You are on page 1of 18

CHAPTE R 4

VIEW PO INT A SPE CT

I begin by discussing the semantic information conveyed by aspectual viewpoints,


and semantic tests for the meaning of a viewpoint. The next section, 4.2, presents the
main types of aspectual viewpoints, with examples of variations that orcur in the five
languages of this study. 4.3 sets out several arguments for the independence of view-
point and situation type. Finally, in 4.4 1 discuss pragmatic inferences and conven-
tions of use associated with the viewpoints.
Aspectual viewpoints function like the lens of a camera, making objects visible to
the receiver. Situations are the objects on which viewpoint lenses are trained. And just
as the camera lens is necessary to make the object available for a picture, so view-
points are necessary to make visible the situation talked about in a sentence.
The main types of aspectual viewpoint are treated here as categories of Universal
Grammar. Universal Grammar provides a general schema fot each viewpoint. These
general schemata underdetermine the properties of the viewpoints; along certain
dimensions they are realized differently in individual languages. The evidence pro-
vided by input from a given language enables the learner to fix the parameter for that
vi e wpoi Il t. 1 I will use the terms ‘perfective', ‘imperfective’ and ‘neutral’ to refer both
to the general viewpoint categories and to language-particular instances of them.
Aspectual viewpoints have consistent semantic meanings for the language in which
they appear. This is close to the view of Roman Jakobson in spirit. Jakobson claimed
that categories such as the imperfective in Russian have an invariant semantic mean-
ing, a Gesamibedeutung which accounts for all uses of that category (1932, 1957). The
implementation is rather different from Jakobson’s, however: the viewpoint meanings
give positive information, and are complemented by pragmatic meanings.
Pragmatic interpretation makes an essential contribution to the interpretation of thc
viewpoints. At the pragmatic level, semantic meaning interacts with such factors as
62 V I EW PO I NT AS PEC T THE PA R AM ETER 0 F AS PECT 63

tion of their viewpoint systems. There are important language-particular conventions progressive, and its composite temporal schema, represented in (lb). An internal
in addition to the general pragmatic conventions of inference and world knowledge. stage of the telic event [Mary walk to school] is visible.
The information conveyed by viewpoint constitutes its semantic meaning.
Semantic meaning is conveyed by linguistic rorm and cannot be cancelled. View- (1) a. Mary was walking to school.
points span all or part of a situation: more precisely, the temporal schema of a situa- b. I . . ////// . . Fp
tion. Therefore what information a viewpoint presents is affected, and limited, by the
structure of the situation talked about.2 However, viewpoint must be stated indepen- I claimed above that sentences like (la) have an open interpretation. I now justify the
dently of situation type. This point is essential to the two-component theory. It is claim by showing that ( Ia) does not entail that a complete event occurred. I will use
based on evidence from the viewpoints of different languages as well as theoretical the technique of indirect proof, conjoining the sentence with a clause that asserts non-
considerations. I will make the case for the independence of the components in completion. I then ask whether the conjunction is reasonable or contradictory. If (la)
Section 4.3, after discussing the range of aspectual viewpoints in some detail. semantically conveys the completion of the event, such a conjunction should be con-
The analysis of viewpoint is at the heart of the two-component approach. I make tradictory. But if not, the conjunction should be reasonable. Consider (2):
several new proposals. I propose that the basic inventory of viewpoints includes the
Neutral viewpoint, a default in many languages; the proposal extends the tange of the (2) Mary was walking to school but she didn' i actually get there.
theory to languages without grammaticized viewpoints. I show that viewpoint is inde-
pendent of situation type. 1 distinguish the semantic and pragmatic levels of interpre- (2) is in fact entirely reasonable. This shows that (la) does not present a complete
tation: I posit explicit, positive meanings at the semantic level, and the traditional event: it does not mean that Mary actually completed her walk to school. In our terms,
notions of contrastive meaning and inference at the pragmatic level. the final endpoint of the event is not linguistically presented.
There is one inference about endpoints that is licensed by the information of (la).
By a default inference, one can conclude that the initial point of the event has
4.1 Semantic Information and Aspectual Viewpoints occurred. The inference follows from the fact that part of the event is visible. One
might make other inferences about the event, of course. Fot instance, one might infer
The two-component theory requires that all sentences have a viewpoint, since situa- on pragmatic grounds that the final endpoint occurred. This second inference might
tion type information is not visible without one. This theoretical requirement has the be reasonable if no information to the contrary is given, or if one knows Mary’s habits
interesting consequence that sentences with no explicit aspectual morpheme must well. The two inferences are quite different in basis. The inference of an initial end-
have an aspectual viewpoint. I posit the Neutral viewpoint as a default for such sen- point is semantically entailed by the material that is visible; whereas the inference of
tences. The default viewpoint gives partial information, which allows for the inter- a final point is pragmatically based. The former cannot be over-ridden by other
pretations that arise. In addition to the languages discussed here, I draw on the gen- information, whereas the latter can be. Receivers use both types of inference in inter-
eral studies of Comrie 1976, Dahl 1985, and the references in the Introduction. preting aspectual meaning.

