Valence at the clause level
As we have seen, valence is described in Cognitive Grammar in terms of correspondences between the component structures that make up a grammatical construction. These correspondences are accounted for in terms of autonomy, dependence and elaboration. As an illustration of how these ideas can be applied to the clause, consider the verb see. This verb expresses a temporal relation or PROCESS and has a schematic TR and LM as part of its representation. The schematic TR and LM are the elaboration sites. In a clause like (8), the NP George elaborates the schematic TR of see and the NP his childhood friend elaborates the schematic LM.
(8) George saw his childhood friend.
Because the verb relies on the two NPs to elaborate its schematic TR and LM, the verb is conceptually dependent and the two NPs are conceptually autonomous. As this example illustrates, the Cognitive Grammar account of the constituency of the clause rests upon the same assumptions as the account of word structure and phrase structure. The only difference is that the com ponent parts of these grammatical constructions become increasingly complex as we move from word, via phrase, to clause. As we saw in the last section, the pattern of autonomy and dependence illustrated by example (8) represents the prototypical grammatical construction where a dependent RELATION (here, the verb) relies on an autonomous THING to elaborate some schematic aspect of its structure. While see is a ‘typical’ transitive verb and has two elaboration sites, intransitive verbs like die have only one elaboration site which corresponds to the TR, and ditransitive verbs like give have three elaboration sites. We return to each of these clause types in more detail below. As we will see in the next section, in Cognitive Grammar a subject is the unit that corresponds to the TR of the verb and the object is the unit that corresponds to its LM.
Recall that Langacker describes valency relations as binary, an idea that captures the ‘layered’ structure of complex constructions (section 17.1.6). This idea also accounts for the ‘layered’ structure of the clause. The dependent verb and the autonomous object combine to form a complex unit (VP), of which the verb is the profile determinant. In other words, the VP profiles a PROCESS. The object elaborates the schematic LM of the verb. This is an instance of the head-complement relation. However, the resulting PROCESS (VP) remains a dependent unit, since the verb still has a schematic TR that requires elaboration. The VP then combines with the subject which elaborates its TR. This also represents the prototypical valence relation between a dependent RELATION and an autonomous THING, despite the fact that it is not strictly speaking a head-complement relation since the dependent relation is itself a complex grammatical construction. In Cognitive Grammar terms, however, the valence relation between V and object NP is the same as the valence relation between VP and subject NP: in both cases, the profile determinant is a PROCESS, and in both cases that PROCESS requires the NP to elaborate some aspect of its schematic structure. The resulting construction still profiles a PROCESS, which means that the verb is the profile determinant or head of the clause as a whole, an idea that is central to many theories of grammar. The structure of the prototypical transitive clause is illustrated in Figure 17.3.
Of course, this account suggests that the construction of a clause is a ‘step by-step’ process. However, recall that the usage-based model assumes that even clause-level constructions are stored as wholes if they represent well entrenched units, and thus give rise to schemas that mirror their structure. From this perspective, viewing a construction in terms of the ‘order’ in which it is ‘built’ is rather meaningless. However, Langacker (1987: 319) suggests that the natural intonation pattern of a basic transitive clause supports the traditional partition of the clause into subject and predicate, or NP and VP, an idea that we discussed in Chapter 14.
Arguments versus modifiers
The participants that are required by the verb to complete its meaning are described as the arguments of the verb, which is the predicate or semantic core of the clause. (This sense of the term ‘predicate’ should not be confused with the traditional grammar sense of the term which refers to the verb phrase.) As we have seen, arguments of the verb are autonomous. Consider the examples in (9).
