Aspectual Grammar and Past Time Reference

Laura A. Michaelis

• Publisher: Routledge
• Year: 2002
• eBook ISBN: 9781134730063
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

This study presents a semantic framework for analysing all aspectual constructions in terms of the event state distinction, and describes the grammatical expression of aspectual meaning in terms of a theory of grammatical constructions. In this theory, grammatical constructions, like words, are conventionalized form-meaning pairs, which are best described not only with respect to their intrinsic semantic values, but also with respect to the functional oppositions in which they participate.

1: ASPECTUAL MEANING

1.1 TOWARD A THEORY OF ASPECT

The study of aspect, and of English aspect in particular, requires one to confront a number of problems of categorization. Since aspectual distinctions are not part of verbal inflection in languages like English, it is not clear that English can be compared to those languages in which aspect receives reliable coding in the verbal syntagm. Zandvoort in fact doubts whether aspect is relevant at all for the analysis of English:

The attempt to transfer the category of “aspect” from Slavonic to Germanic, and from there to Modern English grammar, strikes one as an instance of misdirected ingenuity.

(Zandvoort 1962:19)

At the very least, the relevant aspectual categories do not appear commensurate across languages. For example, some analysts have identified the French imperfective with the English progressive. It is evident, however, that the two categories are not identical: one can say Il avait de l’argent (“He had (imperf.) money”) but not *He was having money. In describing English aspect, therefore, it may be difficult to avoid inappropriate analogies. In addition, it may be difficult to analyze aspect in any language as a unitary phenomenon, like modal expressions, exponents of aspectual distinctions are drawn from many areas of the grammar. Aspectual meaning is expressed at the level of lexical-verb semantics, adverbial semantics, verbal morphology, and argument structure. How then can one describe aspect as a component of grammar?

It becomes especially difficult to maintain a distinct aspectual component in light of the fact that tense markers may express notions otherwise identified with aspectual categories. Past-time reference and perfective aspect are difficult to disentangle on logical grounds, since the only completed events are those located in the past, and on formal grounds, since analysts focusing on English are forced to regard simple past-tense predications (e.g., I went) as instances of perfective aspect. Even within the aspectual component, should we succeed in circumscribing it, we find areas of semantic overlap. For example, how can one distinguish in English between the perfect form (I have gone) and perfective aspect (I went)? Both forms seem to be in the business of expressing completion, and Frawley (1992) views the English perfect form as the English exemplar of perfective aspect.

Some authors seem to have abandoned the goal of devising a constrained theory which addresses the aforementioned problems. Frawley (1992) presents an aspectual classification containing six overlapping “major” categories (e.g., telic-atelic and imperfective-perfective) as well as five “minor” categories, e.g., inceptive and terminative. Such a classification can hardly be more than a catalog of grammatical distinctions; it does not appear to represent a part of the conceptual system of speakers.

In this Chapter, I will outline a semantic and functionally based model which provides a straightforward and cognitively plausible account of English aspectual phenomena. This model presumes that all languages contain: (a) the conceptual categories event and state, whether or not these are overtly distinguished, and (b) grammatical mechanisms by which a speaker can override the canonical representation of a situation as event or state. In the next section (1.2), I will discuss the semantic basis of the event-state distinction. In that section, we will explore a three-tier model of aspect in which viewpoint and phasal aspects function to override situation aspect, thereby providing presentational flexibility to narrators. In section 1.3, I will examine formal-semantic tools that can be used to represent aspectual distinctions. In section 1.4, I will show how the three-tier model of aspect works with respect to English.

1.2 THE ASPECTUAL CONSTRUCT AS A GRAMMATICAL AND CONCEPTUAL CATEGORY

The aspectual categorization of a situation is not necessarily encoded grammatically.¹ Therefore, one cannot always regard aspectual meaning as the semantic value of “aspectual marking.”² To illustrate this problem, let us examine the manner in which two typologically disparate Indo-European languages, Latin and English, manifest a fundamental aspectual distinction: the imperfective/perfective contrast. As a preamble to this illustration, we will first examine the conceptual basis of the contrast.