Visibility: the aspectual information conve yed by a sentence Semantic tests for visible information
Aspectual viewpoints focus all or part of a situation. what is in focus has a special sta- As tools of linguistic investigation I introduce several simple semantic tests which
tus, which I will call ‘visibility’. Only what is visible is asserted. Visible information allow us to investigate the aspectual meaning of a sentence. No one test completely
about an event is available to the receiver of a sentence for truth-conditional issues determines aspectual meaning, but together they are quite informative. Conjunctions
and entailments. The visible information of a sentence is conventional and cannot be and questions show whether a sentence presents an open or closed situation.
changed, or cancelled. Receivers may make additional inference; these are conversa- Temporal clauses function as diagnostic contexts. The tests demonstrate well-known
tional and can be cancelled. The contrast between conventional and conversational properties of familiar perfective and imperfective viewpoints. They are also useful for
meanings, which can and cannot be cancelled, is due to Grice 1975. studying properties of unfamiliar viewpoints.
The main semantic difference among aspectual viewpoints is in how much of a sit- Conjunction tests are based on the compatibility of two assertions. To test whether
uation they make visible. Perfective viewpoints focus a situation in its entirety, includ- an aspectual viewpoint is open or closed in a given sentence, we conjoin it with an
ing endpoints; Imperfective viewpoints focus an interval that excludes endpoints; assertion about the situation in question. Open situations are compatible with asser-
Neutral viewpoints include the initial point and at least one stage of a situation. tions that the situation continues, or was terminated without completion (if tht situa-
In analyzing aspectual viewpoints, I rely on evidence from semantic tests for the tion is telic). If a sentence is reasonable in conjunction with such assertions, we can
meanings that are conventionally conveyed. The following familiar example illus- conclude that it presents an open situation. One version of this test was used above in
trates the kind of evidence that I will adduce. Consider tht interpretation of (I a), a sentence (2), repeated here as (3a); (3b) gives another version of the same test.
64 VIEW PO I NT AS PECT TH E PA RA METER OF AS PECT

(3) a. Mary was walking to school but she didn’t actually get there. In the (a) sentences, the main clause events were in progress at the time of the other
b. Mary was walking to school and she’s still walking. event (swimming, preliminaries to leaving). This shows that the imperfective view•
point does not semantically include endpoints. In the (b) sentences the main clause
Both conjunctions are reasonable, showing that the imperfective viewpoint does not events began or took place at the time of the other event. The perfective clause is
entail that Mary actually completed her walk to school. The final endpoint of the taken as an inceptive in (6b’) because swimming is a durative event. It is not plausi-
event is not visible. ble that a swimming event occur in its entirety at the same time as a bell ringing, but
In contrast, the perfective viewpoint is not compatible with assertions of continu- entirely plausible that it hegin at that time.
ation and incompletion, as (4) shows: Despite the flexibility of when, the sequential reading does not arise for sentences
with an imperfective viewpoint in the main clause. Thus sentences with when-claus-
(4) a. I Mary walked to school but she didn’t actually get there. es function as diagnostic contexts for the interpretation of viewpoint. There is a pre-
b. # Mary walked to school and she’s still walking. ferred pattern of interpretation for perfective sentences. If the situations are succes-
sive, the event of the w'heii-clause is taken as preceding (Steedman 1981). As I will
The impossibility of conjunction here shows that the perfective presents a closed sit- show below, the test allow us to discriminate between imperfective and neutral view-
uation. points: the former does not allow a sequential interpretation, whereas the latter does.
Perfective and imperfective sentences are both compatible with assertions that an Sentences with before- and after-clauses are also useful. They require a sequential
event is closed, for different reasons. The perfective asserts closure, while the imper- or overlapping relation between situations. They are odd, even ungrammatical, with
fective is compatible with an inference of closure. (5) illustrates: a main clause in the imperfective viewpoint. Compare the sentences of (7):

(5) a. Susan built a house and now it’s finished. (7) a. John left after Mary broke the glass.
b. Susan was building a house and now it’s finished. b. John was angry after Mary broke the glass.
c. ?*John was singing after Mary broke the glass.
Both of these sentences are reasonable, though (5a) is somewhat redundant. Con-
junctions are not informative about whether a closed reading is semantic or pragmat- The context of a beyore- or over-clause indicates whether the viewpoint of a sentence
ically licensed. This limits their usefulness in determining the information that is is imperfective.
semantically conveyed by an imperfective sentence. Questions can also function as tests that delimit the semantic meaning of a sen-
Sentences with temporal clauses give information about the semantic properties of tence. If the sentence presents an open situation, questions about its continuation are
aspectual viewpoints. They do so because they present situations in temporal relation reasonable; if the situation is closed, such questions are not reasonable. (8) illustrates:
to each other; temporal relations depend on whether the situations are presented with
or without endpoints. The minimal requirement for successiveness is that an endpoint (8) a. Martin walked to school.
of one situation must follow that of the other (Heiniimaki 1974). Before- and ofier- b. Martin was walking to school.
clauses require a sequential interpretation; they are diagnostic contexts for whether c. Did he get there?
the final endpoint of a situation is semantically visible in a sentence. The main claus-
es of such sentences must have a closed viewpoint. lY/ieri-clauses are flexible, allow- The question is a reasonable one for the imperfective sentence, but strange with th
ing several interpretations. Both types of temporal clauses are useful in studying perfective because the latter gives the answer.
viewpoints. These tests and variants along the same lines will be used to establish the stman•
When imposes no particular temporal relation on situations. The situations pre- tic meaning of the viewpoints.
sented may be taken as simultaneous, overlapping, or successive, depending on view-
point and situation type. Consider the examples of (6). The temporal clause presents
an Instantaneous event; the main clause events differ in duration, and in viewpoint: 4.2 Families of Viewpoints

