Grammatical Relations Research Paper

Grammatical relations fulfill distinguishing and characterizing functions. First, they permit identification of the various participants in the event (what was sent, to whom and by whom it was sent), secondly they classify the NPs in the given sentence in terms of a restricted set of characteristic properties, allowing the identification of NPs in different sentences. Thus, in (3) A message was sent to Mary (by Peter), (4) Mary was sent a message (by Peter), both semantically related to sentence (1), the subjects will be, respectively, the noun phrases a message and Mary; in (5) That he won is fine, there is a clausal subject that he won; while in (6) It is fine that he won and (7) It is dark, the subject is the pronoun it (the so-called dummy subject). The concept ‘subject’ as a homogeneous syntactic relation unites different kinds of specific constituents. Each grammatical relation has a specific function and specific linguistic means for its expression.

1. What Do Grammatical Relations Mean?

Despite the widespread use and obvious necessity of the terms subject and object, traditional grammar did not succeed in defining their meaning adequately. Definitions for ‘subject’ of the type ‘what the sentence is about’ are at first sight suitable for sentences (1)–(4), but they are dubious for (5), while the subjects in (6) and (7) clearly do not fall under a definition of this type.

In a series of investigations of the last decades (cf., for instance, the pioneering work of Keenan 1976) it has been established that in the most elementary sentences subject NPs correlate statistically with the meanings of various autonomous semantic domains.

First, subjects are linked to the semantic role properties of participants in the given situation. The subject usually encodes the agent, if there is one in the sentence. If there is no agent in the sentence, then the prime candidate for subject position is the patient. In the presence of an agent the patient is the prime candidate for the position of direct object. Sentences with an agent and patient constitute the prototypical transitive clause with a Subject and direct object. Sentences with participants in other roles are assimilated to varying extents to the structure of transitive or intransitive clauses, or use the position oblique object. The participant with the role of recipient beneficiary usually occupies the position of indirect object.

Second, there are statistical correlations with the discourse (communicative) domain. Thus, subjects often communicate known (given) information, while objects communicate new information; subjects often coincide with the focus of empathy, i.e., with that participant from whose point of view the situation is described, and with the topic (theme), while objects often coincide with the comment (rheme).

Third, the ontological domain, connected with the properties of referents, influences the subject or object position of NPs. Nps denoting animate, and especially human participants tend towards the subject position, inanimates towards the object position (Grammatical Gender). Within the set of humans the greatest tendency towards subject position is shown by the speech act participants (concepts from the deictic domain).

In addition, concepts from the referential domain, which classify kinds of referents denoted by NPs and facilitate their correct identification by the addressee, also correlate with grammatical relations. Thus, the subject has a presumption of existence, usually has autonomous reference, and is definite and specific, while the direct object is often nonspecific.

However, it is not possible to assign any of the above-mentioned meanings as obligatory for the subject or direct object; in each given sentence these grammatical relations satisfy only some of these meanings.

What, then, is the invariant meaning of grammatical relations, if indeed one exists? There is no generally accepted answer to this question. However, the following can be offered as the most plausible answer. Grammatical relations order NPs in the sentence by rank in accordance with their overall communicative value in a concrete utterance. The ordering takes place by means of a system of rules which takes into account the concrete meanings of NPs according to all the relevant semantic domains, so that the first rank is assigned to the subject, the second to the direct object, then the indirect object and oblique objects, as shown in the so-called ‘relational hierarchy’:

where Su represents ‘subject,’ DO ‘direct object,’ IO ‘indirect object,’ and OO ‘oblique object.’ The subject occupies a special position in this hierarchy. The subject is the most essential component of the sentence, and in some languages it is structurally necessary, in contrast to all other grammatical relations. The other grammatical relations are ranked primarily according to their degree of difference from the subject. In this way, grammatical relations have not an absolute, but rather a relative syntactic value in the sentence, and the relation of this value to meanings of various kinds is indirect in nature.

Grammatical relations are a means of cumulative expression of permitted combinations of more elementary meanings, primarily from the role and discourse domains. Given this, then, first, the same grammatical relation in different sentences of the same language can have a different set of meanings; and second, in different languages grammatical relations can be organized differently, since different combinations of these meanings may be grammaticalized into different grammatical relations.

