6. Socio-historical analysis of science in the modern period: even if this is not one of the main objectives and it was originally conceived as a way of complementing
2.1. Definition of nominalization
2.1.1. The concept of transference
Transference has been a key idea in the way scholars have understood nominalizations until very recently. The first insightful remarks on nominalizations in modern linguistics dealt with the idea that nominalizations are transformations of a verb into a noun
(Chomsky, 1970; Lees, 1960;) After an initial exclusive concern on the form of nominalizations, scholarly attention shifted to functions in texts in the 1980s.
“Grammatical metaphor” was the expression used by functionalists (Halliday, 1985) to study nominalizations, as well as other linguistic features. The idea of transference, however, remained there, as all kinds of grammatical metaphors imply the transference of meaning of one unit into the form of another –in the case of nominalizations, the meaning of process usually codified in verbs is transferred into the shape of a noun. In the following pages, the concept of transference, as described by Transformational-generative grammar (TGG henceforth) and Systemic Functional Linguistics (SFL henceforth) will be briefly analyzed.
TGG theory is characterized by the assumption that a universal grammar categorizes language. The scope of study of generativism hence is not concerned with performance, but with competence and language is conceived to made up of two basic components (Chomsky, 1957, 1965, 1970). In the lexicon we can find the semantic information of the words existing in a language, as well as the phonetic and graphic representation of the realizations of those words. Grammar, on the other hand, deals with the relations between the words in the lexicon and it is composed of a base structure and a series of transformational rules that turn the base structure into other types of sentential or phrasal configurations. Nominalizations are indeed considered a transformation that turns a VP into a nominal form (Grefenstette & Teufel, 1998) and scholars have consistently tried to provide theories about it (Chomsky, 1970; Grimshaw, 1990; Hazout, 1995; Jackendoff ,1975; Lees, 1960; Newmeyer, 1971; Siloni, 1997;
Zucchi, 1993).
In the nominalization process, the verb is ejected from its syntactic role into a nominal position. Thus, in (3)
(3) The enemy destroyed the city.
a full sentence made up of a verb (destroyed), its subject (the enemy) and its direct object (the city) is turned into a NP in (4)
(4) The enemy’s destruction of the city.
where through the addition of a suffix, the verb becomes a noun (destruction) and is flanked by its former valencies, now turned into modifiers (the enemy’s and of the city).
This nominal form can be fulfilled by either nominal form of the verb (a deverbal noun) as in (5)
(5) The destruction of the city surprised the enemy.
or a gerundive as in (6)
(6) The destroying of the city surprised the enemy.
Concerning motivations for the use of nominalizations, generativists usually point out stylistic reasons like avoiding repetitions and awkward uses of verbs (Chomsky, 1970).
According to Banks (2005a, p. 78), nominalization can be described as a “form of grammatical metaphor whereby a process, which could be encoded as a verb, is encoded non-congruently as a noun.” Other authors have produced similar definitions within the
SFL tradition (Albentosa, 1997; Guillén, 1998; Halliday, 1985, 2004; Ravelli, 1988;
Ventola, 1996). Several terms, namely grammatical metaphor and non-congruent realization, are keywords in the understanding of the definition suggested by Banks (2005a).
The first concept, grammatical metaphor, recalls on the basics of the functionalist approach. Guillén (1998, p. 368) defined grammatical metaphor as “the transference of the linguistic representation of the semantic components of a situation between different lexicogrammatical categories”. It may be noted here that embedded in the definition we find again the word transference, which echoes the concept of transformation studied by generativists. Again, nominalizations are believed to originate after some form of modification from a pure, more basic structure of language. In this case the tension is not produced between different lexicographic realizations but rather situated between semantics and lexis. Ravelli’s definition (1988, pp. 135-136) categorized grammatical metaphor as “an alternative lexicogrammatical realization of a choice in the semantics.”
From these definitions it can be inferred that there are some prototypical realizations of the semantic components in terms of lexical categories and that grammatical metaphors disrupt this prototypical configuration and assign new linguistic realizations to semantic components.
The study of congruency, that is determining whether the function of a word is being congruently realized by its form, is one of the principles of SFL. Halliday (1985) acknowledged that there is a tendency to establish a one-to-one correspondence between the form and the function25 of a word, which results in the endowment of meaning to the
25 It may be noted here that in SFL function is associated with the semantic properties of a word, that is, with the semantical function that the linguistic unit has in the sentence. It must not be confused with the syntactical function of the word in the sentence.
form of a word and the establishment of a closed system of what he calls “congruent realizations of functions”. These are typical, more natural ways of saying things and encoding functions into particular word class realizations. These more natural, congruent encodings are listed in table 1.