(6) a. Mary was swimming when the bell rang. Knowing a language includes knowing the semantic and pragmatic values of the
b. Mary swam when the bell rang. viewpoints in that language. Perfective viewpoints are closed informationally, in the
sense that they present situations as complete with both endpoints. Imperfectives ers
a’ Bill was leaving when the bell rang. open. Neutral viewpoints allow readings that either both closed or open, although the
b’ Bill left when the bell rang. information given by a neutral viewpoint is not identical to that of an imperfective.
66 V IEW PO I NT AS P ECT TH E PA R A M ETER OF A S PEC T 67
These meanings have an important effect on the information structure of a sentence: This general schema represents the unmarked perfective viewpoint. As such it does
imperfectives are open to inference in a way that perfectives are not. Further, sen- not apply to stative situations, because endpoints do not appear in the temporal
tences with the neutral viewpoint are open in a way that imperfectives are not. schema of a state. Nor does the schema include perfectives with a span greater than
The distinction between closed and open situations pertains not to real time but to that of the situation schema focused. Perfective viewpoints that have these addition-
conceptual or narrative time. What a sentence conveys is a conceptual presentation. al properties are marked in relation to the prototypical perfective.
More formally, linguistic information licenses the semantic representation associated I now demonstrate some of the semantic properties of perfective sentences. I will
with a sentence as open or closed. Situations presented with and without endpoints be interested in the closed readings that are typical of perfectives, using English as an
affect interpretation when they appear in isolation. The presence of endpoints also has example. Since we are concerned with the meanings conveyed by linguistic forms,
consequences in discourse. “The difference in representations with and without end- we must distinguish between semantic meaning and the implications that may arise
points ought to show up in the possibilities ... for the interpretation of sentences from its presence in a sentence. We ask whether an interpretation is due to semantic
which follow them. That is, one would . . . expect that when trying to place the..con- meaning alone, or whether it includes pragmatic inference. To answer the question we
stituent introduced by the next sentence within the narrative time structure we have will put the sentences in contexts where their interpretation might be expected to vary.
already constructed, we would be able to make use of ... end points” (Kamp & If an interpretation remains invariant. it is conveyed by the linguistic form and is part
Rohrer 1989: ch 1, 15-16). of the semantic meaning of the sentence.
Viewpoint aspect is expressed by a grammatical morpheme associated with the The English perfective viewpoint is often called simple aspect because it is sig-
main verb of a sentence. The morpheme may simply indicate the viewpoint, or may nalled by the simple form of the main verb; the imperfective is signalled with the aux-
have lexical content as well. Of the languages studied here, Chinese and Russian have iliary be+•!"8 The perfective is incompatible with an assertion that the event conün-
perfective morphemes with clear lexical content. In analyzing an aspectual system it ued. Consider the interpretation of non-stative sentences with the perfective viewpoint:
is necessary to consider temporal location (indicated by tense, adverbials, and
modals) as well as viewpoint. Temporal location is discussed in Chapter 5 and in the (10) a. Lily swam in the pond. (Activity)
analyses of individual languages presented in Part II. b. . Mrs Ramsey wrote a letter. (Accomplishment)
Situations provide the locus for aspectual viewpoints. There are different patterns c. Lily coughed. (Semelfactive)
of interaction between viewpoints and situation types of a language. They can be d. Mr Ramsey reached the lighthouse. (Achievement)
understood in terms of the notion of dominance: a language may or may not have a
dominant viewpoint. Some systems are asymmetric: one viewpoint is limited and There are two points to notice about the interpretation of these sentences. They pre-
another is not, e.g. English and Russian. In these languages the ‘dominant viewpoint’ sent the events as closed, with initial and final endpoints (for durative events); and the
is available for all situation types. In other languages all viewpoint choices are avail- events are taken as terminated or completed depending on the situation type of tlu
able to all situation types, and no viewpoint is dominant. French is an example. In sentences. ( I0a,c) present terminated events, while (10b,d) present completed events.
others, viewpoint choices are available only for non-statives, so that statives are in The interpretations are due to the semantic meaning of simple aspect and not to
effect outside the viewpoint system; Chinese and Navajo are of this type. pragmatic factors. To substantiate this claim. I put the sentences in contexts that are
The viewpoint of a sentence presents all or part of a situation: more precisely, all not compatible with the closed readings just given, by conjoining them with asser-
or pan of the temporal schema of a situation. Temporal schemata are associated with tions that the events continue. If the closed readings are due to inference they will
the viewpoints, as well as with the situation types. The aspectual meaning of a sen- change in this context to open readings. But the readings will not change, and the con-
tence is a composite of the situation type schema and the viewpoint schema. I now junctions will be contradictory, if the interpretations are conveyed semantically. The
discuss the main viewpoint types, with examples of viewpoints in the languages ana- examples of (11) demonstrate.
lyzed in Part II.
(11) a. # Lily swam in the pond and she may still be swimming.
4.2. I Pe rfecti ve viewpoints b. # Mrs Ramsey wrote a letter and she raay still be writing it.