It can easily be seen that the subject and direct object are furthest removed from the semantic role of a NP and that their basic characteristic is their rank position in the hierarchy. Together with the predicate they form the nucleus of the predicate–argument structure. Oblique objects, in contrast to subject and direct object, have a specific semantic role (e.g., place, time, cause, manner), and in a clause there can be more than one oblique object. The indirect object has an intermediate status. On the one hand, it is considerably more loaded, in terms of semantic role and ontological semantics, than are the subject and direct object, and in this it is like oblique objects, but on the other hand it is the easiest to incorporate into the nucleus, compare (1) and (8) Peter sent Mary a message. In (8), in contrast to (1), Mary occupies the position of direct object.

2. How Are Grammatical Relations Encoded?

Languages use different means to encode grammatical relations. The best known morphological category specifically designated to encode grammatical relations is case. Case marking is present in many languages, e.g., in Latin and Greek, in the Slavic and Dravidian languages, and in the languages of Australia. Under case marking the grammatical relation is expressed by the morphological form of the head of the NP (i.e., of the noun, as in Latin and Russian) or by the forms of other structural constituents of the NP (e.g., forms of the article and noun in German). Alongside inflectional morphology many languages make use of grammatical words or particles (usually prepositions or postpositions), which combine with a NP to give analytical constructions that indicate the kind of grammatical relation (e.g., the combination of pre- position to and NP, such as to Mary in (1), encodes the indirect object). The linear order of NPs in the sentence is the positional equivalent of inflectional or analytical marking and is calculated according to the position of the NP relative to the verb, relative to the beginning or end of the sentence, or relative to other NPs. Morphological marking may be indicated not on NPs, but on the verb by means of agreement: the form of the verb expresses information about the grammatical properties of that NP which stands in a particular grammatical relation to the verb. An example in English is agreement of the verb with a third-person subject in number, combining with the indication of grammatical relations through linear order. Some languages have polypersonal agreement of the verb with two or more NPs. Thus, in Abkhaz the verb has three agreement positions indicating agreement with specific syntactic types of NPs.

3. The Syntactic Functions Of Grammatical Relations

The question of which grammatical relation a particular NP occupies in the sentence can by no means always be decided on the basis of the coding technique. The same means may be used to encode different grammatical relations and the same grammatical relation may use different coding techniques. Thus, in English an immediately postverbal prepositionless NP is not always a direct object: (9a) He was sleeping all night, cf. the equivalent Russian sentence with a time adverbial in the accusative: (9b) On spal sju noc , where the accusative is the typical case of the direct object. Conversely, in Russian the direct object can stand in the genitive case in the context of negation, e.g., (10a) On kupil knig-u (acc.) ‘He bought a book’ vs. (10b) On ne kupil knig-i (gen.) ‘He didn’t buy a book.’ For this reason the question often arises of the criteria by which one can assign different grammatical relations to identically marked constituents or assign the same grammatical relation to differently marked constituents.

The decisive diagnostic role in identifying grammatical relations is played not by coding means, but by syntactic tests. The point is that the coding technique is merely the outer packaging that helps identify grammatical relations in the sentence, whereas the grammatical relations are really components of the syntactic structure of the sentence; their identity or nonidentity is determined in the last analysis on the syntactic level. In many languages the system of rules of syntactic organization operates to a considerable extent on the basis of units characterized in terms of grammatical relations, and thus the identity of syntactic behavior of constituents differing in form is an important criterion for uniting them as a single grammatical relation.

Within the clause, voice transformations are oriented to grammatical relations. Thus, the difference between active and passive voices is that different grammatical relations adhere to the same NPs in active and passive versions. Passivization is usually characterized as the promotion of the syntactic status of the direct object to that of subject and/or demotion of the syntactic status of the subject. In English both processes occur in passivization, cf. (1) and (2). The subject NP of the active loses its subject status (it is expressed by the prepositional phrase by Peter), while the direct object NP (a message) becomes subject (promotional passive). There are also languages in which passivization involves only the demotion of the status of the subject NP (demotional passive). However, in all cases passivization is defined in terms of grammatical relations, irrespective of the coding techniques used in the particular language.

