What It Means to Verbalize: the Changing Discourse Functions of the English Gerund

What It Means to Verbalize: The Changing Discourse Functions of the English Gerund

Lauren Fonteyn12, Hendrik De Smet1,
and Liesbet Heyvaert1

Abstract

The English gerund system consists of two types of gerunds: a nominal gerund (the learning of a language), and a verbal gerund that developed out of the nominal gerund (learning a language). While the formal aspects of this diachronic verbalization of the gerund are well documented, much remains to be said about the discourse-functional side of the change. In this paper, it is argued that the formal verbalization of the gerund is accompanied by an important change in the discourse-functional organization of the gerund system. Based on functional characterizations of NP behavior in the literature, the prototypical behavior of complex NPs is operationalized as (i) functioning as manipulable discourse participants that are important enough in the following discourse to be susceptible to anaphoric targeting and (ii) being inaccessible to anaphoric targeting of internal participants. The results of an analysis ofa set of nominal gerunds, verbal gerunds and 'regular' complex NPs covering the period 1640–1914 (taken from the Penn Parsed Corpora of Early Modern and Modern British English) showthat the increasingly clause-like appearance of the verbal gerund is in fact accompanied by atypical NP behavior. Moreover, the paper makes clear that the changes in the discourse-functional organization of the gerund system did not only affect the verbal gerund, but also had some implications for the nominal gerund. These findings shed new light on the (diachronic) processes of verbalization and nominalization, and on what they mean on a discourse-functional level.

Keywords

historical linguistics, gerund, nominal gerund, verbal gerund, nominalization, verbalization, Early Modern English, Late Modern English, functional linguistics, history of English

Introduction

This paper discusses the development of nominal and verbal gerunds from Early Modern to Late Modern English with a focus on their discourse-functional behavior. The English gerund is a deverbal structure ending in -ing, which can take two different forms: nominal gerunds, as in (1), have the internal syntax of a noun phrase (NP), whereas verbal gerunds, as in (2), have the internal syntax of a clause.

(1) There is, first, the dryness inseparable from the learning of a language (…). (Bain, 1878, Penn Parsed Corpus of Modern British English [PPCMBE]1)

(2) I feel so relieved at having found a master. (Thring, 187X, PPCMBE)

At the same time, the nominal gerund in (1) and the verbal gerund in (2) can synchronically be defined as nominalizations, i.e., they serve to make a verbal form function as a noun phrase (or NP) in a clause and involve both decategorization and recategorization (Hopper & Thompson 1984; Bhat 1994; Malchukov 2006). Both nominal and verbal gerunds show signs of external reclassification in that they adopt—be it to a varying extent—the distribution of an NP (i.e., the grammatical function that an NP serves in a clause, e.g., subject, object, prepositional object, etc.). They are also subject to internal reclassification (McGregor 1997; Heyvaert 2003) or formal decategorization (Hopper & Thompson 1984), which manifests itself, for instance, in the absence of overtly realized morphosyntactic properties such as inflection for person or tense and in the integration of internal formal noun-like features and referential behavior in the discourse (Langacker 1991; Heyvaert 2003).

While synchronically, the verbal gerund, like the nominal gerund, is considered a nominalization, it is in fact the result of an instance of diachronic verbalization of the nominal gerund, which existed long before its verbal counterpart (Tabor & Traugott 1998; Malchukov 2004:119-121).2 Verbalization (and nominalization for that matter) in this diachronic sense is to be understood as a historical process in which a form gradually acquires verb-like (or, with nominalization, noun-like) characteristics, often combined with the loss of features of its original category (Malchukov 2004:119). The diachronic verbalization of the gerund has thus far mainly been defined in morphosyntactic terms, as earlier studies focused on the lengthy process involving the reconfiguration of the NP structure of the nominal gerund into that of a nonfinite clause (Jespersen 1940; Mustanoja 1960; Visser 1963–1973; Emonds 1973; Tajima 1985, 1996, 1999; Donner 1986; Jack 1988; Houston 1989; Van der Wurff 1993; Fanego 1996a, 1996b, 1998, 2004; Miller 2002; Kranich 2006). In that process, the gerund gradually acquired verbal properties, such as the ability to:

a.Govern an object or a predicative complement (e.g., She always hated his teasing the girls / my being an atheist);

b.Be modified by adverbs or adverbials which only co-occur with verbs (e.g., by gently stroking the cat);

c.Show tense and voice distinctions (e.g., the harm in having eaten the plastic wrapper / dreams of being chased);

d.Be negated by means of the VP-negating particle not (e.g., my father’s not caring); and

e.Take a subject in a case other than the genitive (e.g., I don’t like him being around mom).