Sentences with a perfective viewpoint present a situation as a whole. The span of the The conjunctions are contradictory, showing that the closed readings are based on the
perfective includes the initial and final endpoints of the situation: it is closed infor- semantic meaning of simple aspect. (This statement oversimplifies the interpretation
mationally. This is the basic property of the perfective; the general schema is given of Activity sentences. In discourse contexts the interpretation is more flexible. The
in (9). perfective imposes an implicit bound, but the event need not actually terminate.°)
Now consider the type of final point conveyed by the perfective sentenccs: do they
(9) General schcma for the perfective: I F c°n*ey termination or completion? The interpretations vary with situation type. Th¢
Activity sent nc° conv°x• termination fLilY stoooed swimmina) whereas the Accom•
68 VIEW PO I NT ASP EC T TH E PA R A METER 0 fi ASPE CT 69

plishment conveys completion (Mrs Ramsey finished the letter). The difference may b. # Wo zuotian xie-wan-Ie yifeng xin, keshi mei xie-wan.
be due to the semantic meaning of the sentences, but it may also be due to inference I yest’day write-LE oneCL letter, but not write-finish
which augments that information. It would be natural to infer that the telic event had I wrote a letter-finished a letter yesterday but didn’t finish it
been completed, since we are given no information to the contrary. Conjunction is
again useful. If the simple viewpoint semantically conveys completion, an Accom- (14a) entails that the event was completed, as (14b) shows. Resultative Verb
plishment sentence should be incompatible with the assertion that the event was ter- Complements in Chinese have lexical content. One might relate them to panicloi
minated rather than completed. This prediction is tested in (12): such as up, which have a perfective force as in wash yp, eat up), other such mor-
phemes give information about the resultant state of a telic event. Hindi has a similar
(12) a. 8 Mrs Ramsey wrote a letter, but she didn’t finish writing it. system, with an independant morpheme that unequiv cally indicates completion
b. # James fixed the clock. but he didn’t finish fixing it. (Singh 1991).
c. # Mary opened the door, but she didn’t get it open. Russian also has perfective morphemes with lexical content. They appear in
Activity sentences with the perfective viewpoint. The perfective viewpoint is indicat-
The conjunctions are contradictory, showing that the implicature of completion can- ed by the prefixes po- and pm-. There prefixes are perfective in viewpoint, and give an
not be cancelled. The conjunction tests show that the implicature is conventional explicit bound for the Activity event. (15) illustrates, with an Activity verb begol (run).
rather than conversational. Conventional implicatures are conveyed by the meaning
of linguistic forms: thus, perfective viewpoint Accomplishment sentences in English (15) Ivan pobegal dva casa.
semantically convey completion. Ivan ran°< two hours
The interpretations vary according to the situation type of a sentence in English Ivan ran for two hours
simple aspect. If the situation type schema has a natural endpoint, so does the per-
fective sentence. This interaction between situation type and the perfective viewpoint The prefixes, known as delimitatives, indicate that an Activity is specifically limited,
occurs in French and a number of other languages. It does not occur in all languages, They also contribute an additional nuance of meaning. Po- indicates that the duration
however: Chinese has a different system. of the event is short; pro- suggests that the interval is unexpected in some way. The
The Chinese perfective morphemes (there are two) convey termination for all non- claim that these prefixes are part of the viewpoint system is defended in the discus•
stative events. The notion of completion is expressed by a set of optional, indepen- sion of Russian. The Russian perfective requires that the boundary of an event be lez•
dent morphemes known as Resultative Verb Complements. The examples illustrate ically indicated. The delimitative prefixes anchor an Activity to a specific interval of
with the perfective morpheme -fe and an Accomplishment sentence. (13a) does not time (Flier 1985: 49). The other perfective morphemes occur with telic verb constol•
entail that the event was completed, as (13b) shows. famous and do not have lexical content.

(13) a. Wo zuotian xie-ie yifeng xin. The parameter of statives arid the perfective viewpoint
I yest’day write-LE oneCL letter Perfective viewpoints prototypically include the endpoints of a situation, so that the
I wrote a letter yesterday schema of the Universal Grammar perfective does not appl y to statives. The end•
b. Wo zuotian xie-ie yifeng xin, keshi mei xie-wan. points, or changes into and out of a state, are not part of the state itself. No prediction
I yest’day write-LE oneCL letter, but not write-finish about perfectives and statives is made by Universal Grammar: this parameter varies
I wrote a letter yesterday but didn’t finish it according to individual languages. There are three different relations between statives
and the perfective viewpoint in the languages studied here. The perfective included
If the conjunction in (13b) were contradictory, we would conclude that the perfective changes into and out of a state and thus applies to the stative sentences (French); the
-Ie necessarily indicates completion. In the absence of information to the contrary, - perfective does not include the endpoints of states, departing from the canonical
Ie often suggests completion. But the suggestion is conversational. Completion is structure by presenting open stative situations (English); the perfective does not apply
indicated unequivocally by another morpheme: the completive morpheme -wan, a to statives at all (Russian, Chinese, Navajo). This variation is not a problem for the
Resultative Verb Complement, which may appear in addition to -le. (14) illustrates; theory, or for learners of these languages. The child learning the aspectual systeM of
the English translation cannot be made with a simple verb. each language can conclude from positive evidence whether and how the perfective
viewpoint applies to states.
(14) a. Wo zuotian xie-wan-ie yifeng xin. In French the perfective viewpoint is available for all situation types with a con•
I yest’day write-WAN-LE oneCL letter sistent closed interpretation. ( 16a) is a Stative sentence in the Posed Compoc4, one of
Yesterday I wrote (and finished) a letter the perfective past tenses of French. The conjunction in ( 16b) shows that the eituatio0
ir presented as clorcd: ( I 6b) ie contradiciory.
70 VTEWPOINTASPECI TO E PM R A M E TF R OF A S PECA

