what someone has said is contentless.

Additional axioms need to be met but they need not concern us here.

And to this one should add, as noted by Bonami and Godard (2008), the existence of predicates incompatible with direct quotation complements. But we will not attempt to capture this pattern here, which extends to verbs of cognition—an analogy much emphasized by Recanati.

And we could extend this somewhat to allow for speech events like, say, grunts that might not be illocutionary in a strict sense.

For a more detailed discussion of GM’s approach, also developed in Maier (2007) and subsequently, see De Brabanter (2010).

See Asher (2011) for such a type theoretic framework, though it is not integrated with the discourse semantics, in contrast to TTR/KoS.