Reduction and expansion of a clause

A clause can constitute a sentence, and a sentence carries an illocutionary force. Whenever this is not needed, a clause may be shrunk to a phrase and be downgraded to a component of a sentence. According to its role in this sentence, it may be converted into different phrasal categories such as a nominal, an adjectival or an adverbial. Since the morphological categories and syntactic features distinguishing a sentence from each such phrasal category are many, the process of reducing the former to the latter is gradual. This process is called desententialization.

Conversely, words of one of those non-verbal categories – nouns, adjectives, adverbs – may have valency and may accept modifiers, so they can be expanded into phrasal categories. Moreover, members of each of these categories can be produced by conversion of a verbal base. Such deverbal nouns, adjectives and adverbs may be expanded into complex phrases even more easily than basic nouns, adjectives and adverbs.

We thus get complexity at the phrase level which may be viewed as partial reduction of a clause or as an expansion of a word on the basis of analogy to a clause. For syntactic processes of nominalization, adjectivalization and adverbialization, the reductive perspective may be more fruitful. For the phrasal syntax of deverbal derivates, the expansion perspective may be more fruitful. In what follows, primarily the first perspective will be taken.

Reduction processes look very different depending on which kind of construction, more specifically which kind of grammatical relation exists between the clauses:

• In coordinative constructions, there are reductive processes, but they have nothing to do with desententialization in the sense relevant here. They are treated in the section on fusion of clauses.
• In subordinative constructions, the subordinate clause may be desententialized. This is the topic of the next section. Alternatively, the main clause may be reduced by grammaticalization. That is treated in the following section.
• In cosubordination, likewise one of the members may be reduced by grammaticalization. This is the topic of the last section.

Reduction of subordinate clause

The reduction of a sentence to a noun/adjective/adverb is desententialization. In this process, all those properties that distinguish a sentence and a clause from a word get gradually lost. This is represented in the following diagram (adapted from Lehmann 1988).

The main steps in the process of desententialization are the following:

• finite clause
• non-finite clause
• deverbal derived stem.

These, however, are interlingual grammatical categories in the sense defined elsewhere. This means that

• a language does not necessarily have anything corresponding to each of these steps,
• the boundaries between these steps may be drawn at different positions for different languages.

To the extent that a clause is desententialized, it changes into a word of a non-verbal category, chiefly noun, adjective or adverb, and in the course it gradually assumes the corresponding new grammatical properties. By nominalization, for instance, it acquires number and gender, categories alien to a clause.

Depending on whether desententialization reduces the verb to a non-finite conjugation form or to a stem of a different word class, desententialization amounts to one of the following:

The distinction between syntactically relevant morphology, i.e. inflectional morphology, and lexical morphology, i.e. stem formation, in familiar languages yields a subdivision of the results of the three processes involved as indicated in the last two columns of the above table. However, different languages make different distinctions there.

These processes have a double goal, a positive and a negative one. The positive goal is to create a member of one of those other categories. The negative goal is to suppress the properties of a clause. Desententialization is, in fact, not only a contingent consequence of the acquisition of nominal, adjectival or adverbial properties; it is a goal in itself. The goal has often been called “backgrounding”. This notion, however, is not precise. Let us assume that the speaker constantly distinguishes between those parts of his expressions that are important for him and those that he only mentions because he cannot omit them in the face of grammatical constraints. Then it is true that he lends emphasis to the first kind and pushes the second kind into the background. Lending emphasis means expanding the expression and highlighting it syntactically and prosodically. Pushing it into the background means reducing it in size, syntactic complexity and prosodic prominence. Now this distinction is relevant for many purposes, but not for sententiality vs. desententialization and, thus, for the distinction between main clause vs. subordinate clause. This is shown by sentences such as E1.

A normal use of E1.a is in a situation where all the information of the main clause is presupposed, while the information of the subordinate clause is asserted and may even be focused upon. Thus, there can be no question of backgrounding the subordinate clause in the sense explained. The goals of the speaker here are compatible with an even greater desententialization, with complete nominalization of the subordinate clause, as in E1.b. This means that information-structural foregrounding is fully compatible with subordination and even desententialization.

What is happening here can be circumscribed better with condensation and typicization. In abstracting away from the other kind of information that comes with a finite clause, the speaker concentrates on the situation core, in semiotic terms: on the predicate. Thus the issue is not one of backgrounding the subordinate clause as against the main clause, but rather one of backgrounding, in the subordinate clause, all information that is generally coded by verbal dependents and verbal morphology and to concentrate on the sheer concept of the predicate. See the section on nominalization.

To the extent that the desententialized clause loses internal grammatical structure, it also loses information structure. There is no longer the possibility of lending relief to one part against the other parts. As a consequence, all those syntactic processes that are applicable at the sentence level and also at the level of the finite clause gradually get constrained and fall away. This is formulated in an unforgettable way in the Penthouse Principle (Ross 1973:397):

More goes on upstairs than downstairs.

For instance, from a certain point of desententialization onwards, such information structure processes as left-dislocation or clefting become impossible. More in general, word-order freedom shrinks in subordinate clauses. All of this is a consequence not of backgrounding, but of condensation.

Reduction of main clause

Very often the bulk of the relevant information is in the subordinate clause, to which the main clause only serves as a support. This is so in the causative construction of E2.

Here the main clause predicate is just a function verb (alias light verb) that governs a non-finite clause which contains the important information. Such a configuration is at the basis of the reduction of the main clause predicate to a function verb and, finally, to an auxiliary.

When the main clause verb becomes an auxiliary, it combines with the subordinate non-finite verb into a periphrastic verb form. This distroys the boundary between the main clause predicate and the subordinate construction, and the erstwhile subordinate construction gets “raised” onto the level of the main clause. E3 shows this with the Spanish progressive. The (synchronically illicit) bracketing is, as it were, diachronic. The blue brackets show the construction which is presupposed, where a form of estar ‘stand’ is the sole predicate of the main clause and the gerund alone governs all the oblique dependents. This is then reanalyzed in the way just described. The orange brackets show the structure of this kind of expression in modern Spanish.

Reduction, i.e. grammaticalization, of the main predicate is the principal path along which verbal periphrases come into existence. They no longer belong in the grammar of the complex sentence, because there is only one clause with one verb form; neither do they belong in the grammar of junction, because there is only one proposition. However, the delimitation is not easy, as the relevant structural difference between E2 and E3 is minimal.

Reduction in verb series

The same problem of delimitation between a construction of two clauses and one clause arises in verb series. These undergo the two basic types of reduction processes, grammaticalization and lexicalization. By grammaticalization, one of the serial verbs reduces to a grammatical formative. This is illustrated by E4 from Middle Chinese (Li & Thompson 1976: 485):

At an earlier stage, ba ‘seize’ is a full verb taking a direct object and forming a regular verb series with the verb following this verb phrase. This construction is translated as #a. At a subsequent stage, ba is an accusative preposition which, together with its complement, constitutes the direct object to the following verb. This situation is rendered by translation #b. Depending on the stage of grammaticalization of this word, we deal with a complex sentence or a simple sentence.

By lexicalization, a series of verbs becomes one compound verb, with the usual symptoms of lexicalization. At the same time, the grammatical boundary between the two verbs disappears, and the result is one clause. E5 illustrates this result.

One of the symptoms of clause-union in E5 is the fact that there is only one tense marker.