(16) a. Marie a vécu â Paris. Other marked perfective viewpoints


Marie lived in Paris There are marked perfective viewpoints which have a span beyond the situation in
b. # Marie a vécu ii Paris et elle y vit encore. focus. The most common is the Perfect, a complex construction that involves tempo-
Marie lived in Paris and she still lives there. ral location as well aspectual information; it is discussed in Chapter S. Another
marked perfective in the languages investigated here is the -quo viewpoint of
The stative cannot have the interpretation that Marie continues lives in Paris as pan Chinese, sometimes known as the 'experiential -gun'. The -8•° •' wpo int extends
of the same ongoing state. beyond the final endpoint of a situation.
In English the perfective viewpoint appears with stative verb constellations. Such The -guo viewpoint presents a telic event and asserts that the final state no longer
sentences present situations that are not necessarily closed. Rather, Stative sentences obtains. Thus the span of the viewpoint goes beyond the final endpoint of the event.
in English allow either an open or closed interpretation. (17a) for instance may felic- (19) illustrates: the meaning of the viewpoint cannot be conveyed directly in an
itously accept either of the continuations below: English translation.

(17) a. Jennifer knew Turkish. (19) a. Mali shang-ge yue qu-guo Xiang Gang.
b. .... but she has forgotten it all (closed) Mali last CL month go-GUO Hong Kong.
c. ..... and she still knows it (open) Mali went to Hong Kong last month (no longer there on the same trip).

Both readings are possible because the first moment of a state is like every other b. Wo bing guo.
moment. The analysis is discussed further in Chapter 8. I sick GUO.
The perfective is not available to statives in Chinese, Russian, and Navajo. These I was sick (no longer sick with the same sickness).
languages have no perfective sentences with stative verb constellations and the inter-
pretation of a basic-level stative situation type. Stative verb constellations do allow These sentences convey semantically that Mali is no longer in Hong Kong, that I am
the perfective viewpoint when they undergo a shift in situation type. They appear as no longer sick. The presentation of the viewpoint requires a discontinuity between the
inchoatives, in derived telic sentences. As such they present a change into the state final stage of a situation and the current state of affairs. The final stage of the situa•
which the verb constellation lexically denotes. In situation type they are either tion no longer obtains. At a more fine-trained level of analysis, the notion of discon•
Achievements or Accomplishments, depending on the feature of duration. The tinuity bas different forces wilh different types of events. (19) presents telic events
inchoatives vary depending on the pattern of the language. In Chinese they appear with transitory resultant states. They involve an affected subject or object. Ttie notion
with some statives but are ungrammatical with others. as (18) illustrates: of discontinuity is vacuous ror atelic situations and telic situations with non-transito-
ry final states and experiencer subjects, such as eating a meal or reading a book, us
(18) a. Mali bing-Ie. Yeh (1996) points out; see Chapter 11.
Mali sick-LE The conjunction test demonstrates that the interpretation of discontinuity is seman-
Mali got sick tically required by the -quo perfective. If it arises pragmatically, the interpretation
should disappear in the context of an assertion that the state continues. But in fact the
b. * Mali congming-ie. -quo perfective cannot be conjoined with an assertion of the continuing final state.
Mali intelligent-LE (20) illustrates; the first conjunct is the -8•° rf ctive of (19a):
Mali became intelligent
(20) # Tamen shang ge yue qu-guo Xiang Gang; hai zai nar.
The possibility of a derived inchoative is partly predictable from the semantic class they last CL month go-LE Hong Kong; still at there
of the stative predicate. Stage-leveI stative predicates can appear with viewpoint mor- Last month they went to Hong Kong and they are still there
phemes on a telic reading, as above; but individual-level predicates do not allow this
construction. The coming about of a predicate such as con8 ming C an only be The sentence is contradictory, as predicted by the analysis given here.
expressed with a lexical verb like chengwei (become); see Chapter 11 for discussion. To account for the span of this viewpoint we need a representation that is not depen-
In Russian the possibility of derived telic sentences with basic-level stative verb dent on the event itself. The representation must allow the viewpoint to span an interval
constellations depends on moiphol ogy. If a perfective form is available for a stative that includes the event and a subsequent stage that differs in the relevant way. The span
constellation, it functions as a derived telic with that form. In Navajo, an inceptive oL -gyo is represented in (21). The top line represents the temporal schema of an event.
prefix on a verb base is required for the inchoative interpretation. Such forms are I and P are its endpoints, F+1 indicates a stage distinct from the final stage; more for-
available for many but not all statives. mal statement of the viewpoint will specify that P+I requires a change of state from P.
72 VIEW PO I NT ASPECT
TH E PA RA METER 0 F AS PECT 73

