the 1980s was to formulate an account that allowed for the observed phenomena without reintroducing the excessive
promiscuity of the earlier hypothesis. More generally, the frequent readjustments in Chomsky's theories over the years
have been an attempt to propose as lean a version of Universal Grammar as possible; the Minimalist Program
(Chomsky 1995) is by far the leanest.
Some of this constant re-evaluation has been driven by explorations into what forms of grammar are mathematically
learnable. In particular, many of the early constraints on transformational rules were in part inspired by learnability
considerations (Gold 1967; Wexler and Culicover 1980), which applied specifically to theAspectsmodel. And later
researchhas oftenavertedtolearnabilityconsiderationsas well(Baker and McCarthy1981; Pinker 1989; Gleitman and
Landau 1994; Van Valin 1994; and many others).
What I hope the reader can appreciate fro mthis brief discussion is that hypotheses about the content of Universal
Grammar constantly raise complex empirical issues about how a multitude of linguistic phenomena are to be
described. Continued examination of more and more linguistic phenomena, with attention not only to what happens
but also to what doesnot happen, has led to the many reformulations of linguistic theory over the years, with
concomitant rearticulation of the content of Universal Grammar.
4.4 Substantive universals, repertoire of rule types, and architectural universals
Another aspect of the proble mof universals is addressed in a section ofAspectscalled“Formal and Substantive
Universals”(Chomsky 1965: 27–30), which distinguishes two different facets of Universal Grammar. By“substantive
universals,”Chomsky means the basic building blocks of linguistic structure: phonological distinctive features and the
notion of syllable in phonology, and parts of speech and the notion of syntactic tree in syntax. These parts are used
differently in different languages, but one cannot construct a human language without them. Chomsky therefore
wishes to attribute the mto the brain's prespecification.
By“formal universals,”Chomsky means the overall organization of the grammar. These might be divided into two
subcategories. First, the child has to have a repertoire of rule types—what kinds of rules a language might have for
combining the basic units into complex structures. Chapter 3 distinguished phrasal formation rules, derivational rules,
severalvarietiesofconstraints, lexicalformationrules, lexicalredundancyrules, and inheritancehierarchies. Allofthese
rule types have to be in the child's repertoire, ready to befilled with content so they can develop into the rules of the
child's very own grammar.