Table 1: Congruent functional realizations.
Clearly table 1 does not give a full account of all the possible matchings between word-class and function available in English. Those cases in which congruency is disrupted –for instance in a nominalization, where a process is encoded in a noun and not in a verb– are called congruent or metaphorical realizations. Despite non-congruent forms may be more frequent or even function as the norm, Halliday (1985) proclaimed the superiority of congruent realizations. Indeed, grammatical metaphors are widely used in adult speech and in some cases they have even become the default form.
Other SFL scholars have produced similar theories about congruency. Thus, in her article about the packing of information in scientific discourse, Ventola (1996)
established three basic ways in which function types can be encoded, namely the simple-congruent, the complex-congruent and the incongruent or metaphorical codings.
The most basic way of packing information is using the simple, congruent coding, which relies on the congruent realizations shown in table 1. The complex, congruent coding increases the amount of information included in the sentence without altering its functional structure. The congruent nature of the unit is kept even though this does not imply simplicity in the understanding of the content. In nominal groups, this complexity is achieved by the addition of extensive pre- and postmodifiers or by means of rankshifted postmodifiers, whereas in the case of verbal groups, auxiliaries are used to increase its density. Finally, the metaphorical coding to which Ventola (1996, p. 182) referred is what “in systemic-functional literature [has] been called grammatical metaphor.”
The relationship between the semantic component and grammatical categories concerning grammatical metaphor and the disruption of standard formulas was analyzed by Ravelli (1988). He set eighteen different categories of grammatical metaphor, which can be summarized in nine basic groups. These categories cover not only instances of nominalization –the first category in his classification– but practically all grammatical categories. Examples include cases like for that reason (type 5), in which a prepositional phrase functions as a logical connection –a function usually fulfilled by nominal or verbal groups–, or historical experience (type 9), in which an adjective replaces a PP and fulfills a circumstantial function. This classification also includes examples of rankshift (type 8) in which clauses function as phrases within a bigger clause, as in (7)
(7) [All it can do] is [to retaliate].
The existence of such a number of different kinds of grammatical metaphor does not minimize the importance of the process of nominalization itself, because nominalization is, indeed, “the type of metaphor of which there is the greatest awareness” (Ravelli, 1988, p. 140). The common feature for all the different kinds of nominalization can be said to be the convergence of the function of process and the grammatical form of noun. Thus, nominalization could be defined as a structure “in which verbal processes are coded in nominal structures” (Ventola, 1996, p. 153).
Prototypically, the unmarked function of nouns is to express an entity or thing and that of verbs is to express a process. In nominalizations, however, the function of expressing a process is realized by a nominal group (Banks, 2005b). The process of nominalization could be systematized in the following way:
Figure 2: Prototypical/unmarked functions of nominalizations.
The merging of function and form results in one of the most characteristic features of this process. In nominalizations the semantic component of verbs, that of expressing a process, is encoded into a nominal group. In other words, nominalizations retain the semantic component of the process but they present it as an entity. Banks (2005b) compared grammatical metaphor to semantic metaphor. According to him, both grammatical and semantic metaphors are similar processes in the sense that both intend to achieve some effect on the discourse by the alteration of either the function or the meaning of a particular word. Thus, whereas semantic metaphor keeps the form of the word but alters its meaning, in grammatical metaphor the meaning is retained –with subtle modifications– but the form is changed.
However neither TGG nor SFL of approaches to nominalization were complete.