(21) Span of the perfective -quo: I F F + 1 or termination rather than the occurrence of an event as a whole, but need not do to
(as Comrie shows: he presents a useful set of counter-examples to some general
claims about the perfective 1976: 17-24). I summarize the chief endpoint differences
The guo perfective viewpoint appears with the full range of situation types. The dura- among perfectives in the languages studied hete. The perfective may vary in mean•
tive event in (21) is representative only. ing according to the endpoint type of a situation: telic events have natural endpoints,
atelic events have arbitrary endpoints. Perfectives vary this way in English, French,
Punctual preseitiation and perfective viewpoints Russian, and Navajo. But in Chinese perfectives termination and completion are
Perfective viewpoints tend to present situations as punctual. The impression of expressed separately for all situation types. Perfective morphemes have lexical con-
punctuality arises from the closed nature of the perfective presentation. We conceive tent in Chinese and Russian, but not in the other languages studied. For Activities,
of a punctual situation as a simple, closed structure which appears at a point of time. dutative atelic events, Russian requires the explicit presentation of a specific final
When situations follow each other in time we may think of each situation as occu- endpoint; most languages do not. This list does not include differences in conventions
pying a point whatever its actual duration. Given a time line studded with dots, a dot of use, which also affect the interpretation of viewpoint aspect.
may stand for a war, a battle, an explosion. But to say that a punctual event occu-
pies “a point of time” in this way makes no claim about what literally happens in 4.2.2 Imperfeetive viewpoints
the world, or about the time course of the idealized event. As Lyons puts it, the
notion of punctuality is subjective and pragmatic (1977: 708-710). Punctual pre- Imperfective viewpoints present part of a situation, with no information about its end-
sentations of situations allow the speaker to convey narrative or other pragmatic points. Thus imperfectives are open informationally. The unmarked imperfectivt
meanings. spans an interval that is internal to the situation. This conforms to the general princi-
The perfective linguistically presents situations as punctual whether or not they ple that the span of a viewpoint coincides with all or part of the temporal schema of
actually have internal structure, and whether or not they actually take time. the situation. (23) gives the span of the unmarked imperfective.
Achievements and Semelfactives, which are instantaneous and have no internal struc-
ture, are perhaps prototypical punctual situations. But situations that take a moment, (23) General imperfective temporal schema: I . . ///// . . F
a few minutes, or years, can all be presented as punctual. Evidently, a ‘punctual situ-
ation’ does not correspond to an actual point, either conceptually or truth-condition- This skeleton schema conveys the information of imperfective viewpoints. Marked
ally. In fact, such points may be entirely impressionistic: there may not be such min- imperfective viewpoints span the preliminary stages of events or the resultant stages
imal points in actual theory or in practice. Many scholars view time as dense rather of telic events.
than discrete; if this view is correct, there is no smallest unit of time and the notion The two most common imperfectives are the general imperfective and the pro-
of a temporal point is essentially subjective and arbitrary 4 gressive. The former focuses intervals of all situation types; the latter applies only to
Not all perfectives are punctual in presentation. Duration can be explicitly assert- non-statives. In 4.3.2 I consider the relation of the imperfective to states, a frequcnt
ed in sentences with the perfective viewpoint: topic in current semantic discussions of aspect.
The French /mpaJnir exemplifies the general imperfective viewpoint; it is a past
(22) a. The king reigned for thirty years. tense with imperfective aspectual value. The viewpoint appears with sentences of cll
b. 1 wrote the sonnet in 5 minutes. situation types with internal stages, as (24) illustrates. The superscript indicates view-
point.
The examples show that the perfective presentation of a situation is entirely compat-
ible with expressions of durativity. This seems odd only if the properties punctual and (24) a. La mer était calme.
durative are in contrast. In fact they do not contrast in the perfective viewpoint. The sea wasImp* calm (Stative)
Semantically, the perfective presents events as closed, with no information about b. L’enfant pleurait.
duration. The punctual interpretation is strongest and most natural when duration is The child was crying'=r'
not explicitly mentioned in a sentence. This notion of ‘punctual’ is quite different c. Ils bâtissaient une cabine.
from that of an event that is instantaneous in principle, or adverbials that refer to a They were building'=P' a cabin (Accomplishment)
point in time. To avoid confusion I will refer to the perfective as presenting ‘closed’
situations, reserving the word ’punctual' for the impressionistic perfective presenta- The English translations are, of course, only approximate: the English progressive ifi
tion. I call Achievements and Semelfactives ‘instantaneous’, and for adverbials refer- the closest translation equivalent but it differs subtly from the Imparfoit. 1 use tht
ring to a point I usc the word ‘momentary’. conjunction test to ask whether the lmparfait presents events as continuing, or open
Perfectives vary considerably across languages. They may emphasize completion (25) conjoins (24b), an Activity sentence in the lmpurfait, with an assertion in the
74 U IEW PO I NT ASPECT T H E PA R A M ET E R 0 F ASPECT 75

Présent that the event may still be in progress. The result is reasonable, showing that (28) a. John sang when Mary knocked at the door.
the viewpoint of the first conjunct is open. b. John was happy when Mary knocked at the door.

