The following “conversation” consists of passages from Roberta D’Alessandro’s 2021 commentary (on Lingbuzz) on my 2021 paper on general linguistics (to appear in Theoretical Linguistics), plus some reactions from me. I thought that such an immediate reply to some of the points might be useful, as D’Alessandro’s commentary has proved to be very popular (hundreds of downloads within a few days).
Martin: Many thanks, Roberta, for commenting on my forthcoming paper “General linguistics must be based on universals (or nonconventional aspects of language)”. I will react to some of your points directly below (even though a more formal reply is planned for the journal Theoretical Linguistics). So let us consider some of your criticisms.
Roberta: (¶4) “Some of the things [you] write about the methodology [of generative linguistics] are not quite accurate; this makes me wonder: Are [you] simplifying the line of argumentation for the sake of rhetoric…, or are we not making ourselves understood?”
Martin: I have long been struggling to understand what everyone else means, because I’m not happy to simply be “in the other camp”. We should have a universalist linguistics also in that we at least understand each other’s most basic points, even if we don’t adopt the research strategies of others. So I fear that it’s the latter – generativists are often unclear about what they mean by “theory”, so I am insisting that we need to distinguish between p-theories and g-theories (see further my comment on your ¶9 below).
Roberta: (¶6) “What we are asking is not “what happens”, which is … rather trivial, but “what cannot happen. This is one of the core points of the generative enterprise and of contemporary linguistic research…”
Martin: Yes, I agree completely – and this is why I am interested in generative linguistics, at least of the type of research that people like Ian Roberts, Jonathan Bobaljik and Mark Baker engage in. We want to know what the limits of grammatical diversity are, and we want general theories about this. (In generative grammar, the general hope is also that a theory of acquisition will simultaneously serve as a theory of the limits of variation; I’m not sure about this, and since you don’t mention it, I will not talk about this further.)
Roberta: (¶9) “Abstractions and descriptions in structural terms are not a formal theory yet. A formal theory is built through the convergence of a number of proven hypotheses… We reduce [several] phenomena to a more general phenomenon…, until we hit the general “law”… The general law could be due to cognitive requirements, it could have functional explanations. Only if we do not find any such functional explanation or general cognitive explanation can we attribute this “law” to [UG].”
Martin: Yes, I agree on the latter point, and in fact full-heartedly. There are many facts of grammars that seem eminently functional (e.g. that accusative markers are often limited to referentially prominent patient arguments), so we need not attribute them to a special mechanism. UG explanations of the type that have the greatest prestige in generative grammar should be a last resort. Since there may not be a good functional explanation for the absence of “agreement in color features”, this may indeed be a candidate for an aspect of innate knowledge.
However, there are different kinds of theories: General theories (about Human Language in general, g-theories) and particular theories (about particular languages, p-theories). You may not want to call the latter type “theories”, but as I noted in my paper, this has been the general practice in generative linguistics, from Chomsky (1957) through Adger (2003). It does not really matter, though, what we call such complexes of ideas – what matters is whether we get closer to our goal of arriving at general theories by using a highly technical metalanguage for descriptions of particular languages (e.g. Arregi & Nevins 2012, a book that only discuss a single language, Basque). I have strong doubts about this, whereas broad comparative work like Bobaljik (2012) is quite likely to lead to general-theoretical insights.
Roberta: (¶6) “It totally escapes me why the general take on linguistics nowadays… is that you need many tokens of something for it to be worthy of consideration… Imagine that a phenomenon is produced by [only] one speaker of a language.. Of course, an explanation is required as to why the speaker is the only person producing that phenomenon, but this does not mean that this phenomenon should be disregarded altogether.”
Martin: Again, I agree fully: Everything that we observe should be taken into account, and to the extent that it’s systematic, it needs to be described. Linguistic systems of speech communities, but also systems of individuals, are worthy of theoretical description, but such descriptions/analyses are p-theories, with no immediate relation to g-theories.
