A 700-year-old argument for a syntactic transformation

Michael A. Covington


Program in Linguistics, The University of Georgia

Noam Chomsky pointed out in Cartesian Linguistics (1966) that philosophical grammarians in past centuries anticipated some of the insights of generative grammar. In what follows I shall examine some arguments offered by the modistic grammarian Radulphus Brito (c. 1300) for what amounts to a transformational analysis of the plural feature on conjoined singular NPs such as (in Latin) Socrates et Plato.

Modistic grammar

The Modistae (Pinborg 1967, Rosier 1983, Covington 1984) were a group of grammarians that flourished around Paris from about 1260 to 1310; they are so named because they wrote treatises de modis significandi "on modes of signifying." Having inherited the traditional grammatical description of Latin from the ancients, the Modistae set out to make grammar a science in the Aristotelian sense, i.e., to explain language, not just describe it. They called their endeavor "theoretical grammar" (grammatica speculativa, where speculativa translates Greek theo:re:tike:). Important modistic grammarians include Martin of Dacia, Boethius of Dacia, Siger de Courtrai, Thomas of Erfurt, and Radulphus Brito. The last of these left the most elaborate surviving treatise, a book-length set of quaestiones disputatae.

Whereas the initial goal of Chomsky's TG was merely to specify the set of grammatical sentences, the Modistae always took it for granted that syntax and semantics were closely linked. They called grammatical features modi significandi because they saw each feature as an aspect of the way meaning is encoded. They did, of course, acknowledge that some features, such as grammatical gender, are "accidental" in that they belong to the arbitrary structure of Latin rather than the essence of human language.

Like much of traditional grammar, modistic grammar is implicitly generative; that is, it describes whole sentences by describing how the parts are put together. However, modistic syntax is not transformational; each sentence has only one structure.

Instead of phrase-structure rules, modistic grammar uses a dependency framework; each word-to-word link is called a constructio and there are two criteria of headship. The syntactic head, called primum or prius, is the governor or the modified element, the one presupposed by the other. The semantic head, or terminans, supplies or points the way to the referent. Syntactic headship was introduced into modern transformational grammar by Chomsky (1970) and Jackendoff (1977), and, subject to certain conditions, dependency trees are equivalent to X-bar trees (Covington 1992). Semantic headship is, as far as I know, not used in modern work, although the theory of discourse referents bears some relation to it.

Conjoined structures

Dependency grammarians and X-bar theorists of all epochs have noticed a problem with conjoined phrases such as Socrates and Plato or runs and laughs. Specifically, none of the three words appears to be the head. The two conjoined elements are of equal status; neither has a clear claim to headship over the other. If anything, the conjunction appears to be the head word, but it does not determine the syntactic properties of the overall phrase; when you conjoin two NPs you get an NP, not a ConjP.

Even worse, when you conjoin two singular NPs you get a plural NP even though none of the words within it is plural. The fundamental assumption of dependency grammar and X-bar theory, that the features of every phrase are those of its head word, breaks down.

Radulphus' analysis

Radulphus Brito analyzes the sentence Socrates et Plato currunt ("Socrates and Plato run") in two of his Quaestiones.

In Quaestio I.24 the question is whether all constructions are either transitive or intransitive, i.e., whether in every construction the syntactic and semantic heads are either the same or different. The alternative would be to have constructions in which one type of head is missing or undefined. The issue is therefore whether all constructions are endocentric.

Radulphus argues that the grammatical relations in Socrates et Plato currunt are doubled up. Both Socrates and Plato are subjects of the verb, which assigns nominative case to both of them. (Thus also Tesnière 1959, who calls this phenomenon dédoublement.) In modistic grammar, the sentence is the maximal projection of N, not V, so the result is a two-headed structure with one verb depending on both nouns.