(25) Ce matin l’enfant pleurait; peut-être qu’il pleure encore. The Activity main clause in (26a) has a closed interpretation only. The main clause
ThiS mornin g tht• C hil d C r y' rf'; pt•ï haps he iS Still C rying*e• in (26b) has both an open Stative and a closed event interpretation. On the Stative
reading John was already happy before Mary knocked; the open interpretation is typ•
The information in the first conjunct would also be consistent with an assertion that ical of English perfective statives. On the event reading, which is closed, John
the child is no longer crying, of course. Imperfective viewpoints and Instantaneous became happy at the time of her knock. Closed readings are typical of English per-
events are discussed in 4.2.2 below. fective non-statives. The two readings depend on whether the verb constellation is
Progressives focus on the internal stages of non-stative events. Dahl notes that pro- taken as a Stative or as a shifted inchoative. The contrast between (27) and (28)
gressives tend to appear in all tenses if a language has tense (1985:92). In contrast, shows that sentences in the progressive are not available for a closed interpretation.
general imperfectives tend to be limited to past tenses. Among the 64 languages for The interpretation of when-clauses depends on viewpoint, situation type, and prag-
which Dahl collected information, at least 28 had progressives. Of the languages stud- matic factors (Boertien 1979, Dowty 1979, Steedman 1981, Smith 1984).
ied here, English has a progressive, French and Russian have general imperfectives,
Chinese a progressive and a resultative, Navajo a progressive and an imperfective. Marked imperfective viewpoints
The English examples illustrate for Activities and Accomplishments, which are In this section I discuss imperfective viewpoints that focus on the external stages of
both events that have internal stages; (26c) a Stative, is ungrammatical. a situation. focusing the preliminary and/or the resultative stages of a situation. Such
viewpoints are marked.
(26) a. Kelly was singing. (Activity) Since they have no internal stages, we would predict that imperfective viewpoints
b. Ross was climbing a tree. (Accomplishment) do not apply to instantaneous events. In fact the imperfective is often available for
c. *Bill was knowing the answer. (Stative) Achievements. Imperfectives present the preliminary stages of the event; there is no
suggestion in such presentations that the Achievement actually takes place. English,
The progressive viewpoint has meanings that do not arise for other types of imper- Prench, Russian and Navajo have imperfectives of Achievements, whereas Chinese
fective. Nuances of activity, dynamism, and vividness are often associated with sen- does not. I present English and Chinese examples; Lai is the morpheme of the
tences of this viewpoint. The connotations can be traced to the requirement of non- Chinese progressive.
stative events for the progressive. The stereotypes of such events involve activity
and change; the connotations invoke the stereotypes. In contrast, the Imparfait (29) a. The team was reaching the top.
appears with all situation types, and lacks the connotation of dynamism typical of b. She was winning the race.
the progressive. c. *Lao Wang zai dao dingfeng.
Imperfective viewpoints do not linguistically present closed situations, although Old Wang ZAI reach summit.
they allow inferences about beginnings and endings. The distinction is brought out by Old Wang is reaching the top.
linguistic contexts that involve open or closed readings. In such contexts closed read- d. *Ta zai ying sai pao.
ings arise only with perfective viewpoints. I illustrate with En glish progressive sen- she ZAI win race run.
tences in contexts that invite and require a closed reading, a alien-clause and an over- She is winning the race.
clause.
The temporal schema for the Achievement situation type includes preliminary
(27) a. John was singing when Mary knocked at the door. stages, for languages where the imperl‘ective may focus preliminary stages.
b. ?* Herbert was hiding the loot after the telephone rang. Semelfactives, which are also single-stage events, never appear with imperfectives
with the interpretation of a preliminary interval. The temporal schema of a
In these examples the main clauses are progressive. With the when-clause only an Semelfactive has no preliminary stages. This device accounts for the facts, but dots
open reading is available (27a); with the acer-clause, which semantically requires a not explain them. I suggested in Chapter 2 that this difference between
closed main clause, the sentence is ungrammatical (27b). These examples contrast Achievements and Semellactives may be due to the conceptual primacy of changes
with the sentences of (25), in which the main clauses are perfective and have a closed of state for human beings.
interpretation. In (2Sa) the events must be successive, in (28b) they may be succes- Preliminary focus for durative events is also possible. Navajo sentences allow such
sive ‹ir ‹overlapping. a focus with contextual support, as (30) shows; the verb base denotes a durative event:
76 V I EW PO INT AS PECT
TH E PA R A METER 0 F AS PECT 77
(30) Shidibe k’ad‹;q da’adlj. (33) Resultative imperfective viewpoints: 1 .... F .. //// ..
sheep almost drink 'mpf B: mr,
The sheep are about to drink. Sentences with a resultative imperfective focus are distinct from verb constellations
that lexically denote the resultant states of a telic event. (34) illustrates the latter:
In Navajo preliminary focus requires the adverb k bdp9 (almost) for durative events;
see Chapter 12. Russian also allows this interpretation. which is known as ‘conative.’ (34) a. Mary cooked the roast medium rare.
The other languages require additional lexical material for sentences of this type. b. Richard sliced the carrots into rounds.
Resultative imperfective viewpoints present a state that follows the final point of
a telic event. More precisely, such viewpoints focus an the interval after the change These sentences lexically denote the resultant state of a telic event.
of state. Resultatives occur in Chinese and in English, as the examples illustrate; -the Imperfective viewpoints may inherit or impose temporal properties on the stretch of
is the morpheme for one of the Chinese imperfective viewpoints: a situation that they focus. We have already discussed examples from English. The pro•
gressive viewpoint inherits the propeny of dynamism from the events to which it neu-
(31) a. Ta zai chuang shang tang-zhe. trally applies. Progressive stative sentences are marked, and convey the dynam-isrn of
he at bed on lie ZHE an event. There is even stronger evidence in Chinese, which has two imperfective view-
He is lying on the bed points. The analysis of imperfectives must be rich enough to account for the Chinese
impezfectjves zzz and -the, which differ in distribution and in meaning. Zai, a dynamic
b. Men shang xie-zhe sige zi. progressive, is
door on writeZHE 4 CL character able neutrally available
to stativesneutrally to non-statives; -zâr, a stative imperfective, is avail-
and has a relatively wide derived range. The difference
Four characters are written on the door between these viewpoints can be handled by the formal device of inheritance. Thy
resultative viewpoint indicated by -the has a static propeny, which is imposed on all sit•
c. Your socks were lying on the bed. uations that the viewpoint focuses. We provide for this by associating a static feat
d. The statue is standing on the corner. with the viewpoint. Through inheritance, we provide that properties of one represents-
cion are carried over, or inherited, into a representation which is derived from the fir$l
The English sentences (31c-d) are stative syntactically and semantically, although Analysis of the Chinese imperfective viewpoints shows that progressives are dis-
morphologically indistinguishable from progressives. They are discussed in Chapters tinct from statives. I return to this question in Section 4.3.2 below. Using inheritance,
2 and 8. Resultatives present a state via the event that brings it about: the lexical span we can say that each imperfecti ve inherits properties of the situations that it focuses,
oY the verb constellation includes a change of state. fact that their distribution is different explains the properties they inherit. The
When transitivity and morphological factors allow, a sentence can be ambiguous progressive zni inherits the stage property from the temporal schema of events. But
between an eventive and a resultative reading. In the former case the imperfective or
since states are the basic domain of the stative imperfective-zhe, this viewpoint does
progressive viewpoint focuses an internal interval of the situation; in the latter, an
not inherit the stage property. The presence or absence of the stage property cormct-
external interval. (32) illustrates for English and Chinese: ly distinguishes intervals focused with Lai and -she.
The extended, marked uses of a viewpoint can be accounted for by allowing it io
(32) a. John was sitting in the chair. impose a temporal property on the interval that it focuses. In its marked use the pro•
gressive imposes the stage property on stative situations. Similarly, the imperfective
b. Tianli zhong-zhe huar. -the imposes a stative coloration on non-stative situations. See the sections on these
land-in plant-ZHE flower viewpoints in Chapters 8 and 11 for further discussion.