1.2.1 The semantic basis of the event-state distinction

According to Langacker (1987:258), the distinction at issue has a “primal character,” because it is linked to a basic cognitive capacity: the ability to perceive change (or the lack of change) over time. In general, as Langacker notes (1987:261), “the covariant properties of change and bounding can be regarded as two sides of the same coin (as can their opposites, namely constancy and open-endedness.” Imperfectively described situations (also known as states) obtain thoughout the interval at issue, possibly overflowing the boundaries of that interval. Perfectively described situations (i.e., events) are bounded insofar as they terminate within the relevant interval. To report the occurrence of an event is to report its cessation. An event predication denotes a change. Change implies the presence of distinct phases of the situation in question. The final phase of an event may be equated with a goal state, upshot, or outcome.

Events having goal states, like reaching the summit or reading a book, are typically termed telic situations (see Dahl 1981). Telicity and the presence of an episodic construal have often been equated. As Langacker notes, however, “some processes are internally homogeneous but construed as occurring in limited episodes (e.g., sleep, wear a sweater, walk)” (ibid.). Such processes have been termed activity verbs (Vendler 1967, Dowty 1979, 1986). Activity verbs partake of semantic properties associated with imperfectivity and properties linked to perfectivity. Like states, activities lack any intrinsic endpoint. Activities also share with states the properties of cumulativity and distributivity. The cumulativity property involves the possibility of conjoining two phases of a given situation: such conjunction will yield an identical situation. For example, two temporally contiguous instances of walking activity, taken together, count as an instance of walking. Of course, as C. Fillmore notes (p.c.), it is difficult to think of situations as being “conjoined” without invoking an explicitly linguistic operation, A more accurate description of the cumulativity property, which acknowledges the centrality of speaker inference to aspectual categorization, is the following: cumulativity is that property of situations which allows a conceptualizer who knows of two or more temporally contiguous instances of a given situation to conclude that these contiguous instances can be presented as one situation. Thus, if a speaker is asked to tell what happened during a certain phase characterized by two contiguous periods of giggling, the speaker can respond by saying He giggled. However, this same speaker, when asked what happened during a period characterized by two temporally contiguous instances of falling down, would not respond by saying She fell down.

The distributivity property, termed internal homogeneity by Langacker, can be described as follows: every temporal subpart or “component moment” of the state is an instance of that same state. In the case of activities, however, one typically finds only “weak distributivity” (see Herweg 1991b): one cannot reasonably infer, for example, that running is taking place at all subintervals of the overall interval in which that activity goes on. Running can be said to entail successive leaping motions, involving an alternating “trail leg.” One leap does not count as an instance of running.

Like events, activities are dynamic. The property of dynamism arises from one of two factors: (a) the presence of change over time or (b) the salience of terminal points. Some activities can be regarded as unfolding over time (e.g., a “cycle” of running involves successive hopping motions). Such activities therefore share with events the property of having distinct subphases. We will call these heterogeneous activities. Other activities, like wearing a sweater, are internally homogeneous, but involve stasis of limited duration. Such limitations are typically culturally imposed. For example, a given article of clothing is worn during waking hours only. As Langacker observes (1987:262), “the onset and termination of an otherwise homogeneous process can be interpreted as representing a type of change.” The type of change involved here is not that characterizable as a course of evolution through time (as in the case of a heterogeneous activity like breathing, which involves, minimally, successive events of inhalation and exhalation), but as a pair of momentaneous transitions from one state (e.g., that in which situation A obtains) to another (that in which situation A does not obtain). As we will note below, the dynamism which characterizes both activities and “events proper” affects the feasibility of present-tense reporting upon such situations.