In (9a), the verb put requires three participants to complete its meaning and therefore has three elaboration sites, a schematic TR (elaborated by Lily), aschematic LM (elaborated by the letters) and a schematic ‘destination’, also known as the secondary landmark. In Cognitive Grammar terms, the verb is conceptually dependent upon the arguments that elaborate its sites. Arguments are often given labels in terms of semantic roles. For example, the NP Lily that elaborates the schematic TR has the role AGENT, which describes an entity that acts with volition and intention. The NP the letters that elaborates the LM has the role MOVER, which is Langacker’s term for an entity that undergoes a change of location. The relational unit (PP) in the bin elaborates the schematic destination or secondary landmark and has a LOCATION role. We discuss semantic roles in more detail in the next section, where we will see how they interact with the grammatical functions subject and object. While Langacker calls both the subject Lily and the object the letters in (9a) nominal complements, he calls the PP in the bin a relational complement because its profile determinant is a RELATION.
Example (9b) has a rather different structure from example (9a). This is because the PP across a crowded room does not elaborate any part of the verb’s substructure. The verb see requires two arguments to elaborate its schematic TR, (Lily), and its schematic LM, (George). This means that the PP is a clausal modifier or adverbial. In other words, the verb is autonomous in relation to this PP. It is the modifier that is conceptually dependent in this type of relationship. The modifier relies on the verb plus its arguments, the clause-level PROCESS, to elaborate its schematic PROCESS TR. This explains the optionality of such modifiers in contrast to the obligatory presence of subjects and objects. Clause level modifiers are not always PPs. They can also be adverb phrases (very sincerely), NPs (these days) or other clauses (humming a happy tune). What these all have in common is that they have a schematic PROCESS as part of their meaning which is elaborated by the (main) clause.
Copular clauses Recall from Chapter 14 that some clauses do not contain a prototypical content verb. These are clauses headed by the copular or ‘linking’ verb be, which takes a subject predicative complement. Consider the examples in (10).
As we saw in Chapter 14, the bracketed constituent in each example is called the subject predicative complement, because it defines or describes the subject, occurs as the complement of be and is predicative in the sense that it contributes the semantic core of the clause. Some grammars limit the term ‘predicative complement’ to the NP (10a) and the AP (10b), describing the PP (10c) instead as an adverbial complement because it expresses the same kind of meaning as an adverbial (for example in describing the place of the subject).
Given that most grammatical theories, including Cognitive Grammar, view the lexical verb as the ‘heart’ of the clause, questions arise concerning the formation of copular clauses. In some formal approaches, the copula be is treated as a semantically empty verb that does not have independent argument structure (in other words, does not behave as a predicate). This semantically empty verb licenses the combination of subject and predicate (NP, AP or PP) by enabling the formation of a clause, which it does by contributing finiteness (in English, main clauses have to be finite). From this perspective, the copula ‘mediates’ between subject and predicate by licensing a well-formed finite clause structure, complete with agreement, tense and so on. This entails that the copular verb has a subject and a complement in structural terms, even though the constituents in these positions are not semantically related to the copula. Instead, the subject and the predicative complement are semantically related.
The Cognitive Grammar account is reminiscent of the formal account in the respect that the verb be is described as maximally schematic. According to this analysis, the verb be designates a schematic stative PROCESS. However, the verb be has its own semantic structure in Cognitive Grammar: like a contentful lexical transitive verb, it has a schematic TR and LM. These are elaborated by subject NP and by either nominal (10a) or relational predications (10b)–(10c), respectively. In other words, a rocket scientist is a nominal complement of the verb be in (10a), while drunk in (10b) and under the table in (10c) are relational complements of the verb be. Langacker points out that the correspondences between the subparts of the copular construction entail that the subject is not only subject of the verb be, but also subject of each component part of the VP and subject of the VP as a whole. In this way, the Cognitive Grammar analysis captures the same intuition as the formal analysis concerning the predicative nature of the subject complement and the semantic relationship between subject and complement.