Well, one might say that if a phenomenon is found in at least one speaker, this at least demonstrates that it is possible in Human Language (HL), and that it is relevant to general linguistics in this sense. This is true, as you say: “What we observe in languages is a possible instantiation of “Human Language” (¶6). But we want more than just unstructured heaps of phenomena that are possible in Human Language – we want structured theories, and for these, we need systematic cross-linguistic comparison.
Roberta: (¶8) “If we limited ourselves to observing subject clitics as a superficial category, and if we did not look into their structure and the features they encode, and we did not make any generalization about what they represent, and we did not proceed by reductionism, we would not be able to make predictions about what can and cannot be found across languages.”
Martin: Now here, it seems that I no longer agree with all you say, because I think that we can make cross-linguistic predictions about subject clitics without “looking into their structure”, at least if by “their structure”, you mean (i) their language-particular structure, or (ii) the UG features (the innate building blocks) they are made up of. Each language has its own categories (for instance, some languages have polite pronouns that are extremely culture-specific), so we need comparative concepts for systematic comparison (this is a logical necessity). As an alternative, you might say that you compare subject clitics with respect to their UG features, and while this is logically coherent, it is not a practical possibility at present, because we don’t know what the UG features are.
But I must say that I’m not quite sure what you mean by “superficial category” and by “encoding of features”. This is a typical case of the genuine problems that I often encounter with understanding generative grammarians. (I never use the term “surface”, and I never talk about “encoding of features”, because I don’t understand these concepts; what I understand is the idea of encoding a meaning.)
Roberta: (¶12) “What I claimed is that wh-movement … exists in languages, and therefore it is a possible grammatical phenomenon in language. Therefore, since it is possible, it can be found in languages other than English because we are all human, and we all make use of the same restricted set of grammatical tools.”
Martin: Yes, I think I understood this – but whether we all make use of the same restricted set of tools is an empirical question, and the jury is still out (as you note yourself: (¶11) “Have we succeeded in finding the general theory of HL? Not yet”). This is why I keep emphasizing the need for standard definitions of comparative concepts, so that we can learn from each other. One of my criticisms of generative grammar is that too many things are presupposed. Even if you replace the metaphorical term “wh-movement” by “copying” and “double interpretation”, it will not be clear what exactly falls under it. Do you identify a question construction as “wh-movement” if the question pronoun has positional flexibility but statistically prefers a fronted position, in a language which generally has very flexible word order? This does seem similar to the English Wh-Movement construction, but how would we interpret this similarity? I think that we need clearly defined comparative concepts to make progress here.
Roberta: (¶13) “Imagine you have to describe an undocumented language… What do you do? Do you start looking for pronouns or not? Do you only consider absolute universals? …Should one start from scratch for every new language description[?] This [would] mean that all our knowledge is going to be completely useless when describing a new language, because we are not allowed to postulate the existence of a category just because it exists in another language.”
Martin: You are right that we want to make use of our knowledge of other languages when beginning the task of describing a new language. But as I said earlier, we cannot presuppose that cross-linguistic categories exist (because we don’t know what the innate building blocks are, if there are any). So we need to be careful and use our knowledge for inspiration, not for specific hypotheses (I discussed this point more extensively in my 2020 paper on the value of comparison for description – which consists in (i) inspiration and (ii) terminological transparency).
For example, when studying a new Romance variety, we may be interested in whether it has subject clitics or not (this is a question close to your own research interests, e.g. D’Alessandro 2017). But how do we distinguish between subject clitics and full subject pronouns? If we use different criteria in different languages, we will quite likely end up comparing apples and pears (though we could be lucky and hit upon the right UG categories). So I think that we need a clear comparative definition of what a “clitic” is (I wrote about the problems with the “clitic” concept in my 2015 paper). But while this will answer our comparative question (which varieties have subject clitics?), it is not really relevant to language description, because the “clitic” concept may not be needed for the description. (For example, grammarians have often successfully described German varieties with object clitics without using the term “clitic”, focusing on other properties such as the special position, the phonological properties, etc.).
Roberta: (¶13) “Isn’t it easier to start from postulating that Italian has no number and no wh-movement based on Chinese, and then look for evidence to the contrary, … to conclude that there is wh-movement in Italian? This way the description is as thorough, but we don’t need to go back to Aristotle … to start describing a new language.”
