Re: RT list: VS: Ray Gibbs on logical form

Ray Gibbs observes that there is no need to posit a relevance-theoretic
concept of logical form, and Thorstein Fretheim asks whether there is a
framework to hand which replaces the concept of logical form as a
context-independent output with a competitive alternative. As Chris
Lucas has already pointed out, the answer is that indeed there is.
This is the Dynamic Syntax framework of Kempson et al 2001, Cann et al
2005, Cann et al 2010. Dynamic Syntax involves an explicit
characterisation of ellipsis and split utterances of the type discussed
in this debate; and those of us working on this framework have been
advocating for some time that no concept of logical form is warranted.
On this view syntax itself is the set of mechanisms for incremental
projection of a propositional formula as identified in context, with
both speakers and hearers projecting a propositional structure relative
to their own context (with actual choices being determined by relevance
constraints) The shift of perspective is radical in that one aspect of
"performance" is incorporated into the model of grammar -- the concept
of incrementally projecting structural representations of content in a
way that reflects real-time dynamics. So underspecification and
mechanisms for progressive update become core parts of the grammar.

As noted by Chris Lucas, contributors to this debate might like to
access the Dynamics of Conversational Dialogue website, where papers
accumulated during the progression of our research project of this
name, along with other Dynamics Syntax related papers:
> http://www.kcl.ac.uk/research/groups/ds/publications.html

Ruth

  On 16/02/2011 12:32, Thorstein Fretheim wrote:
>
> ------------------------------------------------------------------------
> *Fra:* Thorstein Fretheim [mailto:thorstein.fretheim@hf.ntnu.no]
> *Sendt:* 16. februar 2011 13:26
> *Til:* 'relevance@linguistics.ucl.ak.uk'
> *Emne:* Ray Gibbs on logical form
>
> Dear all,
> I was thrilled by Ray Gibbs' submission to the RT list on February 15.
> He asks a rhetorical question, "Is there ANY empirical/experimental
> evidence to suggest that people compute anything like a 'logical form'
> during their online understanding of language?", and he concludes
> that, "RT works just fine without needing to posit logical forms as a
> starting place for linguistic understanding". I am sympathetic to the
> idea that 'logical form' with its current properties, understood as
> the interface between language and communicated thought, should be
> abandoned, but what is it to be replaced by? The demise of 'logical
> form' was argued for quite explicitly by Noel Burton-Roberts in his
> paper "Varieties of semantics and encoding" in the volume /Pragmatics/
> that he himself edited (Palgrave, 2007). The big question is what
> you're going to put in its stead. Like Recanati who has criticized the
> concept of 'logical form' in RT in "Pragmatics and logical form", a
> paper that appears in his new book /Truth-conditional Pragmatics/
> (Oxford U.P., 2010), I am not so sure that I'm ready to renounce
> *semantic composition*. Is Gibbs prepared to do that?
> Recanati presumes that semantic composition is a context-dependent
> process and that a workable notion of 'logical form' cannot be the
> relevance-theoretic, context-free "proposition schema", a level of
> representation that is semantically gappy so that it doesn't count as
> a "propositional form". Rather, a logical form must be a
> post-pragmatic level for pragmatic modulation or enrichment to operate
> on, says Recanati. Free ("top down") pragmatic processes must be
> allowed to impact on the input to rules of semantic composition.
> Exactly how different from the encoded linguistic semantics must this
> input to semantic composition be in order to meet Gibbs' reminder to
> us in yesterday's mail: "Language is processed in a dynamic,
> incremental fashion where the build-up of meaning occurs
> moment-by-moment without any need to start off with some abstract,
> logical structure onto which one can build linguistic and contextual
> meanings." ?
> I believe that semantic composition does not operate on 'logical
> form' understood as the output of the grammar module. In a couple of
> forthcoming papers - "Description as indication: the use of conceptual
> meaning for a procedural purpose" and "Relevance Theory and Direct
> Reference Philosophy - a suitable match?" - as well as one published
> paper, "Conceptual pointers to antecedent information and reference
> resolution" (in /Meaning, Content and Argument/, ed. by J.M Larrazabal
> & L. Zubeldia, University of the Basque Country Press, 2009), I
> advocate a view that I believe to be keeping with Direct Reference
> philosophy (cf. Kaplan 1898 on "Demonstratives", but also Kripke,
> Perry, and Recanati's book /Direct Reference/, Blackwell, 1993) and
> with George Powell's approach to reference in his important book
> /Language, Thought and Reference/ (Palgrave, 2010). In my opinion we
> have a reason to lament the fact that RT permits no encoded conceptual
> meaning to be eliminated as input to those pragmatic processes of
> enrichment that are necessary for the hearer's 'mental representation'
> a fully-fledged proposition. .Powell makes an interesting distinction
> between a speaker's "informative intention" and her "derivational
> intention" (2010: 29), saying that, "... separate from the informative
> intention itself, a speaker will have intentions concerning the route
> via which her audience should reach the content of her informative
> intention." The (primary) role of certain linguistic items is to give
> an indication of the speaker's derivational intention in Powell's
> sense. Such items encode a "procedural" meaning in the sense of
> Relevance Theory. In my recent work on reference I discuss sets of
> natural language data which I believe to show that in referential
> expressions even lexical items and descriptive phrases that do encode
> a conceptual meaning (according to the standard view) deliver nothing
> to the content of the utterance, though they may provide important
> information about what is in Powell's terminology the speaker's
> "derivational intention"; in other words, they offer exclusively
> procedural information for the benefit of the hearer, in spite of the
> conceptual meaning that the output of the grammar module provides.
> Best regards,
> Thorstein Fretheim
> Norwegian university of Science and Technology
Received on Wed Feb 16 14:32:41 2011