The relation between et and the conjuncts is less obvious, but again Radulphus resorts to doubling up. He argues that et is a modifier of both of the conjuncts and therefore depends on both of them. It follows that (in modern terms) Socrates et Plato is not a constituent; instead, Socrates et and et Plato are two noun phrases that share the word et (Fig. 1). This gives an ill-formed dependency tree, but no more ill-formed than it was already; once you've admitted dédoublement, you should use it everywhere it's applicable. Radulphus' sense of theoretical economy is reminiscent of modern generative grammar.


Fig. 1. Dependency structure and equivalent X-bar tree for Radulphus' analysis.

In Quaestio I.69, Radulphus tackles the plural verb agreeing with two conjoined singulars. He notes that two singulars do not always make a plural: "Socrates or Plato" takes a singular verb, and "runs and disputes" takes a singular subject. Clearly, we're looking at a rule that applies only to NPs joined by "and." This alone is enough to put it outside the regular system for building syntactic structure.

Next Radulphus argues that there is no real semantic plurality in Socrates et Plato. Plurals, in his system of semantics, signify per modum unius plurificati ("in terms of one thing multiplied") but Socrates et Plato signifies two things as singulars. On this semantic ground, Radulphus rejects as misguided any attempt to treat Socrates et Plato as a plural phrase. He already had no place to put the plural feature, since et is not the head word, Socrates and Plato are not plural, and in fact Socrates et Plato, on his analysis, is not a constituent (has no single head). Nor can the verb respond directly to the presence of "and" since it has no syntactic link to it.

What's left? Only the radical alternative that, strictly speaking, the verb should be singular, and something outside the basic grammar pluralizes it. This is what Radulphus concludes: "Socrates and Plato run," with a plural verb, is a figure of speech, but the figure is obligatory. That is, the grammar specifies a singular verb, but, obligatorily, speakers of Latin must change the verb to plural in this situation. He cites Priscian (c. 500) and Petrus Helias (c. 1140) in support of this analysis, comparing it to conceptio personarum ("combining of persons," as in "You (2nd pers.) and I (1st pers.) run (1st pers.)," which he also considers to be a figure of speech.

What is crucial here is the idea that figurative construction, i.e., deviation from the basic grammar, is obligatory. The grammar generates the sentence in one form and then its form gets changed by the obligatory figure. This is the only transformational syntactic analysis I have encountered in modistic grammar, although the seeds of the idea were present in the ancient concept of figures of speech, and may have germinated at more than one time and place. Centuries later, a different classical figure, ellipsis, formed the basis for the account of conjunction in Chomsky's Syntactic Structures.


Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton.

Chomsky, Noam. 1966. Cartesian linguistics. New York: Harper and Row.

Chomsky, Noam. 1970. Remarks on nominalization. Readings in English transformational grammar, ed. R. A. Jacobs and P. S. Rosenbaum, pp. 184-221. Waltham, Mass.: Ginn.

Covington, Michael A. 1984. Syntactic theory in the High Middle Ages. Cambridge: Cambridge University Press.

Covington, Michael A. 1992. GB theory as dependency grammar. Proceedings, International Congress of Linguists.

Jackendoff, Ray S. 1977. X' syntax. Cambridge, Mass.: MIT Press.

Radulphus Brito (Raoul le Breton). c. 1300. Quaestiones super Priscianum minorem, ed. Heinz W. Enders and Jan Pinborg. 2 vols. Stuttgart-Bad Cannstatt: Frommann-Holzboog, 1980.

Robins, R. H. 1967. A short history of linguistics. London: Longman.

Rosier, Irène. 1983. La grammaire spéculative des modistes. Lille: Presses Universitaires.

Pinborg, Jan. 1967. Die Entwicklung der Sprachtheorie im Mittelalter. (Beiträge zur Geschichte der Philosophie und Theologie des Mittelalters, 42.2.) Münster: Aschendorff; Copenhagen: Arne Frost-Hansen.

Tesnière, Lucien. 1959. Éléments de syntaxe structurale. Paris: Klincksieck.


The content and opinions expressed on this Web page do not necessarily reflect the views of,
nor are they endorsed by, the University of Georgia or the University System of Georgia.