States differs from events in lacking distinct subphases. States have the properties of (strict) cumulativity and distributivity (Herweg 1991a, 1991b). The distributivity property arises from the atemporal nature of states. As Bach argues:

The atemporality of states…can be brought out in the following Gedankenexperiment: Imagine a possible history (world) with only one time (which in a sense amounts to having no time); it is possible to think of various states that might obtain in such a world, but impossible to imagine events and processes that occur or go on in such a world.

(Bach 1986:71)

Thus, while a state can be said to obtain at a single instant, an event “takes time” in the sense outlined above. The atemporal quality of states finds an analog in the spatial dimension: as noted by Bach, among others, it is typically impossible to specify the location within which a given state obtained. While it may make sense to specify the general location of a state that happens to be subject to the influence of environment (Mary was happy in Cleveland), it is difficult to imagine circumstances under which a person could plant a landmark at or draw boundaries around the precise location of that state. Just as states cannot be said to “take time,” so they cannot be said to occupy any expanse of space. Events, by contrast, may take up both time and space—a situation that is readily represented by means of the two-dimensional system of spatio-temporal coordinates described by Cooper (1986). In Cooper’s system, events, like one dog’s chasing another across a lawn, are resolved into a series of successive states. Each state represents a point in space and time, where space—time location is computed as degree of displacement from some spatial and temporal origin (while “displacement” along the time line is unidirectional, displacement along the spatial axis is multidirectional). A collection of specific space-time locations constitutes the spatio-temporal “envelope” within which the event took place.

While the properties of cumulativity and distributivity are commonly treated as “physical attributes” of states, this mode of description obscures the fact that aspectual categorization is a product of speaker construal. As Herweg notes (1991b:977): “the same factual situation is often open to alternative conceptual categorizations…. So the fundamental aspects perfective and imperfective are linguistic devices that express the special perspective on a situation taken by the speaker.”

The properties of cumulativity and distributivity are best regarded as modes of inference available to the interpreter of a text. Thus, for example, the property of cumulativity enables the interpeter who encounters a stative assertion like Betty was ill yesterday to imagine a larger interval, subsuming the time period under discussion, which is characterized by the presence of that same state. The claim that aspectual constructs license certain modes of inference coheres with a recent trend in aspectology (manifested in both truth-conditional and functionally based theories) toward examining aspect within the framework of text construction and interpretation (Dowty 1986, Hopper 1979, Kamp and Rohrer 1983, Partee 1984, Polanyi 1985). Theories of this type examine the manner in which tense and aspect are used to encode the temporal interrelations among situations presented in a narrative (see section 1.2.2.4 ). Such relations link up “a series of successive instants in the narrative world which correspond to the moving reference point in the narrative construction of that world” (Polanyi 1985:10).

According to narrative-based aspectual theories, the categories event and state are not (as Vendler seems to have assumed) rigidly defined over particular classes of lexical predicates (i.e., verbs), or even over situation types—the latter of which we represent as tenseless propositions like Harry fry a fish. The demands of creating a coherent temporal discourse will often induce the narrator to override the “inherent categorization” (equivalently, Aktionsart) of a situation, e.g., to impose a stative construal upon an event predication.³

The aspectual means by which overrides are signaled will differ from language to language (see Smith 1986). These differences derive from the manner in which a given language encodes default aspectual categorization; they will be discussed in section 1.2.2.4. An examination of the means by which default aspectual categorization is grammatically signaled will return us to the question raised at the beginning of this section: is the perfectiveimperfective division properly regarded as a grammatical distinction? As noted, some languages do not manifest the distinction directly, i.e., certain languages contain no means of indicating within the verbal morphosyntax whether the situation denoted by a clause constitutes an event or a state. One such language is English, which contrasts in this regard with Latin. Note the Latin examples in (1.1–1.2):

(1.1) Statim invenit auxilium. immediately find:3sg:perf:act:ind help:A “He found help immediately.”

(1.2) Se Catilina credebat himself:A Catiline:N believe:3sg:imperf:act:ind posse urbem incendere be-able:pres:act:inf city:A burn:pres:act:inf “Catiline believed that he could set fire to the city.”

