the week of february 18

lectures

On Monday, I reviewed the story that had emerged at the beginning of last week. We had seen some support for the hypothesis that common noun roots in natural languages pick out kinds. Japanese-style classifiers and English-style nominal number morphology might then denote functions that map a kind to a set whose members are certain privileged parts of that kind. In this way, we arrived at a unified analysis of Japanese and English noun roots. Here are the slides that formed the background for our discussion. During the second part of Monday's lecture we speculated about how noun roots may have become hooked up with their referents, and the recently observed birth of a new language (Nicaraguan Sign Language) provided us with a realistic backdrop for our speculations. We concluded that as long as we could assume certain conceptual prerequisites, a rather simple 'naming game' could explain how a group of people who have no language in common can converge on a common vocabulary in a very short time. The slides are here. The main puzzle that occupied us on Friday was the fact that humans are able to pick out the kind of thing denoted by a common noun (or common noun root) on the basis of a single specimen of the kind. Most surprisingly, field linguists or children acquiring a language do not seem to make a lot of mistakes in extending a word used to describe a particular specimen to the intended kind. The only possible explanation for this surprising ability seems to be that humans work with a set of constraints that cut down the hypothesis space dramatically. When you teach me the word gavagai by pointing to a particular instance of a rabbit, my default assumption is that gavagai names the species 'rabbit', and not the kind 'mammal' or 'animal', for example. I am also excluding the possibility that gavagai names stuff related to rabbits, including not only the rabbits themselves, but also typical rabbit food like carrots, or the places where rabbits live, like rabbit holes. I am discarding the possibility that gavagai is a name for rabbit parts, like rabbit ears or rabbit legs. And I am not considering the possibility that gavagai may actually be a verb describing hopping movements. The move from specimen to kind may be a theoretical nightmare, but it is a move that doesn't cost humans any effort at all. Here are the slides from Friday's lecture.

lab session

During the lab session we did some actual fieldwork on the German discourse particle ja. You had already read Matthewson's article on semantic fieldwork (PDF) and I started the session by summarizing the points that were essential for us. After I showed you some examples of appropriate and inappropriate uses of ja, you split into two groups with one native speaker each, trying out various hypotheses about appropriate and inappropriate uses of ja. Here is the handout with the instructions and initial examples for your fieldwork. And here are my comments on the fieldwork you did.

take-home one

Take-home one was given out on Friday. A point of clarification: I write "/A/" to talk about the cardinality of a set A. Partee et al. use the notation "|A|" instead. Use my office hours on Monday or Wednesday to ask me any question you may want to ask about the take-home. We will not do e-mail consultations about the take-home, so you have to plan ahead. Florian will be on the road again next week, so he must cancel his office hours on Wednesday. He will be available on Friday for consultations about your projects.

video: chomsky on semantics

If you have a high-speed internet connection, watch this fascinating video (almost two hours): Noam Chomsky on "On Referring revisited". On Referring is a famous article by the philosopher Peter Strawson, but the video is really about semantics in a very broad sense and addresses many of the topics we have already been talking about in this class. Chomsky comes across as fairly combative in this video, but if you listen closely and interpret what you hear as it was intended, most of what he is saying is fairly uncontroversial. Chomsky draws a very helpful comparison between questions that are interesting for a phonetician and what might be interesting questions for a semanticist. Taken literally, some of what Chomsky says may sound shocking, for example his statement that "there is no notion of reference in natural language". But what he really seems to mean (if you consider what he says in the rest of the lecture and the discussion) is that if all you are doing in semantics is saying things like Shakespeare refers to Shakespeare or water refers to water, you are not saying anything interesting, you are, in fact, completely missing out on what could be interesting questions in semantics. We are getting a glimpse into what interesting questions may be when Chomsky talks about the way humans construct the notion of an individual that a name like London refers to, for example, or when he turns metaphysical questions about personal identity into puzzles about human cognition. We discussed similar questions when we talked about shared modes of individuation as a prerequisite for the successful use of proper names by the members of a speech community. The conclusion to draw, I think, is not that we should no longer talk about reference and truth in natural language semantics, hence stop wondering about how humans manage to talk about the world. This would be like telling a phonetician to stop talking about the acoustic properties of sounds and ban all acoustical terms from his or her theories. If you are interested in sound perception, you need to talk about how certain acoustical properties, hence actual properties of sounds, are perceived. If you are interested in semantics, you need to talk about how humans relate linguistic expressions to things in the world that they refer to or situations in which they are true. This is ultimately not incompatible with Chomsky's internalist view holding that the semantics for natural languages would work in exactly the way it actually does if there was no world outside of our brains. If this is what defines an internalist view of meaning, few practicing semanticists would object. Even if there was no world outside our brains, we would still think there was, and native speakers' assumptions about how the expressions of their language relate to the world would be exactly what they actually are.