According to the literature, the acquisition of these verbal features appears to have unfolded roughly as follows. The verbalization of the gerund started during the Late Middle English period, when the deverbal noun in -ing began taking adverbial modifiers of various kinds. Around 1300, the first gerunds with direct objects instead of of-phrases appeared, yet their occurrence was limited to (a restricted set of) prepositional contexts. The first instances of verbal gerunds with non-genitive subject “occurred sporadically from Late Middle English, but remained very rare for a long time afterwards” (Fanego 2004:9). In Early Modern English, the frequency of verbal gerunds received a boost, which was accompanied by a more gradual spread of the verbal gerund from prepositional to other grammatical contexts. Around the same time the verbal gerund also acquired the ability to express distinctions of voice and tense/aspect through the auxiliaries be and have (Tajima 1985:111-113, Fanego 1996a:129).

While the formal aspects of the verbalization of the gerund are well documented, much remains to be said about the discourse-functional side of the change. The view to be developed in this article is that the verbalization of the gerund was not merely a morphosyntactic process, but was also reflected on the discourse-functional level. Not only did the verbal gerund adopt the morphosyntactic features of a non-finite clause, it also increasingly adopted new functions in discourse. Previous discourse-functional analyses of the gerund mainly focused on the types of reference that the gerund can realize (e.g., specific vs. generic reference) (Schachter 1976; Heyvaert 2003, 2008; De Smet 2008). In contrast, taking our inspiration from functional approaches to the noun phrase (Du Bois 1980; Givón 1985; Hopper & Thompson 1984, 1985), we will operationalize the discourse-functional perspective in terms of “referent manipulability” and “internal participant targeting.” In the process, we show that, with respect to these functions, verbal gerunds are or have become less NP-like. This discourse-functional change can only be revealed if the verbal gerund is studied against the background of its interaction with the nominal gerund construction (cf., De Smet 2008), and against the background of its interaction with regular NPs. Therefore, our argument is based on a comparison of the discourse behavior of verbal gerunds, nominal gerunds, and regular NPs across two historical periods. One consequence of this approach is that apart from revealing functional change in verbal gerunds, our analysis shows that nominal gerunds underwent functional change too. In sum, then, we show (i) that the morphosyntactic verbalization of the gerund went hand-in-hand with a functional shift towards less NP-like behavior in discourse and (ii) that nominal gerunds appear to have initially anticipated this functional shift but later retracted again towards more noun-like behavior.

In what follows, we will first describe the particular discourse-functional perspective that is taken in this paper in more detail, developing the notions of referent manipulability and internal participant targeting. Next, we discuss the historical corpus data and the methodology used in our research and present the results of our analysis. Finally, we summarize our findings and discuss their implications for the notions of verbalization and nominalization.

Towards an Alternative Discourse-Functional Analysis of the Gerund

In previous synchronic (Schachter 1976; Heyvaert 2003, 2008) and diachronic (De Smet 2008) studies, the discourse-functional perspective on the gerund was operationalized by treating gerunds as NPs or discourse referents that somehow fit in with the nominal paradigm in terms of their distribution, or in terms of the types of reference they can express (e.g., specific vs. generic reference). In this paper, we have operationalized the discourse-functional analysis of gerunds differently, in thatwe analyze the degree to which they function as regular or prototypical NPs by looking at referent manipulability and internal participant targeting.

