To the editors:

Noam Chomsky’s Aspects of the Theory of Syntax, published in 1965, was highly influential and inspired many scholars to think in new ways about the grammatical structure of languages. But it was hardly a landmark achievement because it contributed no lasting insights and ignored a number of important earlier insights. In her essay, Anna Maria Di Sciullo treats the book almost like a higher revelation, and thus gives the contemporary reader a good sense of the enthusiasm that Chomsky’s ideas created for a decade or two. Until this day, Chomsky has a substantial number of followers who are trying to carry on this enthusiasm—and still see themselves at the forefront of the field. But linguistics has not made progress commensurate with the hopes and aspirations of the 1960s.

While Chomsky emphasized deep questions, the progress that has actually been made concerns breadth of description rather than depth of understanding. Of the 6,000 or so extant languages, about 2,000 now have fairly good grammatical descriptions,¹ and research on small and often disappearing languages has acquired unprecedented prestige. This new attention to the small languages and worldwide diversity is a consequence of the universalist spirit that became widespread in the 1950s, symbolized by the first conference on language universals in 1959.² Chomsky’s interest in universal grammar, as first articulated in 1965, was in harmony with this general movement, though his own focus was always on English syntax.

In this letter, I would like to highlight another aspect in relation to breadth of description that is seldom discussed, namely the completion of syntactic paradigms. The term “paradigm” in grammar is typically associated with inflectional paradigms of Latin, such as mensa (table), mensae (of the table), and mensam (the table [accusative]). In nineteenth-century linguistics, the study of such paradigms and their historical dynamics was at the forefront of research. After 1945, attention shifted to English in the United States, probably because of the decreased importance of Europe and its classical roots. English lacks complex inflectional paradigms, but it has equally complex syntactic paradigms. Analogous to mensa, mensae, and mensam, English has paradigms of constructions such as Mary pleased John, John was pleased by Mary, and John is easy to please.

Paradigms such as this one can be applied to a wide range of verbs, but where exactly is the limit of each construction? In noun inflection, linguists were used to setting up classes of first conjugation, second conjugation, and so on, but for syntax, it had not been done systematically before. For example, how does the verb “give” behave in these three constructions? We can say John was given money by Mary, but also The money was given to John by Mary, with an additional complication: the preposition “to” is now required. In *The money was given John by Mary the asterisk indicates unacceptable sentences, and its use was one of the most important practical innovations of generative grammar. In addition to analyzing occurring sentences, grammarians increasingly asked how to account for gaps in the space of logically possible sentences.

Just as earlier linguists who focused on the inflectional possibilities of words wanted to have full paradigms, syntacticians started to investigate the complete range of syntactic possibilities, and they found more and more surprising results. An example comes from coordinated nominal phrases in which the noun is shared. In many languages, the second noun can be omitted, but in English, an anaphoric element, one, is needed:

*the three big horses and the three small _

the three big horses and the three small ones.

But when it is the first noun that is omitted, the anaphoric one is not needed:

the three big _ and the three small horses.³

Is the empty position in this sentence simply equivalent to the anaphoric one? Consider further possibilities, such as nominal phrases with relative clauses:

the three big horses that I own and the three small ones that you own.

When an anaphoric one is used, it is still possible to have a relative clause at the end. But when the empty position is used, the result is unacceptable:

*the three big _ that I own and the three small horses that you own.

The syntactic rules for English must be formulated in such a way that this outcome is predicted. Complete syntactic description has turned out to be an extremely difficult task because the interactions are complex and we keep running into new factors that play a role. In inflectional paradigms, the variation is relatively easy to control, because there are rarely more than a dozen dimensions: number and case for nouns, plus perhaps possessive marking; tense, aspect, and person for verbs, plus perhaps mood and negation marking; and a few others.

In syntactic paradigms, the variation is enormous, and we keep making new discoveries. German has a three-way gender distinction in articles and personal pronouns—er (he), sie (she), es (it)—that applies to animate and inanimate nouns alike. We refer to der Kaffee (the coffee) as er, to die Milch (the milk) as sie, and to das Buch (the book) as es:

But when the two pronouns are coordinated, they can no longer refer to inanimates. The following sentence can only refer to people, like the English personal pronouns he and she:⁴

Er und sie gefallen mir.
I like him and her. (Not: I like that and that.)

Personal pronouns are among the most basic elements of languages, so one might think that their properties in German, a well-studied language for centuries, would be well-known. But this special animacy behavior of German pronouns in coordination constructions had not been observed before. It was only through generative grammar’s emphasis on completing the syntactic paradigms that linguists learned about this fact. Anna Cardinaletti and Michal Starke wanted a full account of the behavior of personal pronouns and systematically asked how they behaved in a wide range of contexts.

