The concept of nominalization

Nominalization s.l. is a grammatical or lexical operation that converts a non-nominal expression into a nominal one. (Derivatively, the term also designates the product of the operation.) For instance, the NP the shooting of the hunters is a nominalization of either the clause the hunters shoot or the clause the hunters are shot. Nominalization is a kind of category transposition or recategorization.

In a static perspective, a non-nominal and a corresponding nominal construction are just in a paradigmatic relationship. In a dynamic perspective, however, nominalization is a directed process, witness its name, which is ‘nominalization’ rather than ‘sententialization’. Such an intrinsic orientation is typical for reductive processes. It should at least be noted that the opposite perspective, viz. expanding a complex nominal into a clause, is useful in certain circumstances.

Since both of the properties ‘nominal’ and ‘non-nominal’ allow for variation, there are multiple ways of restricting this concept, with the result that rather different notions termed ‘nominalization’ are circulating.

Derivational vs. syntactic process

1. Both the input and the output expression may be stems (of lexemes). Nominalization is then a process of stem formation (“word formation”), more specifically, a derivational process. In this sense, e.g., the noun nominalization is nominalized from the verbal base nominalize. This variety may be called derivational or lexical nominalization.
2. The input, and possibly the output, too, may be syntactic constructions. Nominalization is then a paradigmatic syntactic operation (a “transformation”) converting a non-nominal syntactic construction – typically, a clause – into a nominal construction. The latter may be of varying degrees of complexity. In this sense, the transformation of .a into #b or into #c are examples of nominalization. This variety may be called syntactic or grammatical nominalization.

To the extent that the boundary between syntax and derivation is fluid, i.e., that derivation is syntactically regular, the two notions of nominalization may comprise the same phenomenon. In the case of the above exemplification, the formation of a nomen actionis with the suffix -tion may be described as a process of word formation or as a syntactic process. In either case, the change of verbal valency into nominal valency must be accounted for. The difference then boils down to whether this particular type of nominalization is completely regular and productive. (It is not, so this is more appropriately regarded as a process of derivation.)

Verbal vs. non-verbal bases

The input for nominalization, or the base of the derivational process, may be of any non-nominal category. Thus, at the level of derivational processes, destruction is a deverbal nominalization of destroy, sadness is a deadjectival nominalization of sad; outing is a deadverbial nominalization of out.

The nominalization of non-verbal bases typically does not involve much syntax. It is the nominalization of verbal and clausal bases that is mostly treated as a syntactic process.

Conversion and coercion

Both at the lexical and at the syntactic level, an overt operator may apply to the underlying non-nominal construction to signal its nominal category. This is, for instance, the case both in the syntactic nominalization of .a and in its lexical version #b, nominalized by the universal subordinator que and by the derivational suffix -ion, resp.

Things are different in . That walk in .a is a noun rather than a verb gets clear only by the fact that it is preceded by the indefinite article. The article is not a nominalizer; it also precedes expressions already categorized as nominals, and its function is determination. Likewise, the fact that the bracketed expression in #b is categorized as nominal derives from its function as the direct object of the matrix verb. A recategorization not signalled by any overt means is a conversion. A conversion takes place by coercion, i.e. by inserting the expression to be recategorized in a context which only allows members of the target category.

Not seldom, coercion is forced by some particular formative in the context which only combines with expressions of the target category. In the case of nominalization, definite determiners typically play this role. Such a formative may then grammaticalize into a nominalizer. A case in point is the English subordinator that appearing in the translation of .a. Originally, it preceded the subordinate clause as a cataphoric demonstrative. Now it is grammaticalized to a universal subordinator.

Oriented vs. non-oriented output

The concept of orientation is explained suo loco. Here it suffices to recall two relevant examples: administrator is an oriented nominalization, viz. one oriented towards the subject slot of the underlying verb, while administration is a non-oriented nominalization, as it designates the core, rather than any of the participants, of a situation in which somebody administrates something. In the following, the effect of orientation on the category of the product is used for subclassification:

1. If the nominalized clause is not oriented, it has the distribution of a proper noun or NP (cf. the paradigmatic relation between .a and #b). This category is traditionally called substantive clause. Since the middle of the 20th century, it has also been called complement clause. While the term ‘substantive clause’ appeals to the category of the construction, the term ‘complement clause’ appeals to its function, viz. as a clause governed by its head.
2. If the nominalized clause is oriented, it may have distributional properties either of a common noun or nominal or of an adjectival.
1. As explained in the section on orientation, if the nominalized clause codes a closed (i.e. complete) proposition, its orientation can only take place at the semantic level. Such a clause semantically oriented towards one of its components functions as a nominal in the matrix ().
2. If the nominalized clause is based on an open proposition, there are two possibilities:
1. An oriented nominalized clause that does not modify a lexical head is a free relative clause. Its category is ‘nominal’. It may constitute the core of an NP by itself (.a). If it combines with a nominal head, it bears an appositive relation to it (.b).
2. An oriented nominalized clause which modifies a lexical head is a plain relative clause. More specifically and in contrast with adjoined and circumnominal relative clauses, it is an adnominal relative clause. Its category is ‘adjectival’. Its relation to its head is then one of modification, more specifically, of attribution (translation of .b).

In other words, a non-oriented nominalization is of the category NP, while an oriented nominalization is of the category ‘nominal’ s.l. The latter may or may not combine with a nominal head. If it does, it may either bear an attributive or an appositive relation to the head.

The function of nominalization

The operation of nominalization is an essential component of both the functional domains of junction and of concept formation. This is because nominalization varies in strength as discussed in the section on reduction of clauses (cf. also the diagram below): On a low degree of nominalization, clausal properties predominate, with the result that the nominalized clause is integrated with the main clause by the mechanisms available in junction. On a high degree of nominalization, nominal properties predominate, with the result that the desentential noun participates in clause formation like primitive abstract nouns.

Reference to a particular situation is made by a finite clause. Nominalization makes a situation available in the nominal category. It typically involves desententialization and, in the end, non-finiteness: The situation is reduced to the situation core, the clause is reduced to a verbal noun. Those clause components which individuate a particular situation get gradually lost under nominalization. The effect is typicization: instead of a particular situation, a type of situation is being designated. This is useful in a variety of contexts:

• An anaphor is typically a hyperonym of its antecedent. The same goes for anaphoric reference to a situation: While the antecedent may be a full clause, the anaphor may be verbal noun ().
• An entity may be characterized by a type of situation in which it participates. This is done by an oriented nominalization ().
• Hypothetical situations are often not fully specified, but are rather types of situations. Thus, both .a and #b contain a noun formed by conversion of a verbal base (or by category indeterminacy). In #a, the situation designated is specified by one participant which is referential (the interlocutor), while the other participant (by default, the speaker) is inferred on pragmatic grounds. Moreover, the act of thanking presupposes a benefit that occurred in the past. As a result, this is a fully specified situation, and the expression designating it can therefore be used – as it is here – to refer to a particular situation. By contrast, the verbal noun in #b is not specified in any way. The matrix verb selects a situation in the future. As a result, this is just a situation type, and the expression has indefinite non-specific reference or, depending on theories, is not referential.

The relation between the structural operation of nominalization and its typicizing function may be visualized as follows:

This means: The more sentential a nominalization is, the more it is apt to individuate a situation; the more strongly it is desententialized, the more it designates a situation type.