Figure 1 Active and Passive analyses.
Relational Grammar (RG) refers to a formal approach to SYNTAX that takes GRAMMATICAL RELATIONS like subject, direct object, and indirect object to be indispensable and primitive notions. According to Perlmutter (1980), they are indispensable for achieving three goals of linguistic theory:
RG was motivated by syntactic work in a wide range of languages that revealed both language-particular and crosslinguistic generalizations involving subject and object. These included generalizations about morphological processes like agreement and case marking, semantic processes like the interpretation of anaphors, and grammatical processes like passive and relative clause formation. The earliest codified RG work (1974 lectures by David Perlmutter and Paul Postal; see Perlmutter and Postal 1983a, 1983b) argued that the relations subject, direct object, and indirect object were primitives because they could not be defined universally in terms of relations like linear order or dominance. These were the relations which formed the cornerstone of Transformational Grammar (TG), the prevailing theory of syntax in the early 1970s. The proposals of RG chiefly concern the relational structure of clauses.
Relational approaches to syntax were pursued in several forms during the mid-1970s, some a development of the ideas of Perlmutter and Postal, others a reaction to them (e.g., the papers in Cole and Sadock 1977). But the name "Relational Grammar" came to be associated with work that adheres to the program laid out by Perlmutter and Postal, work exemplified in Perlmutter 1983, Perlmutter and Rosen 1984, Postal and Joseph 1990; see Dubinsky and Rosen 1987 for a bibliography. RG, narrowly construed then, conceives of a clause as a network of grammatical relations. A clause might consist of a predicate, a subject, and a direct object. Under such a conception, two clauses that are superficially very different, perhaps because they are clauses from different languages, might in fact be structurally very similar. The fact that agreement, for example, is with the subject, whether the subject is clause-initial, clause-second, or clause-final, is directly statable and the parallelism between the two cases patent.
A crucial assumption of RG, inherited from TG, has been that the description of a clause refers not only to its superficial structure but also to a deeper structure and possibly to several intermediate levels of structure. In RG, these levels are called strata, and the RG position is that there is no one stratum at which all the properties associated with subject or object hold; rather these are apportioned at different strata. Relations in the initial stratum are linked to semantic role (e.g., agent, patient; Rosen 1984), whereas relations in the final stratum determine more superficial phenomena like agreement and word order. Clause pairs such as active/passive that express the same proposition in relationally different ways generally share the same initial stratum, accounting for their synonymy, but diverge in later strata (Perlmutter and Postal 1983a). The active sentence the committee recommended us has a single stratum, one in which the committee is subject and the pronoun us is direct object. The passive version, we were recommended by the committee, has, in addition, a second stratum in which we has been "advanced" to subject, and the committee "demoted" to an RG-specific relation called chomeur (= French "unemployed"). (This terminology reflects the relational hierarchy, a ranking of grammatical relations: subject > direct object > indirect object > other.) The diagrams in figure 1 represent these analyses. A and B are clausal nodes, and the grammatical relations borne by each element to the clause are organized into strata, represented by horizontal rows.
Figure 1 Active and Passive analyses.
Grammatical relations are represented by "1" = Subject, "2" = Direct Object, "Cho" = Chomeur, "P" = Predicate. Differences in word order and agreement between active and passive are due then to their different final strata. Universally, where grammatical relations determine word order, it is final stratum relations that are relevant. This generalization is deeply embedded in RG, which takes word order to be entirely irrelevant to nonfinal strata.
The demotion of the initial subject to chomeur in passive clauses reflects a universal principle of RG, the Stratal Uniqueness Law, which stipulates that a stratum may contain at most one subject, one direct object, and one indirect object. This law prevents the initial stratum subject in passive from persisting as subject when the direct object advances to subject. The fact that the chomeur in English passive is marked with by is, in contrast, a language-particular fact. RG has proposed an inventory of grammatical relations, and a set of principles ("laws") governing the well-formedness of networks (Perlmutter and Postal 1983b). The laws are linguistic universals (goal (1) above), and it is through the laws that RG proposes to characterize the class of grammatical constructions (goal (2)). Under certain conditions, the laws permit one nominal to assume the grammatical relation borne by another at a prior stratum, and they thereby define a typology of relation-changing constructions. Advancement to subject is seen in passive; advancement to direct object is also possible. In RG terms, Every student gave the teacher a present involves advancement of an indirect object, the teacher, to direct object. As direct object, it may advance further to subject in the passive version, The teacher was given a present by every student. The Stratal Uniqueness Law prevents the initial direct object from persisting as such when the indirect object advances to direct object. This explains why it cannot passivize: *A present was given the teacher by every student. Also possible are demotions, as well as various kinds of clause merger (Davies and Rosen 1988). All these constructions are subject to the Stratal Uniqueness Law. The Stratal Uniqueness Law represents an empirically testable claim about syntactic organization and has not been uncontroversial (Perlmutter and Postal 1983b). Other important laws constrain the distribution of chomeurs and impose the requirement that the final stratum of every clause (but not necessarily any other stratum) have a subject (Perlmutter 1980; Perlmutter and Postal 1983b).