Martin: As I said, we should surely be inspired by what we know about other languages, and of course this goes in both ways: Describers of Asian languages should be inspired by what we know about European languages, and vice versa. But if we start with the hypothesis that language A has the same category C that we identified language B, on what basis can we falsify this hypothesis? We need clear criteria, but since the days of structuralism, we have known that different languages have different structures, which means that we cannot apply the same criteria in all languages. A big problem in all fields is cherry-picking of data, and the corresponding problem in linguistics is what I now call “criteria selection bias” (in 2018, I called it “diagnostic fishing”). There is an open-ended range of phenomena that one might take into account in formulating a generalization, and we are often biased toward particular conclusions (for example, in my first p-syntax paper, I was biased toward finding an ergative construction in Lezgian, and to be sure, I found evidence for it). So there’s really no alternative to “going back to Aristotle” (who described Greek without being biased by earlier descriptions of other languages, because he was unaware of the findings of Babylonian linguists).
Roberta: (¶13) “[If one doesn’t need to go back to Aristotle]… one might even spare some time to concentrate on understanding what Human Language is about rather than wasting time on tiny details that sometimes take the attention away from the big picture.”
Martin: Yes, we want to understand the big picture, in addition to paying attention to tiny details about Italo-Romance varieties or Lezgian ergativity. But is it a question of time? I would rather think that there’s always some kind of “big picture” in the background that influences how we think about the “tiny details”. For example, I find your papers about Abruzzese hard to read because I don’t have a good understanding of the big picture that you are presupposing (e.g. why are you using the notions “spine”, “v”, and “T”?). So it seems to me that “understanding Human Language” is not a matter of too little time, and that we will not arrive at better understanding if we ignore the details of language-particular differences. There is a long history of paying too little attention to differences between different cultures, and at this point, we are all aware of the dangers of not respecting such differences sufficiently. Cultural differences tell us about what it means to be Human, and linguistic differences tell us about what it means to be a Human Language.
Adger, David. 2003. Core syntax: A minimalist approach. Oxford: Oxford University Press.
Arregi, Karlos & Nevins, Andrew. 2012. Morphotactics: Basque auxiliaries and the structure of spellout. Dordrecht: Springer.
Bobaljik, Jonathan David. 2012. Universals in comparative morphology: Suppletion, superlatives, and the structure of words. Cambridge, MA: MIT Press.
Chomsky, Noam A. 1957. Syntactic structures. ’s-Gravenhage: Mouton.
D’Alessandro, Roberta. 2017. When you have too many features: Auxiliaries, agreement and clitics in Italian varieties. Glossa: A journal of general linguistics 2(1). 1–36. (doi:10.5334/gjgl.102)
D’Alessandro, Roberta. 2021. Not everything is a theory: A reply to Martin Haspelmath. Theoretical Linguistics (to appear). (https://ling.auf.net/lingbuzz/005659)
Haspelmath, Martin. 1991. On the question of deep ergativity: The evidence from Lezgian. 44/45(1–2). 5–27. (doi:10.5281/zenodo.225289)
Haspelmath, Martin. 2015. Defining vs. diagnosing linguistic categories: A case study of clitic phenomena. In Błaszczak, Joanna & Klimek-Jankowska, Dorota & Migdalski, Krzysztof (eds.), How categorical are categories? New approaches to the old questions of noun, verb, and adjective, 273–304. Berlin: De Gruyter Mouton.
Haspelmath, Martin. 2018. How comparative concepts and descriptive linguistic categories are different. In Van Olmen, Daniël & Mortelmans, Tanja & Brisard, Frank (eds.), Aspects of linguistic variation: Studies in honor of Johan van der Auwera, 83–113. Berlin: De Gruyter Mouton. (https://zenodo.org/record/3519206)
Haspelmath, Martin. 2020. The structural uniqueness of languages and the value of comparison for description. Asian Languages and Linguistics 1(2). 346–366.
Haspelmath, Martin. 2021. General linguistics must be based on universals (or nonconventional aspects of language). Theoretical Linguistics (to appear). (https://ling.auf.net/lingbuzz/005158)