The starting point of this particular discourse-functional approach is that NPs and clauses differ in their typical discursive use. This is because, according to Langacker (2009:151), “the nominal and clausal realms pose different primary epistemic concerns.” NPs represent things or objects, which have the basic tendency to endure. The default expectation is that objects exist, will continue to exist, and simultaneously coexist with many other instances of the given type. For NPs, then, the primary issue is that of identification, signaling which one of the enduring coexisting (groups of) object entities or participants the speaker is referring to. In contrast, clauses profile an event. Events do not endure, but occur in a (typically) short time span and are generally transient and not time-stable. It follows that events do not develop a permanent identity and are not experienced as co-occurring. The primary epistemic concern of clauses is therefore not identification of the event or process, but existence or occurrence, i.e., whether the event in question occurs at all.

The differences in semantics and discourse function between prototypical NPs and clauses manifest themselves in how NPs and clauses contribute to text structure. The textual system of reference (Halliday & Hasan 1976; Lyons 1977; also called “identification” and “ideation” in Martin 1992) is concerned with creating textual coherence through tracking and referring to participants in discourse. In (3), the three cooking apples are anaphorically referred to by the NP the apples or the personal pronoun them.

(3) Wash and core three cooking apples.Put the apples/them into a fireproof dish. (Martin 1992:137)

Such a system logically requires permanent or time-stable entities. In other words, the “entry condition” for the reference or identification network is participant status (Martin 1992:128). Crucially, while all participants are realized through nominal groups, verbs profile a process, which is a non-participant meaning (Martin 1992:129, 138). It follows that verbs are normally not “trackable” or “targetable” through reference, suggesting that verbs typically do not act as participants, but instead report events that involve participants. This does not mean that it is impossible for clausally introduced entities expressing events to be tracked and referred to in discourse. Nonparticipant meanings such as processes can be construed as participant-like things by means of discourse semantics, and one can use it, as in (4), or this,as in (5), or that to refer to them.

(4) [The Queen said:] ‘Curtsey while you’re thinking what to say. It saves time.’ (Halliday & Hasan 1976:52)

(5) Skrine has got the Newdigate and Hamley a College scholarship. This is pleasant. (Thring, 187X, PPCMBE)

This phenomenon is described by Halliday and Hasan (1976:52-53, 66-67) under the heading of "extended reference" (see also Martin 1992:139); Francis (1994:84) speaks of “advance/retrospective labels”; and Takahashi (1997:63) uses the term “summation.” As the term “extended reference” indicates, this is not the default type of reference, and is much less common than reference to participants in their prototypical form, i.e., NPs.3

In the same vein, Hopper and Thompson (1984, 1985) argue that linguistic function in the discourse correlates with degrees of prototypicality in the formal marking of linguistic categories. Cross-linguistically, the extent to which an element is morphosyntactically treated as a noun appears to be strongly related to the degree to which it serves a noun-like function. An element is said to serve a noun-like function if it introduces a participant into the discourse that is going to be prominent in the subsequent discourse (Hopper & Thompson 1985:159; Du Bois 1980; Givón 1981) and has stable rather than processual properties (Cristófaro 2003). High or low categoriality (or the degree to which a form displays the prototypical behavior of its category) is thus associated with how likely it is that an element will be referred to or targeted in the following discourse. This is what we will term here “manipulability” (Hopper & Thompson 1985:160). Prototypical nominal referents, then, are not necessarily those that are semantically most object- or thing-like, but those that are manipulable. Hopper and Thompson (1985) summarize these ideas in the “Iconicity of Lexical Categories Principle” (henceforth ILCP):

(…) to the extent that a linguistic form is carrying out this cardinal function [i.e., to introduce manipulable participants in the discourse], it will be coded as a noun, and will manifest the full possible range of nominal trappings conventional in the language. (Hopper & Thompson 1985:159)

As an illustration of the ILCP, consider (6) and (7).

(6) Early in the chase the hounds started up an old red fox, and we hunted him all morning. (Hopper & Thompson 1984:709)

(7) We went fox-hunting in the Berkshires. (Hopper & Thompson 1984:709)

In (6), the full range of attributes characteristic of nouns is manifested: fox takes a determiner and adjectives, has the potential of being either singular or plural (old red foxes), and can be combined with a demonstrative (that old red fox). In (7), despite the fact that fox still signifies a (visible) thing or entity, it does not formally behave as a prototypical noun at all, as it cannot pluralize or take modifiers or determiners. The reason for this, Hopper and Thompson (1984:708) argue, is that fox in (7) does not “play a role in the discourse in which it figures,” whereas an old red fox does.