The situations presented in (1.1–1.2) are encoded in a manner which reflects their default aspectual characterizations. The speaker determines this characterization by assimilating the conceived or perceived situation to an ontology of idealized situation types (see Smith 1986). Since “finding” is an event (an achievement in the Dowty—Vendler categorization), the verb invenio (“I find”) in (1.1) receives perfective coding, i.e., the stem form and person—number ending characteristic of perfective aspect. As a stative predicate, the verb credo (“I believe”) in (1.2) receives imperfective coding. The English analogs of the Latin verb forms are inflected only for tense.

1.2.2 Grammatical and inferential reflexes of the event—state distinction

While English lacks any overt means of encoding the event—state distinction, there is evidence that perfective and imperfective aspects constitute covert grammatical categories in English. This evidence involves both grammatical phenomena and the licensing of textual inference. Grammatical ramifications of the event—state distinction arise from the distinct semantic properties attributed to events and states above. Thus, for example, event predications but not state predications are compatible with the so-called pseudo-cleft construction (Parsons 1990). The sentence What Harry did was run is acceptable, but the sentence *What Harry did was hate cats is deviant. Such deviance can be attributed to the fact that events, unlike states, are classifiable as deeds or actions; the occurrence of an event entails an output of energy.⁴

Given the semantic basis of grammaticality contrasts like that described above with respect to the pseudo-cleft construction, we might expect that certain grammatical traits will constitute cross-linguistically valid criteria for membership within a given aspectual class. There are, however, certain difficulties inherent in the use of grammatical criteria as universal tests for eventhood or statehood. Foremost among these difficulties is the fact that there is no cross-linguistic semantic parity among language-particular instances of a grammatical construction that might, in a particular language, prove useful as a sorting heuristic. For example, the compatibility of perfective aspect with present-tense inflection will depend upon the semantic range of the present tense in the language at issue. English differs from languages like French in opposing the simple present to a progressive present. In English, present-tense event predications, if intended as reports upon circumstances ongoing at present, must appear in the present progressive.⁵ Thus, a sentence like He falls, while having the potential for a habitual interpretation, is anomalous if interpreted as a report about the present state of things: the moment of speech cannot accommodate the extended temporal profile of the event. The present-progressive sentence He is falling is, however, acceptable. The progressive (as will be argued below) denotes a component state of the overall event; this state can be fully instantiated during the moment at which it is reported. In French, by contrast, the simple present Il tombe (lit. “he falls”) is acceptable, insofar as the present is here interpretable in a manner analogous to that suggested for the English present-progressive. In French, therefore, the simple present covers a semantic territory shared by simple and progressive presents in English. As noted by Smith (1991: Ch. 5), “formalization of the limitation of Present sentences will vary from language to language.” This argument is echoed by Cooper (1986:29ff.), who regards the English present tense as encoding a variety of temporal reference that is uniquely constrained vis-à-vis present-tense analogs in other languages:

English is quite exotic in having the reportive sense of the simple present. In most other languages (even such closely related languages as German and Swedish) the simple present behaves much more like the progressive behaves in English. We could account for this difference by saying that the semantic interpretation of the present in other languages requires the discourse loc...

Table of contents

• COVER PAGE
• ASPECTUAL GRAMMAR AND PAST-TIME REFERENCE
• TITLE PAGE
• COPYRIGHT PAGE
• ACKNOWLEDGEMENTS
• LIST OF TERMS
• LIST OF ABBREVIATIONS
• INTRODUCTION
• 1: ASPECTUAL MEANING
• 2: THE GRAMMATICAL EMBODIMENT OF ASPECTUAL MEANING
• 3: THE ENGLISH PERFECT SYSTEM
• 4: THREE PERFECT FORMS
• 5: INTERPRETATIONS OF THE PRESENT PERFECT
• CONCLUSION
• NOTES
• REFERENCES