Among nearly universal properties are constraints on reflexivization, operating in cases of coreference between two NPs in the same clause, e.g., (11) Mary hates herself, (12) Mary does not think about herself. Here, one of the NPs (Mary) must be in subject position and is a full identification of the referent (the controller of reflexivization). The other, nonsubject, coreferential NP is replaced by the reflexive pronoun (the target of reflexivization).

Under coordination there are also constraints on grammatical relations. Thus, in English coordination reduction is possible only with coreferential subjects. The second coreferential subject is omitted, e.g., (13) John washed his hands and—[ John] began to eat. Another construction formulated in terms of subjects is switch reference, attested in various languages: clauses are expressed differently according to whether their subjects coincide or not (usually by marking the verb of one of the clauses).

As a syntactic test one can use constructions with sentential complements with verbs like ‘want,’ ‘must,’ ‘begin.’ Sentential complements are usually expressed differently depending on the coreference versus noncoreference of the subject of the sentential complement and the subject of the main clause, cf. (14a) Peter wants to congratulate Mary, but (14b) Peter wants Mary to congratulate him. In (14a) the subject of the sentential complement is coreferential with the subject of the verb want, and is therefore not expressed overtly in the subordinate clause. In (14b) the coreferential NP is the direct object, therefore it cannot be omitted, but must be expressed by means of a pronoun.

The formation of relative clauses is constrained, in many languages, by the grammatical relation of the target of relativization. For instance, reduced (participial) relative clauses in English are restricted to subject targets of relatives: (15a) the student buying a book, (15b) the book bought by the student. In (15a) the target of relativization is in subject position (The student buys a book), therefore the relative clause is formed directly from this structure. In (15b) the target of relativization is the direct object (The student bought the book), therefore in order to construct the participial version it is necessary first to passivize the relative clause, promoting the status of the target of relativization to subject position (The book was bought by the student). In some languages relativization is constrained also by the grammatical relation of the controller of relativization.

Not infrequently constraints on the formation of adverbial clauses also rely on grammatical relations. In Russian there is a set of subordinate clauses expressing background activity, the so-called converbal (gerundive) clauses. The subject of the converbal clause must be coreferential with that of the main clause (and is omitted in the converbal clause): (16) Pokupaja etu knigu, student potratil na poiski massu remeni ‘Buying this book, the student spent a lot of time in searching.’ In (16) the omitted subject of the converbal clause and the subject of the main clause is the NP student. Sentence (17) is ungrammatical: (17) *Pokupaja etu knigu, u studenta propal koselek ‘Buying this book, the student’s purse got lost,’ since u studenta, the NP in the main clause coreferential with the subject of the converbal clause, is not the subject (the subject of the main clause is koselek ‘purse’).

To the extent that many syntactic rules are operations in terms of grammatical relations, these rules can be used as tests to determine the syntactic status of a NP in morphologically ambivalent instances.

4. Initial vs. Derived (Primary vs. Secondary) Grammatical Relations

Certain syntactic processes change the grammatical relation of a particular NP or delete it. Such processes change the valency of a sentence. In such instances it is important to distinguish initial and derived grammatical relations of NPs. Thus, in the passive the initial subject loses this grammatical relation, while the initial direct object becomes the subject of the passive sentence. It follows from this that syntactic processes are often formulated in terms of initial syntactic relations which are not present in the observable, derived syntactic structure. Their existence can nonetheless be reconstructed on the basis of the syntactic process that conditions the observed structure. In a series of instances properties of the initial grammatical relation of a NP are partially retained. For instance, in the examples (18a) Uncle Tom built this hut for himself and (18b) This hut was built by Uncle Tom for himself, the controller of the reflexive is in both instances the initial subject Uncle Tom, even though in (18b) the derived, observable subject is this hut.

5. Nonuniversality Of Grammatical Relations

In linguistics the aprioristic assumption of the universality of grammatical relations was long dominant. Many current syntactic theories derive from this assumption, in particular generative grammar and the theory of Relational Grammar. In the last decades, however, as a result of considerable progress in studying the nature of grammatical relations and thanks to the considerable broadening of empirical linguistic material this assumption is no longer self-apparent.