Paradigm completion is a familiar research task in experimental fields: scientists want to know how a diverse range of elements behave under a diverse range of conditions. Linguists did not think of their field as experimental until the 1960s, because they worked with texts, or with speakers of the language, and did not feel the need for controlled laboratory conditions. They often focused on learning to speak, or at least understand, the languages themselves, not on experimental methods. Chomsky and others did not emphasize the experimental nature of their paradigm completion work either. But systematic testing or diagnosing does have some key properties of the experimental approach in science.⁵

The increasing interest in completing the syntactic paradigms has led to an enormous expansion of linguistic knowledge, but how is this knowledge related to the kind of explanatory depth that Chomsky was aiming for in 1965? And was the new experimental approach rooted in his mentalist approach? As for depth of understanding, it must be said that Chomsky’s hopes have not been fulfilled. Linguists do not know much more about universal grammar than we knew in 1965, and our understanding of mental grammars is still quite limited. We can write grammatical descriptions and have an awareness of many new phenomena, often with a rich cross-linguistic perspective, but there is no simple overall picture of the innate faculties that make knowledge of languages possible. There are many features that are broadly shared among the languages of the world, but these are often best explained by communicative efficiency, not by the kinds of abstract innate structures that Chomsky had in mind.⁶ His interesting questions about the architecture of the language faculty—such as transformations vs. phrase structure rules, syntax vs. lexicon, Merge vs. phonetic form—have not led to any convergence among later researchers.

Given Chomsky’s general disdain for methodological questions, it is interesting that the paradigm-completing experimental method should be the most lasting contribution of the revolution that Aspects was felt to be at the time. And indeed, the adjective “generative” indicates this novel aspect of Chomsky’s approach: while earlier linguists had developed formal methods to describe all the attested complex words and sentences, Chomsky was more ambitious. He wanted a formal system that was able to generate all and only the acceptable sentences of a language. This was apparently what led him and his associates to approach syntactic questions in new ways. But was it related to his radical mentalism? This is not clear to me, because the new technological possibilities would have raised the issue of machine translation and generation anyway. If we want to build a machine that can produce language, we need complete paradigms and formal methods. Much of the perceived novelty of Aspects may have been part of the general spirit of the 1960s, though there is no doubt that Chomsky’s book, through its content as well as through Chomsky’s personal charisma and his later work, has played an important role in the history of grammatical investigations.

Martin Haspelmath

Anna Maria Di Sciullo replies:

In his letter, Martin Haspelmath offers a different perspective on Aspects. According to Haspelmath, Aspects was hardly a landmark achievement because “[i]t contributed no lasting insights and ignored a number of important earlier insights.” He adds that “linguistics has not made progress commensurate with the hopes and aspirations of the 1960s.” This short reply provides some reasons to think otherwise.

First, Chomsky did not ignore earlier important insights on language. He pointed to the limits of notional and taxonomic grammars, and identified shortcomings of the behaviorist approach to language acquisition.⁷ He also built on earlier insights about language, going back to Plato, the grammarians of Port Royal, Wilhelm von Humboldt, Isaac Newton, Alan Turing, and Emil Post.⁸

Second, Aspects, stemming from his previous works, provided a definitive form to the rules of the syntactic component of grammar.⁹ It also contributed long-lasting insights on language. One of them is syntactic categories defined in terms of a small set of formal features,¹⁰ which are active in agreement, displacement, and variation.¹¹

Third, linguistics has made commensurable progress since Aspects, exceeding the hopes and aspirations of the 1960s. Aspects left open several questions for further inquiry. Of these questions, perhaps the most interesting are those concerning the properties of syntactic rules, and why certain principles would qualify as first principles. Consider the discovery that syntactic rules apply across categories. In “Remarks on Normalization,” the rewriting rules of the base component postulated in Aspects were eliminated in favor of a general rule schemata.¹² Likewise, the transformational rules were reduced to two general operations in government and binding theory: move NP, displacing nominal constituents; and move wh-, displacing operators such as who, what, where, and when in open question formation. In the minimalist program, the syntactic operations are reduced to their simplest form: a dyadic and recursive set formation operation, whose computational properties are under investigation.¹³

Aspects is undoubtedly a landmark achievement. Building on Chomsky’s previous work, it provided a definitive exposition of the Standard Theory. In this framework, the object of inquiry motivating linguistics investigation is the language internal to the mind. Linguistic theory is defined with the tools of mathematical logic and the theory of computation. Linguistic phenomena follow from principles of Universal Grammar and provide an explanation for language acquisition. Moreover, the Standard Theory led to the creation of an interdisciplinary field bridging linguistics and biology to further our understanding of the human ability for language.