RG played an important part in the evolution of syntactic theory from the 1970s to the 1980s. During this period, languages other than English had significant impact on syntactic theorizing, leading to an increased appreciation of linguistic universals, and the need to distinguish more clearly between the universal and the language-particular. RG represents one early response to these issues, and it was extended in more formal work on Arc Pair Grammar (Johnson and Postal 1980). A number of ideas pioneered by RG were incorporated into other theories of the 1980s. The indispensability of grammatical relations has been a key assumption of LEXICAL FUNCTIONAL GRAMMAR, the disassociation of word order from more abstract syntactic representation was adopted in diverse guises by Lexical Functional Grammar and Generalized Phrase Structure Grammar, and the influential Unaccusative Hypothesis (Perlmutter 1978) was incorporated by Government Binding Theory (MINIMALISM). RG is currently pursued in Mapping Theory (Gerdts 1992), a relationally based typological approach to language difference.
Cole, P., and J. Sadock, Eds. (1977). Syntax and Semantics8:Grammatical Relations. New York: Academic Press.
Davies, W., and C. Rosen. (1988). Unions as multi-predicate clauses. Language 64:52-88.
Dubinsky, S., and C. Rosen. (1987). A Bibliography on Relational Grammar through May 1987 with Selected Titles on Lexical-Functional Grammar. Bloomington, IN: Indiana University Linguistics Club.
Gerdts, D. (1992). Morphologically mediated relational profiles. Proceedings of the Eighteenth Annual Meeting of the Berkeley Linguistics Society. Berkeley, pp. 322-337.
Johnson, D., and P. Postal. (1980). Arc Pair Grammar. Princeton, NJ: Princeton University Press.
Perlmutter, D. (1978). Impersonal passives and the unaccusative hypothesis. Proceedings of the Fourth Annual Meeting of the Berkeley Linguistics Society. Berkeley, pp. 157-189.
Perlmutter, D. (1980). Relational grammar. In E. Moravcsik and J. Wirth, Eds., Syntax and Semantics 13: Current Approaches to Syntax. New York: Academic Press, pp. 195-229.
Perlmutter, D., Ed. (1983). Studies in Relational Grammar 1. Chicago: University of Chicago Press.
Perlmutter, D., and P. Postal. (1983a). Towards a universal characterization of passivization. In D. Perlmutter (1983), pp. 3-29.
Perlmutter, D., and P. Postal. (1983b). Some proposed laws of basic clause structure. In D. Perlmutter (1983), pp. 81-128.
Perlmutter, D., and C. Rosen, Eds. (1984). Studies in Relational Grammar 2. Chicago: University of Chicago Press.
Postal, P. and B. Joseph, Eds. (1990). Studies in Relational Grammar 3. Chicago: University of Chicago Press.
Rosen, C. (1984). The interface between semantic roles and initial grammatical relations. In D. Perlmutter and C. Rosen (1984), pp. 38-77.
Aissen, J., and D. Perlmutter. (1983). Clause reduction in Spanish. In D. Perlmutter (1983), pp. 360-403.
Blake, B. (1990). Relational Grammar. New York: Routledge.
Chung, S. (1976). An object creating rule in Bahasa Indonesian. Linguistic Inquiry 7:41-87. Reprinted in D. Perlmutter (1983), pp. 219 - 271.
Davies, W. (1986). Choctaw Verb Agreement and Universal Grammar. Dordrecht: Reidel.
Dryer, M. (1986). Primary objects, secondary objects, and antidative. Language 62:808-845.
Farrell, P., S. Marlett, and D. Perlmutter. (1991). Notions of subjecthood and switch-reference: Evidence from Seri. Linguistic Inquiry 22:431-456.
Gibson, J., and E. Raposo. (1986). Clause union, the stratal uniqueness law and the chomeur relation. Natural Language and Linguistic Theory 4:295-331.
Harris, A. (1981). Georgian Syntax: A Study in Relational Grammar. Cambridge: Cambridge University Press.
Perlmutter, D. (1983). Personal and impersonal constructions. Natural Language and Linguistic Theory 1:141-200.
Postal, P. (1986). Studies of Passive Clauses. Albany: SUNY Press.
Postal, P. (1989). Masked Inversion in French. Chicago: University of Chicago Press.
Rosen, C. (1990). Rethinking Southern Tiwa: The geometry of a triple-agreement language. Language 66:669-713 .
Copyright © 1999 Massachusetts Institute of Technology