Principles And Parameters syntax posited that along with some principles, there were parametric settings for certain properties, which are either "on" or "off" in a language. Examples are the "head-directionality parameter", the pro-drop parameter, the polysynthesis parameter, and the ergativity parameter, which are assumed to be provided by UG. Was it (ever/generally) claimed that UG supplies a default setting for one or more parameters? My goal is to see how the low frequency of less-common systems (ergative or polysynthetic languages, for instance) was accounted for in that framework.

1 Answer 1


Some technicalities have been cruelly sacrificed for the sake of a partly-okay answer.

From Chomsky & Lasnik (1993):

Principles and Parameters theory is not a precisely articulated theoretical system, but rather a particular approach to classical problems of the stdy of language, guided by certain leading ideas that had been taking shape since the origins of modern generative grammar some 40 years ago.

So, there's room within P&P for the concept of default settings, but because P&P is not a single theory, different theorists can choose to implement 'default' settings or not.

P&P in particular, and generative syntax generally, does not do very well with the problem of typological distribution. Why so few verb-first languages? Under an unfairly simplified P&P approach, the Parameter in question has to do with the movement of the verb: verb-first langues like Arabic or Irish have a Parameter which states that V(erb) always moves up to head-adjoin with C(omplementizer). The difference between Arabic and English is that English only does this in ex. questions, Arabic does this all the time. So the Parameter in English is 'do this with questions', in Arabic it's 'do this all the time', and in Mandarin it's 'don't do this'. But most theories of generative syntax have nothing much to say, at least not convincingly, about why there just aren't so many languages that work the way Arabic syntax does, or even the way English does.

Greenberg solved the problem partway with his concept of 'feature hierarchies'. For example, in relative clauses, English is unusual because we can relativize possessors: This is the dog whose owner never picks up after them. Most languages can't do that. Greenberg has a hierarchy of relativizable positions: Subject > Direct Object > Indirect Object > Adpositional Object > Possessor. This means that a language which can relativize an indirect object can always relativize a subject and a direct object, but a language which can relativize a direct object can't necessarily relativize an indirect object. And so on. A very few languages can relativize nothing but a subject, a very few languages can relativize everything including possessors, but most will naturally fall somewhere in between. If we assume that languages are more or less randomly distributed along the hierarchy, we'll naturally wind up with the statistical rarity of languages like English which can relativize possessors.

But even with Greenbergian hierarchies, many typological imbalances still struggle to be explained in generative syntax. They don't help us much in our Arabic vs. English example. Both English and Arabic can perform to V-to-C adjunction, the difference is when they do it and the grammatical function of this adjunction. A hierarchy like 'never > in questions > always' is incomprehensible: "a language which always performs V-to-C adjunction can never perform V-to-C adjunction"? And it's not entirely clear that the hierarchies are actually part of UG; is Greenberg describing innate Parameters of UG, or is he describing an emergent phenomenon, an interesting quirk of syntactic systems interaction? You'll find both points of view.

The development of the PF and LF distinction in the "Y-model" doesn't help much, either; it only 'passes the buck', so to speak. Under the Y-model, the difference between English, Mandarin, and Arabic is that English and Mandarin actually do always perform V-to-C movement, but this movement is only relevant for LF, whereas for Arabic it's relevant at PF. This replaces the need for a Parameter, but it still doesn't give us a good reason for why the Arabic structure should be less common than Mandarin one. Maybe we can answer the question with 'economy': Arabic syntax is less common because it involves more overt movement to be handled by PF, and syntax generally likes to 'procrastinate'. But if this is the answer, it still doesn't tell us why a language should ever choose the 'uneconomical' path in the first place, or why an 'uneconomical' language type wouldn't naturally stabilize over time to an economical one. Proto-Semitic was a VSO language, and possibly Proto-Afro-Asiatic as well; in all those millennia and all those generations, if the reason for the rarity of VSO was 'economy', you'd think Arabic would've sorted it out by now. And so we're back to where we're started.

When 'Parameters' are boiled down to questions of overt (PF) vs. covert (LF) movement, some theorists might (and do) say that the 'default Parameter' is the one which involves the least overt movement. Just like Optimality Theory presents us with 'Faithfulness Constraints' which dislike us mucking about too much with the phonological input, so too do some versions of generative syntax have 'Faithfulness Constraints' which dislike the PF moving too many parts away from the base-generation. But it still seems like all languages do so anyways, sometimes quite drastically, like in Arabic. You can come up with a Constraint based approach, like OT, which says that for some languages (Arabic), a constraint like, "V must license/be licensed by some feature of C" is more important than the Faithfulness constraint, but in other languages (Mandarin), the Faithfulness constraint is more important. But this still passes the buck...why aren't there more languages which rank "V-to-C" higher than "Faithfulness"?

And you can around and around like this in generative theories of syntax, with each successive theory coming up with a different explanation of the original Parameter, but each explanation not really able to answer the question of typology. The constraint-based one comes closest, in my opinion--the default setting is the one which involves the least overt movement at PF--but it leaves unanswered the followup question, which is, "So why would a language ever choose to be uneconomical? Why wouldn't an uneconomical language type be naturally unstable?"

That's why I tend to lean away from generative grammar when trying to understand typology. My personal favorite 'pet theory' is John Hawkins's Performance-Grammar Correspondence Hypothesis, which endeavors to answer just the very question of why some language types are more common than others, without resorting to UG or Parameters, and instead relying on principles of information processing derived from findings in psycholinguistics and cognitive science, which can be broadly applied across multiple versions of syntactic theory, generative or otherwise, UG or no. Some of his findings even suggest that aspects of structure which generative theory ascribes to UG or Parameters might actually not be innate at all, but rather result secondarily from the ways our brains are naturally biased towards receiving and processing information, and the structures our brains find easier or more difficult to parse.

  • You describe the Keenan & Comrie Accessibility Hierarchy, but attribute it to Greenberg: do you really think it was first proposed by Greenberg? I was hoping for more engagement of the actual P&P literature – I know many ways to derive the tendencies, I'm just looking for theory-internal discussion, if it exists.
    – user6726
    Jun 25, 2021 at 14:59
  • Ah, you are correct, my mistake! You’re right that the specific example I used wasn’t from Greenberg himself—I was trying to tie him to the general notion of feature hierarchies and implicational universals, and I overlooked the source of that particular example. Apologies for the inadequacy of my response!
    – Khove
    Jun 25, 2021 at 16:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.