Wednesday, 11 May 2011

Exciting new directions in formal semantics

(Cross-posted at New APPS)

The Amsterdam Colloquium is a bi-annual event focusing mostly (though not exclusively) on formal semantics and formal pragmatics. Its 18th installment will be held December 19 - 21, 2011 at (surprise, surprise!) the University of Amsterdam. The call for papers is out, and there will also be three thematic workshops: Formal semantics and pragmatics of sign languages, Formal semantic evidence andInquisitiveness.

Especially the first two workshops indicate that formal semantics and pragmatics as a field is moving in refreshingly new directions (and the third one is also bound to be interesting). The workshop on sign languages shows that the field is finally paying attention to the significant dissimilarities between languages expressed in different media: written, spoken, and in this case sign language. Following Roy Harris and others, I am convinced of the crucial importance of incorporating these differences into linguistic theorizing. To my mind, it is deeply misleading to speak of ‘natural language’ as a blanket term covering both speech and writing (and possibly sign languages too). Moreover, I am with Linell in identifying a chronic ‘written language bias’ in language studies in general, which means that features proper to speech (intonation, prosody) tend to be overlooked. Ironically, features proper to writing are also unduly ignored (as argued e.g. in this great paper by Sybille Kramer).

A workshop on sign language clearly indicates the realization that each form of human language must be studied also from the point of view of its specific medium and the features arising from using the medium in question (clearly, one of my motivations to insist on this aspect is my belief in the fruitfulness of embodied approaches in general). There is some work already being done within formal semantics and pragmatics on e.g. intonation (for instance, by my colleague Floris Roelofsen), and now that sign language is receiving attention in the field, it all seems to be going in a very good direction. Moreover, given that most researchers are not competent users of sign language themselves, such studies will necessarily have a strong empirical component, which brings me to the second workshop.

Working in Amsterdam, I’ve always been exposed to large amounts of formal semantics and pragmatics, and while I’ve always marveled at the technical ingenuity displayed, philosophically I always felt a bit uncomfortable with what exactly was going on. What is formal semantics a model of? Does it purport to describe actual cognitive processes of language users? Does it describe the ‘mathematical’ properties of language, considered as ontologically independent from speakers? (Roughly, this seems to have been Montague’s original take.) And as long as it is not clear what exactly the target phenomenon is for these theories, it is also not clear where the evidence to support or refute a given formal semantic theory should be coming from. So I worried about the kind of epistemic confirmation these theories were based on, and similarly about the explanatory power they could have (what exactly were they explaining?). In other words, I felt that formal semantics and pragmatics as a field was in dire need of serious methodological reflection. I was not alone there, as my boss Martin Stokhof, formerly a full-blown formal semanticist (in particular in his joint work with Jeroen Groenendijk: the logic of questions, dynamic predicate logic etc.), has focused extensively on the philosophical and methodological foundations of the enterprise in recent years, taking a rather critical stance (see his papers here; he is also supervising the dissertation of my friend and co-author Edgar Andrade-Lotero on the philosophical foundations of formal semantics).

Now, given the reservations that I’ve had through the years, I am thrilled to see a whole workshop at the Amsterdam Colloquium dedicated to methodological reflection on the foundations of formal semantics (not surprisingly, it is organized by my very talented colleagues Katrin Schulz and Galit Weidman Sassoon). From the description of the workshop:

Formal semantics as a field of linguists undergoes a rapid change with respect to the status of quantitative methodologies, the application of which is gradually becoming a standard in the field, replacing the good old 'armchair' methodology. In light of this development, we invite submissions reporting of high level formal semantic research benefiting from the use of a quantitative methodology, corpora-based, experimental, neurolinguistic, computational or other.

Clearly, this is a debate with deep philosophical implications, and particularly significant against the background of the sustained debates on methodology that have been taking place in philosophy in general over the last years. In other words, this post is not only intended as a plug for the Amsterdam Colloquium :) It is also intended to suggest that the field of formal semantics and pragmatics may be moving in exciting new directions; to my mind, if these directions continue to be pursued, the field will gain substantially in philosophical depth.

7 comments:

  1. Hi Catarina, this is a teeny bit of self-promotion, I suppose, but I gave a talk on this general topic at Edinburgh's Linguistics Society last October,

    http://langsoc.eusa.ed.ac.uk/blog/?p=370

    There's even a video of it. My main argument there is that one needs to distinguish between:

    (i) the individuation of languages (which I think should be extremely fine: the tiniest changes makes a distinct language - my idiolect, for example, is not identical to Roy's idiolect);
    (ii) the cognitive relations of speakers/minds to the language they speak.

    ReplyDelete
  2. Very a propos self-promotion, if you ask me! :) You mentioned before that you were interested in individuation conditions for language, and I entirely agree with you that it's going to be an *extremely* fine-grained story, to the point that hardly anything is going to count as 'the same language' as something else. (A plug for a plug! Here are some incipient thoughts of mine on the topic:
    http://www.newappsblog.com/2011/01/language-variation-and-the-analytic-continental-divide.html )
    And a lot of the methodological mess in formal semantics seems to me to stem precisely in the failure to distinguish different projects, such as describing the 'mathematical' properties of languages as ontologically autonomous entities, and describing the cognitive processes underlying language use by humans (or potentially other beings). I tend to think these are going to be very different stories, and I am personally more interested in the cognitive one.

    ReplyDelete
  3. Hi Catarina, thanks, yes. I think we should separate these two projects sharply, and call them (i) the Individuation Problem and (ii) the Cognizing Problem. I treat languages as mathematical entities, involving mixed functions (e.g., functions from strings to meanings, extension; rues for assertion; etc). This view appeared first in papers by Lewis and Soames, I think.

    As you say in your New APPS blogpost, the second, cognizing, problem is up to its neck in Kripkensteinery. In my talk, I called it "the hard problem of theoretical linguistics"; I think progress on this lies in the distant future cognitive science.

    ReplyDelete
  4. It's really weird, there were two comments here (by me and Jeff) which simply disappeared! Does anyone know what is going on?

    ReplyDelete
  5. Hi Catarina, I noticed they'd gone too. Very puzzled.

    ReplyDelete
  6. As it turns out, I have the missing replies in my inbox, so let me copy them here. First my reply to Jeff:


    Very a propos self-promotion, if you ask me! :) You mentioned before that you were interested in individuation conditions for language, and I entirely agree with you that it's going to be an *extremely* fine-grained story, to the point that hardly anything is going to count as 'the same language' as something else. (A plug for a plug! Here are some incipient thoughts of mine on the topic:
    http://www.newappsblog.com/2011/01/language-variation-and-the-analytic-continental-divide.html )
    And a lot of the methodological mess in formal semantics seems to me to stem precisely in the failure to distinguish different projects, such as describing the 'mathematical' properties of languages as ontologically autonomous entities, and describing the cognitive processes underlying language use by humans (or potentially other beings). I tend to think these are going to be very different stories, and I am personally more interested in the cognitive one.

    ReplyDelete
  7. Now Jeff's rejoinder:

    Hi Catarina, thanks, yes. I think we should separate these two projects sharply, and call them (i) the Individuation Problem and (ii) the Cognizing Problem. I treat languages as mathematical entities, involving mixed functions (e.g., functions from strings to meanings, extension; rues for assertion; etc). This view appeared first in papers by Lewis and Soames, I think.

    As you say in your New APPS blogpost, the second, cognizing, problem is up to its neck in Kripkensteinery. In my talk, I called it "the hard problem of theoretical linguistics"; I think progress on this lies in the distant future cognitive science.

    ReplyDelete