News and views about philosophy, the academic profession, academic freedom, intellectual culture, and other topics. The world’s most popular philosophy blog, since 2003.
My former colleagues at another university in Middle East have also been moved to online teaching indefinitely, with the students…
If much of the interest of high-quality papers lies between the lines—in the metaphorical fire that a paper lights in…
I would also recommend that potential grad students make inquiries into how far the compensation package actually goes towards cost…
It’s a mix. I’m still in the UAE with my family, and we feel safe. But some students and faculty…
In the above comment, Michel wrote: “As an aside, every once in a while I check out how the chatbots…
I could imagine LLMs having saved me a *ton* of time in graduate school–e.g., by having supplied reasonable answers to…
The McMaster Department of Philosophy has now put together the following notice commemorating Barry: Barry Allen: A Philosophical Life Barry…
MOVING TO FRONT FROM SEPTEMBER 28–A VERY INTERESTING COMMENT FROM A LINGUIST IN RESPONSE TO MY (ADMITTEDLY) OPEN-ENDED REQUEST. I HOPE OTHERS KNOWLEDGEABLE WILL WEIGH IN NOW.
The interview continues (earlier). I would be interested in reactions from readers knowledgeable about these issues.
Your expression of interest deserves a comment, but so far nobody has commented; I can't say I blame the world, because to start out on such a comment immediately opens up a multitude of fundamental questions that would take hundreds of pages to say anything sensible about on even an introductory level. I'm not a philosopher, but a linguist, and not a Chomskyan but a descriptive linguist, thus one of those Hornstein says are not seeing a need to follow what Chomsky and his followers have been saying recently, because his research program is pretty much regarded as irrelevant to our concerns, which have to do with the description of natural language systems. As a "young lion", however, Chomsky's main contribution was to wrench linguistics out of its straightjacket of Machian phenomenalism and to focus on a unified generative system of principles that made possible the phenomena that linguists recorded as data in their notebooks, although Chomsky quickly attended only to the "idiolect" of an individual speaker, where descriptivists would have preferred a focus on the norms belonging to a speech community, which any new individual speaker meets as a preexisting condition; and would have insisted that, unlike Chomsky, making this distinction is important, because the descriptivist is interested in how this community-level norm came to be what it is. So for a descriptivist, the Chomskyan program is a highly developed exploration of fundamental principles that are insufficient to be regarded as underlying any quest for a full understanding of language as a human phenomenon.
The main reason for this is that, in spite of his interest in the mental and the generative, he has remained constrained by the Machian views of language prevalent in that era, influenced by people like the early Hilbert and Carnap, and perhaps impressed by Turing. The fundamental assumption that most stands in the way of a full understanding of language is his dogma of the "autonomy of syntax", where "syntax" is understood as it was by the Formalists, Logical Empiricists and the developing field of formal languages, i.e., as pertaining to an uninterpreted system of inscriptions, the "semantic" interpretation of which is provided by a translation into a model. It doesn't have to be this way. (The problems of this approach were hashed out at the time in the Frege-Hilbert controversy.)
The descriptivist, however, especially in the field research context, has to be interested in concrete language activity not just as the production of noises, but as the production of further significance, otherwise known as "expressing something". Chomsky himself has said that the function of language is to express thought; I would agree with that, but Chomsky never says anything about the "thought" expressed, let alone "thought about the world", which is what thought mostly is. "How is it possible for a system of noises to "express thought"? (and let's take "thought" in the Fregean sense) is not a question Chomsky ever addresses. "What makes possible the production of the thought expressed?" is not even a possible question for Chomsky. The descriptive linguist notices that if you have a sentence in L1 and a translation equivalent in L2, what is expressed by the two sentences is not completely equivalent; there are interesting differences. Where do the differences come from?
Michael Dummett asked, "What is it for linguistic expressions to have senses?" We could focus the puzzle further and ask, "What is it to express something significant?" How does that work? If we're not addressing questions like these we will never have a full understanding of language as a human phenomenon. According to Hornstein, Chomsky has remained uninterested in questions of what is expressed by sentences or questions about the relation between what is expressed by sentences and the world. E.g., in his dismissal of the "myth of the referentialist doctrine", instead of knocking down the straw man of the simplistic account from an introductory textbook, he could have engaged with philosophers who have taken the question seriously, like Frege and Husserl, or in recent times Dummett or Putnam.
So I have a question for Prof. Hornstein: I have always wondered, wrt basic operations like "Merge", on what level of analysis do these operations take place, out there in the world, and wrt the entities that enter into these operations, what sorts of entities are they, out there in the world? The symbols and the manipulations that constitute "computations", the "features" that are "checked", etc., are these biological-level events and objects (the "cognitive" seems to be an extension of the "biological" for Chomsky)? If so how does all of that work, and what is the mechanism that connects that to what is expressed? These are, in spite of the expressions of defeatism from Prof. Hornstein, possible empirical questions. An approach that aims to be empirical should be able to say something about the referents of basic terms, out there in the world.
Alas, my being bothered by the Hornstein interview has resulted in an excessively long comment that I'm hesitant to submit in this philosophical forum. But in any case, I do think that a sensitivity to historical philosophical concerns by descriptive linguists could result in interesting insights about the relation between systems for expressing meaning (languages) and the world.
Very nice, very much appreciated.