It is important to point out that the status of the bracketed units in (10) is rather different from the status of these units in other types of constructions. For example, while it is not unusual for a NP like a rocket scientist to occur either as subject or object of a clause headed by a lexical verb, there is an important difference between its occurrence in that type of construction and its occurrence as a subject complement. This difference relates to grounding. Consider the following examples:
In (11a), a rocket scientist is the direct object. In (11b), the same NP is the subject. In both these examples the NP is grounded. In other words, the indefinite article a grounds the noun and the resulting NP and designates a specific individual, even though this individual is not familiar to the hearer. In cases like these, the NP is a referring expression, which means that the speaker uses the expression to pick out a specific individual in the world. In contrast, the NP a rocket scientist in (10a) is not grounded. An indefinite predicate nominal, in other words an indefinite noun phrase that functions as subject complement, is not a referring expression. Instead, it describes a property of the subject and in this respect might more accurately be characterised as a RELATIONAL PREDIC ATION than a NOMINAL PREDICATION (see Langacker 1991: 65–6). This idea captures its predicational role in the clause but raises questions about the nature of the autonomy-dependence relation between the predicate nominal and the copula.
Observe that if the predicate nominal is marked as definite, something interesting happens to the clause:
The definite NP the rocket scientist in (12a) is now grounded, which means that it now functions as a referring expression and picks out a particular individual that is equivalent to Lily. This type of copular clause is described as equative rather than predicative and is characterised by reversibility (12b). As Taylor (2002: 361–2) observes, it is not clear why the ungrounded predicate nominal in examples like (10a) should have an article at all in light of this discussion. Indeed, languages including French, German and Spanish omit the article in constructions like this, and some English expressions also license the absence of the article:
Observe that what the constructions in (13) share in common is that the predicate nominal designates a unique role: the expectation is that there is only one Queen of England, chairperson, crayon monitor and so on. Despite these differences between predicate nominals on the one hand and subjects and objects on the other, the NPs in all these functions share the same property of conceptual autonomy in relation to the verb (with the possible exception of the indefinite predicate nominal), hence their status as complements in Cognitive Grammar.
In contrast, an attributive adjective (14a) and a predicative adjective (14b), while both profiling ATEMPORAL RELATIONS, do not have the same status with respect to autonomy and dependence. The attributive adjective daft in (14a) is dependent in relation to the autonomous noun and therefore a modifier. The predicative adjective daft in (14b) is autonomous in relation to the dependent copula verb and is therefore a complement:
It is worth observing here that not all adjectives can occur in both attributive and predicative positions (15). Furthermore, some that can occur in both positions take on a distinct sense in each position (16).
Embedded clauses As we saw briefly in Chapter 14, clauses can function as subparts of complex sentences. When this happens, the (subpart) clause is called an embedded clause or a subordinate clause. This is illustrated by the examples in (17) where the embedded clauses are bracketed.
While these examples do not represent every type of embedded clause that is possible in English, they do provide a representative sample. Although these constructions look very complicated, they actually represent construction types that we have already seen. This is because clauses can perform the same kinds of grammatical functions that smaller grammatical constructions like nominal expressions can perform. In (17a), the embedded clause is the object, while in (17b) it is the subject. In (17c), the embedded clause is the comple ment of a noun because it completes its meaning, while in (17d) it is a nominal modifier. This is called a relative clause and is identifiable as such by the fact that we can substitute which for that (the love letters which she burnt), which is not possible in (17d) (*the idea which George was a dreamboat). In examples (17a) to (17d), the embedded clauses are all finite. This means that, with the exception of the relative clause, they are all capable of functioning as independent clauses. This means that they are PROCESSES.
In examples (17e) to (17i), the embedded clauses are non-finite. In (17e), for example, the embedded verb is in its bare infinitive form (compare *George could make Lily cries), while in (17f) the embedded verb is in its to-infinitive form. In both of these examples, the embedded clause is the object. In examples (17g) to (17h), the embedded clauses contain participles and lack subjects. While in (17g) the embedded clause is the subject, in examples (17h) and (17i) the embedded clause is an adverbial clause or modifier. Recall from Chapter 17 that Langacker treats infinitives and participles as ATEMPORAL RELATIONS. We will leave it for the reader to work out the autonomous and dependent units in these examples (see exercise 17.4).