First of all, the functional and formal properties of grammatical relations characterized above ( polyfunctionality of content and cumulative expression) testify to their specific semiotic nature. Why should such a technique of identification and characterization of noun phrases be an obligatory property of natural language? Conversely, it is natural to allow that different languages might grammaticalize various concepts of different semantic domains and package them differently in linguistic form. What we identify as grammatical relations might be filled with different content in different languages and, moreover, one might find not a cumulative, but a separative technique of packaging elementary concepts. Finally, there is no a priori basis to assume that the concepts of some or even all the relevant domains should be grammaticalized in each language, given that so far not a single semantic domain has been found that would be grammaticalized in all languages (be it the category of gender, number, tense, aspect, mood etc.). Even as apparently as indispensable ontological category as part of speech has recently increasingly lost the assumption of universality.

Second, uncritical attempts to use relational concepts with respect to different languages often encounter serious empirical difficulties. One may start with the observation that the morphosyntactic means of encoding NPs in some languages is not only different from, but sometimes not even comparable with the encoding characteristic of Standard Average European languages. If one uses the widespread notation for the nuclear arguments with the symbols S, A, P (or O), where S is the nuclear argument of an intransitive clause, and A and P the agent-like and patient-like arguments, respectively, of a transitive clause, then in European languages the S/A arguments have identical encoding, grouping them together as the grammatical relation of subject, in opposition to the encoding of the P argument. Such a system is called accusative alignment. Alongside this possibility, many languages found in various continents have the ergative pattern, uniting the S/P arguments in opposition to the A argument. Here the S/P arguments are usually morphologically unmarked, while the A argument is marked (ergative case or its analog). The ergative alignment of the nuclear arguments creates an empirical problem for the establishment of grammatical relations.

In the languages of the world other patterns are also found. One possibility is a pattern that distinguishes agent-like and patient-like arguments of both transitive and intransitive clauses (active alignment). Under active alignment one finds different expression of the single argument of verbs like ‘run’ and verbs like ‘fall.’ There is even the tripartite scheme, distinguishing all three arguments, S, A, and P. Finally, there is the neutral pattern, which fails to differentiate the S/A/P arguments by means of coding.

Languages characterized by ‘unusual’ nonaccusative patterns are quite varied from a syntactic point of view. For instance, languages with ergative alignment may behave as syntactically accusative, i.e., in defiance of the coding technique S and A arguments are united in a single grammatical relation subject. Much more interesting is the situation where ergative alignment is also realized at the syntactic level: S and P arguments are united syntactically in the highest ranking grammatical relation, equivalent to the subject under accusative alignment. The A argument is also a nuclear argument, i.e., it occupies the second position in the Relational Hierarchy. Finally, there are ergative languages in which it is impossible to rank S, A, and P at the syntactic level. In languages of this type it is justifiable to speak of the absence of grammatical relations.

Among languages with active alignment are also found languages behaving accusatively at the syntactic level, as well as languages lacking syntactic relations. So far there is extremely little information available on languages with tripartite or neutral alignment. It is, however, quite clear that they too are quite varied. They may have or not have grammaticalized grammatical relations. In particular, there is evidence for the existence of languages with neutral alignment lacking grammatical relations. But one can assume the existence of a neutral alignment language opposing nuclear and nonnuclear arguments. In such a language the nuclear arguments would be equivalent to subject, and the nonnuclear arguments to oblique objects. If such a language is attested, then the thesis of the uniqueness of the subject will have been overturned.

Thus, grammatical relations both have a narrower sphere of use and are considerably more varied than is usually assumed.

It should be kept in mind that the thesis of the nonuniversality of syntactic relations is not universally held, and is even ignored in many works. However, the assumption of nonuniversality is preferable by far in empirical research of little-studied languages, since it does not impose an aprioristic standard on the description and helps to see the real state of affairs in each language under discussion.

6. Grammatical Relations And The Functional Approach To Language

The concept of grammatical relation has been known in linguistics since the time of the grammarians of antiquity. However, significant progress in understanding the nature of grammatical relations has been achieved only in the last decades thanks to the orientation towards the functional explanation of grammar, which has permitted establishing the polyfunctional character of the elementary categories which form the clusters known as grammatical relations. The majority of these categories were themselves discovered and investigated in the framework of the functional approach.

