Last week I was at one of the most unusual and stimulating events I’ve attended in a long time – a workshop on “Variation and universals” organized by Roberta D’Alessandro and Marc van Oostendorp, bringing together syntacticians and phonologists, macrotypologists and microvariantionists, and generativists and linguists who were unsure how to describe themselves. The goal was to think at a higher level than usually about the role of typological data and universal claims in understanding language(s).
(The position statements are available from the programme page, and much of the workshop was streamed on Facebook, on Roberta’s and Marc’s pages, where it can still be viewed.)
At the end of the workshop, a colleague said after one of my statements: “Maybe Martin is the most minimalist of us all” – and I was very happy because it was said in a friendly way; after all, it is only by friendly and constructive interaction that we can understand how we differ (or do not differ) in our assumptions, goals and practices.
Of course, I’m not “Minimalist” in the sense that I use the technology of Chomskyan syntacticians, but I certainly agree that we should attribute as little as possible to the biological prerequisites for language – it seems that this is what Chomsky has been saying for quite some time now. (“Maybe only recursion is domain-specific.”)
One of my wordings raised immediate objections among the generativists present: When I said that linguistic categories are cultural categories and should be investigated as such (without assuming that different languages share the same categories), the question was: But isn’t language acquired completely effortlessly and instinctively? Isn’t this quite different from other kinds of cultural learning? Sure, I said, the instinct to acquire a language is quite likely a specific biological trait of humans – but this does not mean that the categories of the various languages must be biological traits as well. Humans also have an instinct to form lasting mating relationships (it seems), but we still find striking differences in the cultural forms of marriage.
I certainly agree that we need to attribute all kinds of sophisticated abilities to speakers: There must be some kind of “generative procedure” (as urged by David Adger), and the ability to set up abstract categories, features, and dependencies between them must be present in advance. So clearly, there is an innate faculty of language – in that regard we’re all Chomskyan.
But what does this have to do with universals, or with the study of specific syntactic and phonological patterns? Here is where different linguists differ quite strikingly: Many use the same categories and mechanisms to analyze different languages, as if the individual building blocks of grammar were part of the biological prerequisites. I call this the natural-kind view of linguistic categories (Haspelmath 2017): Just as our ability to learn a language in childhood is innate, so categories like [NP] or [+speaker] or [+sonorant] are assumed to be biologically given, in advance of any acquisition. If this were the case, then it would make perfect sense to look for a category or mechanism that is found in English also in other languages, e.g. adjectives in Chamorro, adjectives in Quechua, or syllables in all languages. Some of the generative linguists at the workshop agreed: Yes, categories are thought to be natural kinds, innately given (even though this is not really compatible with Minimalism, where only recursion belongs to the faculty of language in the narrow sense).
Now again, I would not exclude the possibility that categories are given in advance, but how can we find out whether this is the case? Support for this view would come from finding the same kinds of categories again and again when no other (weaker) explanation seems conceivable. But for the great majority of categories with identical or very similar properties across languages (e.g. person, number, case, tense, topic, reflexive), it seems that a functional explanation is readily available – and many other categories are not all that similar across languages (so we need language-particular categories anyway).
In general, stronger explanatory factors should be invoked only if weaker explanatory factors are insufficient – and functional-adaptive explanations very often seem sufficient to explain the limits on variation. So I’d say that innate categories should be posited only as a last resort – and in this sense I’m a categorial minimalist. For some reason, the label “Minimalist” tends to be associated with people who adopt a particular technology in syntax , and who take the view that categories are natural kinds. This does not seem “minimalist” to me, but it of course reflects a tradition of generative linguistics going back to the 1960s. It would be nice to see someone defend this tradition explicitly against the more minimalist conception that I have been advocating, but since the natural-kinds view is the majority approach, I can understand that not many people see the need to defend it.
Of course, many other things were discussed in Abruzzo, e.g. the nature of variable grammars and dialect syntax, as well as the nature of phonological universals (for melody properties, Tobias Scheer agreed with me that there are no universal categories, and that whatever generalizations there are derive from factors outside UG), but this blog post is already long enough.
Haspelmath, Martin. 2017. How comparative concepts and descriptive linguistic categories are different. Draft, https://zenodo.org/record/570000.
Dear Hypotheses blogger,
We found your article particularly interesting. To increase its visibility so the community can more easily appreciate it, we made it a headline article on the hypotheses.org slider.
The Hypotheses.org team