The first focused exclusively on internal structure whereas the latter ignored structure and only studied functional implications. Dik (1985) and Mackenzie (1985) were very innovative in trying to join both approaches. Thus they conjoined functional readings with the study of the structure of nominalizations. Mackenzie (1985, p. 32) remarked that the study of nominalizations was centered on the relationship of the internal structure of the NP and its sentential equivalents, which in his view did not reflect the real use and structure of most nominalizations. Consequently he centered his study on the process of valency reduction. According to him, there are three main reasons why a speaker may have reduced the number of valencies by choosing a nominalization instead of a sentence: a) the avoidance of redundancy with the purpose of observing Grice’s Maxim of Quantity (1975); b) the achievement of syntactic versatility and c) the introduction of discourse referents. After Mackenzie’s turning, many scholars produced new theories about nominalizations which also took into account the premises of the
Optimality Theory (Prince & Smolensky, 2004) Although differing in minor concerns, nominalizations were believed to be the result of a conflict between lexical and functional constraints. In transcategorization processes, two independent processes are involved: on the one hand, the decategorization process (Hopper & Thompson 1980) implies the loss of morphosyntactic properties associated with the original primary function of the word (Malchukov, 2006, p. 974). On the other hand, the recategorization process assigns to an item some of the extended properties of the categories to which it functionally belongs (Bhat, 1994). Concerning the application of this process to nominalization, Malchukov (2006, pp. 1001-1002) claimed that
Morphosyntactic properties of nominalizations arise from the interaction of constraints some of which are functional, some structural. On the functional side transcategorial processes are constrained by hierarchy constraints on deverbalization and substantivization.
As a result of the addition of a suffix, nominalizations lose verbal valencies in the deverbalization process (Givón, 1995) and the degree of transitivity of the VG is reduced (Albentosa & Moya, 2000). Paired with this process, the substantivization process assigns new functional properties to nominalizations (Fowler, 1991). All the studies about valency reduction and substantivization together with Mackenzie’s allusion to Grice’s maxim of quantity, Mackenzie (1985) made way to the understanding of nominalizations as linguistic expressions per se and not as transformations of verbs. It must be pointed out that today many scholars do not agree with this vision but its acceptation is becoming a standard. In the following section,
attention will be directed from the relationship between nominalizations and VPs to the actual morphosyntax of this linguistic feature.
2.2. Theoretical considerations about the morphosyntax of nominalizations
One of the main difficulties when it came to outlining this study was the lack of unanimity on the main point of study concerning nominalizations. Thus, whereas TGG has provided massive bibliography on the morphosyntax of nominalizations and has not been concerned with the functions the motivations in the use of nominalizations, SFL scholars have focused exclusively on functions and have given very vague structural information about nominalizations and their typology. This section is concerned with the morphosyntax of nominalizations and, consequently, many of the TGG readings on nominalizations are here analyzed. They include the difference in morphology between gerundives and derived nominals as well as the differentiation between event and result nominals. This morphological study does not only include the study of the noun by itself but also of the relationship of the phrase it governs and the VP (section 2.2.2).
2.2.1. Theoretical considerations about the morphology of nominalizations
Chomsky (1970) attempted to categorize nominalizations according to their form and their semantic and syntactic properties. He found that there were two main types of nominalizations and called them gerundive, that is -ing nominalizations, and derived nominals, which are the rest of nominalizations, not formed by -ing suffixation. Thus, all the sentences in (8)
(8) a. John is eager to please.
b. John has refused the offer.
c. John criticized the book.
can be turned into both gerundives as in (9)
(9) a. John’s being eager to please.
b. John’s refusing the offer.
c. John’s criticizing the book.
and derived nominals as in (10)
(10) a. John’s eagerness to please.
b. John’s refusal of the offer.
c. John’s criticism of the book.
Chomsky (1970, p. 30) described also their differences according to their behavior within the clause. Thus, he claimed that there are three main differences between gerundive and derived nominals:
1. Unlike gerundive, deverbal nominals are not very productive. In this context, productivity is taken as the propriety of being systematically applied to a wide number of structures. This is explained by the fact that not all verbs have a deverbal noun, while they do have an inflected –ing form that can function as the head of an NP.
2. The relationship between the nominal and its preposition is, in the case of derived nominals, idiosyncratic, whereas in the case of gerundives the preposition used is always that of the root verb.
3. Deverbal nominals have the prototypical internal structure of NPs. Gerundives apply some structural restrictions on their phrase. As a consequence, determiners and adjectives are not always admitted, as can be seen in the examples (11) and (12) proposed by Siloni (1997, p. 5)
(11) a. John’s criticism of the book impressed us.
b. John’s sharp criticism of the book impressed us.
c. The criticism of the book impressed us.
(12) a. John’s constructing sailing boats impressed us.
b. * John’s rapid constructing sailing boats impressed us.
c. * The constructing sailing boats impressed us.
Quirk et al. (1985) also distinguished between two types of nominalizations:
verbal and deverbal. Verbal nouns are formed by the addition of the -ing inflection.
