Touring the Language Faculty: An Interview with Norbert Hornstein

did october happen? it seemed to careen right passed me into mid-november. despite this unforgivable betrayal by one of my favourite months of the year I did manage to pull off a mighty fun interview with syntactician, philosopher, and fellow-blogger, Norbert Hornstein.

I originally met this chap when I attended his syntax course at the LSA summer school some years back. listening to him speak on the topic, one is apt to get the feeling that generative grammar is building a cool mad max death truck out of scrap metal and wishes (to borrow a phrase from my flatmate). this is something that is often missing from the average lecture on the topic of generative syntax wherein one couldn't be faulted for getting the impression that the field is trying to do philology with both hands tied behind their back (methodologically and theoretically). Norbert is no philologist though, neither on his blog, the Faculty of Language, nor in this interview.

as usual, I'll take a quick dip into something raised in the interview that caught my attention.

during the latter half of the interview (about 48m) Norbert mentions the distinction between linguistics and philology. elsewhere in his writings, the distinction is made by appeal to such notions as explanation and description. I think that Norbert is right to point out that often enough the concerns motivating a programme of research aimed at a faithful description of a language are orthogonal to those motivating a programme of researched aimed at discovering the organizing principles which underlie language tout court. nevertheless, there can also be a palpable tension between the two. consider for instance, the levels of theoretical adequacy demarcated in Radford (1982):

"a grammar of a language is observationally adequate if it correctly predicts which sentences are (and are not) syntactically, semantically and phonologically well-formed in the language.

a grammar of a language is descriptively adequate if it correctly predicts which sentences are (and are not) syntactically, semantically and phonologically well-formed in the language, and also correctly describes the syntactic, semantic and phonological structure of the sentences in the language in such a way as to provide a principled account of the native speaker’s intuitions about this structure.

a grammar attains explanatory adequacy just in case it correctly predicts which sentences are and are not well-formed in the language, correctly described their structure, and also does so in terms of a highly restricted set of optimally simple, universal, maximally general principles of mental computation, and are ‘learnable’ by the child in a limited period of time, and given access to limited data."   

notice that there is a conflict of interest between descriptive adequacy and explanatory adequacy. the former is in a permanently taxonomic mood, and is primarily driven to record, sort, and occasionally predict particular language forms (and meanings); whereas the latter is in a mood to gloss, to provide the rules in virtue of which languages contain the forms and meanings that they do. and the specific pairings between form and meaning that they do.

the conflict arises when we try to map a chaotic, constantly changing world in which accidental and principled variation are observationally indistinguishable to the world of intelligible theory in which consistency and evaluability are supreme values. to my knowledge, the conflict between the two was first noticed by the ancients. for them, a crucial problem was how to relate the sophisticated geometrical and mathematical models of the time to the chaos of worldly phenomenon such as motion. Galileo was really the first (again, to my knowledge) to show the possibility of applying the concepts of geometry to the highly variable phenomenon of motion. relatedly, Bacon was the first to be recognized for proposing a mode of inquiry for, inter alia, discerning accidental variation from principled variation. namely, to carry out experiments which contrive experiences. the virtue of carrying out laboratory experiments is that it is possible to discover crucial discrepancies in theoretical prediction which can be used to hone in on the essential nature of a thing.

(returning to language & Radford's levels of adequacy)
feel free to substitute whatever variable that concerns you besides the syntactic, and whatever metric by which you'd like to evaluate well-formedness. but notice that the problems of marrying your descriptive analyses with your characterization of the abstract grammar doesn't go away (whether it be a grammar of gesture, or social relations, or morals). this is because psychological (to say nothing of theoretical) objects, grammars among them, are necessarily normative while the data is decidedly not. that is, grammars characterize a set of things which a given speaker (or speech community, if you really insist) will find well-formed with respect to form and meaning. so even a sociolect (a dialect in which linguistic varieties are correlated with sociological factors) is a kind of grammar in virtue of which speakers sort sociolinguistic forms and meanings into the well-formed and the ill-formed.

ultimately then I suppose it wouldn't be too off the mark to encapsulate the tension between linguistics and philology as a tension between accounting for forms (and meanings), which are quite varied and diverse, and accounting for the sense of well-formedness, which is largely stable and shared commonly amongst all humans.

caveat: the centrifugal force between philology and linguistics is, as any sensible researcher would acknowledge, quite often counter-balanced by a centripetal force between the two disciplines. specifically, philological projects set a baseline which any linguistic theory must meet if it is to be observationally and descriptively adequate. symmetrically, theoretical work provides the intellectual scaffolding by which philological work can proceed (think metrics of simplicity; criteria of sorting words into classes, languages into families, and the like; the very decisions about what is important to put in your taxonomy and what is not).


The End.


Notes, Admissions, Qualifications, and Apologies:

1. Radford, Andrew. 1982. Transformational Syntax. Cambridge: Cambridge University Press. 
2. Apologies for the odd clicking that starts at about 35 minutes. we are working on making sure that stops happening. if you have any insight as to where this clicking is coming from or how to get rid of it we would be very grateful.