For verbs, on the other hand, the degree of prototypical verbhood is determined by the extent to which a form “asserts the occurrence of an event of the discourse” (Hopper & Thompson 1984:708). Compare (8) and (9).

(8) After the break, McTravish threw the log. (Hopper & Thompson 1984:709)

(9) To throw a log that size takes a great deal of strength. (Hopper & Thompson 1984:709)

In (8), threw is used in a discourse context which selects a prototypical instance of the verb throw and can freely appear in a variety of tenses or take agreement morphemes showing concord with its subject. This is impossible in (9). Hopper and Thompson (1984:710) argue that in this less prototypical use of the verb in (9) and the noun in (7), “the overt grammatical contrast between N and V is reduced or even cancelled,” as there are very few overt morphological differences between fox and throw. In the prototypical instances (6) and (8), by contrast, the morphological features differ maximally.

It may be worthwhile relating the ILCP to the discussion of the English gerund. Both nominal and verbal gerunds are nominalizations in the synchronic sense, in that they both preserve the verb-like event semantics, but distributionally occur in NP slots. The nominal gerund also formally resembles prototypical NPs in that it realizes its arguments by licensing ‘s and of-genitive NPs (Huddleston & Pullum 2002:658) and lacks the ability to express verb/clause-like notions such as tense and mood. The verbal gerund, on the other hand, is far more clause-like than its nominal counterpart in terms of internal structure.

What may then be hypothesized about the behavior of verbal gerunds in discourse? On the one hand, we hypothesize that the internal formal differences in the prototypical NP features that nominal and verbal gerunds display are reflected in differences in the degree of manipulability in the discourse. If the ILCP holds, the formal coding of the two gerund types should be linked to their discursive behavior, with nominal gerunds being more likely to function as manipulable discourse referents open to anaphoric targeting, much like ordinary complex NPs but unlike verbal gerunds. This paper will therefore examine how frequently gerunds are picked up as a participant in the following discourse, whether the two gerund types differ from each other or from regular complex NPs in this respect, and whether these differences increase over time.

Anaphoric targeting of the gerund is illustrated in the following examples. As the examples show, anaphoric targeting can assume various forms, such as full repetition (10), a definite NP (11), a demonstrative pronoun (12), or the personal pronoun it (13).

(10) Robert is as incapable of doing a foolish thing as he is of doing a wrong thing.

–“Nobody is incapable of doing a foolish thing.” (Wilde, 1895, PPCMBE)

(11) Many firms make a specialty of laying out gardens artistically and naturally, and although some amateurs try their untrained hands at the business, they generally have to call in the aid of the man who knows his plants and their nature and uses by everyday intercourse and experience. (Weathers, 1913, PPCMBE)

(12) So that we cannot discourse of the man’s right, without describing the measures of his duty; that therefore follows next. (Jeremy Taylor, 1762, Helsinki Component of the Penn-Helsinki Parsed Corpus of Early Modern English [HC])

(13) (…) a full-rigged barque or schooner will cast a great shadow over the window where I fain would be, and the talking of the sailors on the deck be plainly audible. This is not, however, to assume it will be understandable. (Bradley, 1905, PPCMBE)

On the other hand, if verbal gerunds have become less manipulable than nominal gerunds and regular NPs, it can be expected that their internal participants have become more open to anaphoric targeting, with the ing-form in the verbal gerund functioning as a relator between manipulable participants. This fits with Langacker’s (1987) claim that relational expressions (a category that includes verbs, but not nouns) are conceptually dependent on the entities they relate, and therefore not only profile a relation but also the entities related (Langacker 1987:215). Notice then that, contrary to Hopper and Thompson (1984), we do not assume that non-finite clauses completely lose verbal functionality. Even though non-finite clauses do not independently assert the occurrence of an event (cf., example 9 above), they retain the verbal function of linking and profiling discourse participants. Indeed, in English verbal gerunds this is iconically reflected in verb-like participant coding. We expect then that verbalization in the gerund will correlate with increased openness to internal participant targeting.