They represent a description of the action taking place and are usually followed by support verbs because they still retain much of the semantic value of the verb they come from. Deverbal nouns, however, describe an action that has already taken place. Their
relation with the root verb is more unpredictable, either from a semantic, syntactic or morphological point of view26.
Concerning the origin of nominalizations, TGG provided two different hypotheses: on the one hand, the lexicalist approach tried to demonstrate that nominalizations are the result of a movement that turns a verb into a noun.
Nominalizations are, therefore, real nouns and even if their origin is verbal there is no reason to consider them transformations of verbs. Indeed, derived nominals are considered nouns in deep structure, not deep-structure transformations. The differences between gerundive and deverbal nominalizations lead Chomsky (1970) to adopt a lexicalist position and consider them two different types of nominalizations.
On the other hand, the transformationalist approach claimed that nominalizations are the result of a transformational movement that turns verbs into nouns.
Acknowledging this transformation it is therefore possible to maintain the structural simplicity of languages. In transformationalist grammar, all possible irregularities and diversions from a universal simple grammar common for all human languages are systematically enunciated in a series of extensions and movements, among which nominalization has been widely analyzed. After Lees (1960), Newmeyer (1971) refuted Chomky’s alliance to the lexicalist approach signaling that there was no sufficient evidence to prove that derived nominals could not be derived by a transformationalist rule in the same way that gerundives were. In spite of Newmeyer’s remarks, the tendency in the following years was to progressively ascribe to the lexicalist approach
26 Indeed, Grefenstette and Teufel (1998) recognized that deverbal nouns tend to establish metonymic associations and become concretized over time, consequently losing their original semantic value. To solve this, they decided to concentrate on true nominalizations, that is, nominalizations that retain their syntactic parallelism with regard to the original verb so that they could develop a method for automatically identifying support verbs for verbal nominalizations.
because, even if it was not so elegant, it could better explain the semantic differences between the two kinds of nominals.
Taking the semantic domain as the starting point, Grimshaw (1990) distinguished between two types of nouns that can form nominalizations: “result nouns”, which express the product of an event and are usually associated with deverbal nominals and
“event nouns”, which denote an event or name a process and in many cases are found in gerundives. As Siloni (1997, p. 5) put it: “event nominals are the output of a process of lexical nominalization and gerunds are the product of syntactic nominalization”. The morphological form of the noun does not seem relevant as there are examples in which the suffix –ing does not refer to an action that is taking place, but to the result of that action. Hence, in (13) and (14)
(13) The destruction of the city surprised the enemy.
(14) The destroying of the city surprised the enemy.
the change of suffix clearly marks the difference between result (destruction) and action itself (destroying). However, in (15) and (16)
(15) The painting was sold.
(16) The painting of the house turned out to be a chaos.
the same word (painting) expresses the meaning of both action and result and has to be disambiguated with the help of the context. The distinction between event and result nominalizations has attracted the attention of numerous TGG scholars. As Peris27 (2012, p. 22) remarked, the two main points of discrepancy for the distinction of event and
27 In an attempt to organize and settle the debate, Peris (2012, p. 32) established up to twelve criteria to distinguish result and event nominals. These include information about the verb (1), modifiers (3, 4) and former verbal valencies (5, 6, 7, 8), among others. Her approach is interesting but it falls out of the scope of this study.
result nominals revolve around argument capacity and lexical representation of nominalizations. Regarding argument capacity Grimshaw (1990) showed that the relationship between the nominal and its preposition (Chomsky’s (1970) second claim) is explained thanks to the opposition event/result nouns. Grimshaw’s claims were backed up by Zubizarreta (1987) and Siloni (1997, p. 3), who explained:
Event nouns obligatorily have an argument structure as part of their lexical representations; they assign specific θ-roles, just like verbs. The lexical representation of result nominals, which do not express an event, does not specify an argument structure; result nouns do not take real arguments, which bear specific θ-roles, but rather a kind of semantic participants that are more loosely associated with them.
Some scholars (Alexiadou, 2001; Mel'čuk, Arbatchewsky-Jumaire, Elnitsky &
Iordanskaja, 1984; Picallo, 1999; Pustejovsky, 1995) disagreed with this view and claimed that both event and result nominals can indeed incorporate part of the verbal argument structure into their phrases.
Concerning lexical representation, scholars debated whether nominalizations should be considered one or two lexical entries. One possible way of representation is to
Concerning lexical representation, scholars debated whether nominalizations should be considered one or two lexical entries. One possible way of representation is to