In this connection the question of the existence of different patterns of encoding S, A, and P arguments merits special attention. Proceeding from the now widely established assumption of the iconicity of the linguistic sign, replacing the Saussurean assumption of the arbitrariness of the sign, one can allow for the existence of a functional motivation for the various coding patterns. These patterns unite arguments with different elementary roles in different ways. The basic prototype of all these groupings is given by the agent and patient of the transitive construction. From this viewpoint active alignment is a natural metonymic extension of the elementary roles of agent and patient, generalizing them to the hyperroles (macroroles) of actor (the participant in the event which performs, effects, instigates, or controls it) and undergoer (the participant in the event which does not perform, effect, instigate, or control it, but rather is affected by it in some way). Arguments with different elementary roles can be interpreted in terms of these hyperroles.

Accusative alignment is also motivated essentially not by grammatical relations, but by the hyperroles principal and patientive. The principal indicates the main participant, the ‘hero’ of the situation, who is primarily responsible for the fact that this situation takes place. In the transitive clause the principal is, of course, the agent-like argument, while in the intransitive clause the single nuclear argument has no competitors for the role of principal. The patientive is the most patient-like participant of a multi-participant event. This role unites, in addition to the patient, arguments of other types.

Ergative alignment articulates the S, A, P arguments by means of other hyperroles, absolutive and agentive. The absolutive is the metonymic extension of the patient to intransitive verbs, it is the immediate, nearest, most involved or affected participant in the situation. In the transitive clause the patient is most involved in the situation. In the intransitive clause there is no competition for the role of closest and most involved participant in the situation. The absolutive in the transitive clause is in opposition to the agentive, the most agent-like participant of the multi-participant event.

Tripartite alignment distinguishes agentive and patientive of the transitive clause, distinguishing them from the single nuclear argument of the intransitive clause.

All these means of generalizing the elementary roles are in principle motivated, but a language has the option of grammaticalizing just one of them as its basic pattern. In addition, a language may have different patterns in different contexts, and examples of this kind have been attested.

7. Grammatical Relations And The Cognitive Approach To Language

The ‘statistical’ correlations between concepts of different semantic domains noted above and which are grammaticalized in grammatical relations and the means of their encoding are not chance, rather they have a cognitive nature, reflecting the human capacity for categorizing reality. Behind the limitless variety of whole grammatical systems lie the same stable cognitive structures of human thought and the pragmatic strategies for their verbalization. The reconstruction of these structures and strategies on the basis of data from different languages is the task of cognitive typology.

Bibliography:

1. Aissen J 1999 Markedness and subject choice in optimality theory. Natural Language and Linguistic Theory 17: 673–711
2. Cole P, Sadock J M (eds.) 1977 Grammatical relations. Syntax and Semantics. Academic Press, New York, Vol. 8
3. Croft W 1991 Syntactic Categories and Grammatical Relations. The University of Chicago Press, Chicago
4. Dik S 1997 The Theory of Functional Grammar. Mouton de Gruyter, Berlin
5. Dixon R M W 1994 Ergativity. Cambridge University Press, Cambridge, UK
6. Dixon R M W, Aikhenvald A Y (eds.) 2000 Changing Valency: Case Studies in Transitivity. Cambridge University Press, Cambridge, UK
7. Dryer M 1986 Primary objects, secondary objects, and antidative. Language 62: 808–45
8. Dryer M 1997 Are grammatical relations universal? In: Bybee J, Haiman J, Thompson S (eds.) Studies in Language Function and Language Type. Benjamins, Amsterdam, pp. 115–43
9. Givon T (ed.) 1997 Grammatical Relations: A Functional Perspective. Benjamins, Amsterdam
10. Keenan E 1976 Towards a universal definition of ‘‘subject’’. In: Li Ch N (ed.) Subject and Topic. Academic, New York, pp. 303–34
11. Kibrik A E 1997 Beyond subject and object: Toward a comprehensive relational typology. Linguistic Typology 1: 279–346
12. Li Ch N (ed.) 1976 Subject and Topic. Academic, New York
13. Perlmutter D M (ed.) 1983 Studies in Relational Grammar. The University of Chicago Press, Chicago
14. Plank F (ed.) 1979 Ergativity: Toward a Theory of Grammatical Relations. Academic, London
15. Van Valin R, LaPolla R J 1997 Syntax: Structure, Meaning, and Function. Cambridge University Press, Cambridge, UK