There is a sort of simple-minded if not very argumentually effective response, no? The danger is that focussing on such formulations as "What is it to express something significant?' or "How is it possible for a system of noises to "express thought"?" might be to insist on a chimera. If we had a complete Chomsky-style grammar for an individual — that is really just 'syntax', including the fine-grainded internalist theta-roles to sort out agent, patient etc — and the individual were 'plugged into' a context (say a Lewisian possible world), is there something left out, the 'thought'?
A brief response to just two points:
1. On the question of the relationship between language/mind and the world, much work has been done on this topic from a Chomsky-like perspective (Pietroski, Glanzberg,and, modesty permitting, myself). Indeed, Chomsky'd position is very much akin to so-called 'pragmatism' and enjoys deep connections with relevance theory and even Charles Travis's brand of pragmatism. Needless to say, Chomsky has been discussing these issues with thorougher reference to the major philosophical thinkers for the past 60 years. I don't see any straw men hereabouts.
2. Merge, as a basic combinatorial operation, is treated as real insofar as the hypotheses involving it prove explanatory. It would be nice to have a higher resolution account of how the brain is sensitive to the operation (there is work in that direction), but the absence of some such account hardly denudes the relevant hypotheses of their explanatory power, just as a lack of understanding how gravity comes about didn't damn Newton or Einstein, still less the current efforts in that direction. Merge might, to be sure, turn out to be a formal artefact. I consider this kind of situation to be perfectly normal in the history of science. Only in the human sciences, it seems, are hurdles set before one that have no place elsewhere.
"Nqabutho" claims that Chomsky has not addressed Putnam's views on reference. In fact, A substantial chunk of Chomsky's New Horizons In the Study of Language & Mind (2000) is devoted to rebutting Putnam on reference.
I'm a linguist trained in a deeply Chomskyan milieu, though my research focuses mainly on non-Chomskyan areas, like the production and perception of speech (which Chomsky's just not that interested in, because he doesn't think it includes much language-specific machinery). I think Nqabutho and John Collins have both given really good answers, but I want to elaborate a bit on Nqabutho's ending question and John Collins' response in his point (2), which I think is pretty crucial here. Merge as I understand it is not meant to be 'real' in any typical psychological or neurological sense. John Collins gets at this, but kind of obliquely. Merge is a mathematical function that (Chomsky claims) is useful for describing the kinds of computational systems that languages are, and for failing to describe the kinds of computational systems that languages aren't (e.g., the famous example of a language that forms questions by reversing the order of the words in the corresponding declarative). It is nothing more or less than that. There is no claim that individual humans processing or producing an utterance in real time go through any process that looks like Merge. But their static, long-term implicit knowledge of which utterances are and are not acceptable in their language is claimed to reflect systematic generalizations that can be modeled with a generative grammar using Merge.
This is a great topic for a thread. I've only read through the interviews quickly so far, and not yet any of the associated papers; I'm looking forward to reading it all carefully and will comment here if I think there's anything further to add. I hope others comment on this thread, since I imagine many readers readers of this blog are very knowledgeable about the areas of linguistics, philosophy of language, and cognitive science that Chomsky has worked on (and/or founded). As a onetime descriptive/historical linguist now working in theoretical computer science, I recently returned to Chomsky's linguistics work, starting from the mid-50s papers (which are incredible) and making my way through the main works up to the present. I used to feel very similarly about Chomskyan linguistics to Nqabutho in comment #1, and I still believe Chomskyan linguistics has little to offer when it comes to many aspects of descriptive linguistics (for instance, the important endeavor of doing empirically accurate lexical semantics, giving the most accurate descriptive account of what/how particular words and phrases mean in a given language). However, now that I've studied Chomsky's works chronologically, with an eye to which questions he is (and is not) trying to answer and to how his theories have evolved, I now see "the point" and I find his approach extremely compelling. (And his criticisms of current approaches to Natural Language Processing and neural-net based AI, which stem directly from his fundamental views on what the language faculty is and how to study it, are also spot on in my view.)
For those who are super interested in these topics, I find this obscure (and very long–19 hours, and Chomsky has lost his voice) set of talks "Language and the 'Cognitive Revolutions'" from 1992 to be a wonderful resource:
Chomsky discusses the issues mentioned by John Collins and Ritwik Agrawal, and a lot else, with great clarity. He deals with many of the core issues in philosophy of language and cognitive science though his criticisms are so extreme that I'm sure that many philosophers of language will not agree with them. I find them wholly convincing, but I am not a specialist in philosophy of language so am not the best person to assess them.
There isn't anything I would want to disagree with here, but I think a wrinkle is appropriate. There is a distinction between processing/performance and competence. Merge, qua a set formation operation, squarely falls under the latter, but will have potentially real effects on processing (e.g., https://global.oup.com/academic/product/minimalist-parsing-9780198795087?cc=gb&lang=en&). Setting aside this distinction, there remains the question of what aspect of a competence theory is merely notational and what aspect is tracking something 'real'. This issue was discussed by Chomsky and Harman in the Replies to Chomsky's BBS target article 'Rules & Representations' (1980), and is taken up in an intriguing paper by Kent Johnson (https://onlinelibrary.wiley.com/doi/10.1111/mila.12076). For example, the familiar 'tree' notation is utterly conventional, but then so are the system of indexes associated with nominals according to many theorists. The minimalist assumption is that we can cut back on the merely notational to identify something like a set of invariances, including Merge, from which the heavy descriptive conditions of early incarnations of generative theory might be seen to follow. This is the way Norbert sees things, which strikes me as correct.
Leave a Reply to Nick Cancel reply