Both sentences are ambiguous in the same way. On the internal reading, an event is 4.2.3 Neytyal Viewpoints
in progress: John is in the process of seating himself, the planting process is going on.
On the resultative reading the events have already taken place, and the resulting state In this section I discuss sentences that are aspectually vague, they lack a viewpoint
is focused. John is already seated and the flowers are already planted. The resultative morpheme. i argue that such sentences, which have neither a perfective not an imper-
is an important pattern in Chinese, and in other Asian languages; it is one of the major fective morpheme, should be analyzed as having the Neutral viewpoint. The Neutral
stative patterns in Japanese (Jacobsen 1982, Talmy 1985, Ogihara 1989, Du 1996). viewpoint is a default with a positive semantic value. It arises in aspectual systems
Resultative viewpoints can be handled naturally within the two-component frame- which Allow sentences without a viewpoint morpheme. There are both eGpirical and
work, since they span a period not included in the event proptr. The schema of (33) theoretical reasons for positing a default viewpoint.
illustrates: Empirically, tht interoretation of asncetuallv vaouo arniencca r»n toe mhm» tn M
78 V I EW PO I NT A S P ECT TH E PA RA M ET0E R F AS PECT

neither perfective nor imperfective. Such sentences are more flexible than either open or closed situations. This range of interpretation is not available for either the per-
viewpoint, in that they allow both open and closed readings. Do we endow such sen- fective or the imperfective viewpoints. Note that both sets of examples are needed to
tences with a new viewpoint, or say that they simply have no aspectual viewpoint? make this point. One could argue that the apparently open reading actually arises from
The latter view seems plausible, since context often indicates the favored interpreta- the 1nceptivc, wh1ch prescrits the beginning of the event has taken place. But this dons-
tion. I will argue for the second, more radical view. Both the range of meanings that n’t explain the preferred open reading of (36). The French Présent tense also has the
is found for aspectually vague sentences, and the theory-internal requirement of vis- Neutral viewpoint; see Chapter 9 for further discussion.
ibility, suggest that they have a viewpoint that is open but not unlimited.^ Cfii1nese UM sentences Have sim1Jar intozpretations. Viewpoint morphemes are
The argument will take the following form: I begin by considering the interpreta- syntactically optional in Chinese, and LVM sentences are not uncommon, especially
tion of sentences that lack a viewpoint morpheme, using the semantic tests introduced in discourse.* They do not normally occur in sentences with when-clauses, because
above. 1 will refer to sentences that lack a viewpoint morpheme as LVM sentences. I »'6en-clauses are generally used to foreground the event in the main clause. The mciv
show that both perfective and imperfective readings arise. This finding rules out the clause, foregrounded, events usually have a viewpoint morpheme. But for speakers
possibility that LVM sentences actually have one of the familiar aspectual values. I who accept them, LVM sentences with w/ten-clauses can be interpreted to present
then investigate further the interpretations that LVM sentences may have. 1 show that open or closed situations. For instance, some native speakers accept sentences like
certain readings do not arise. This finding rules out the possibility that LVM sen- (37) With the two readings given.
tences have no aspectual viewpoint and allow thus free interpretation.
The evidence from interpretation supports the idea that there is a specific view- (37) a. Zhangsan dao jia de shihou, Mali xie gongzuo baogao.
point value for LVM sentences. They allow closed and open readings, but are not Zhangsan arrive home DE time, Mali write work report
entirely unconstrained. To account for the facts I suggest a viewpoint that is infor-
nationally open. It differs from both the perfective and the imperfective. The Neutral b. When Zhangsan arrived at home, Mali wrote the work report
viewpoint is weaker than the perfective in allowing open readings. It is stronger than c. When ZliangSan arrived at home, Mali was writing the work report
the imperfective because it allows closed readings.
I now present examples showing that LVM sentences allow both open and closed On the closed reading the main clause is inceptive: Mali began writing at at time
readings, depending on context. I use examples from French. Chinese, and Navajo. roughly simultaneous with Zhangsan’s arrival (37b). On the open reading, the ttt@p
Consider first the future tense of French, the Futur, which conveys no viewpoint clauses presents an ongoing event: Mali was already writing at the time of Zhangsan'i
information.^ arrival (37c),
In the context of a when-clause the Latur presents situations that can be taken as Questions also bring out the flexibility of interpretation that is typical of LVM sen•
open or closed. The determining factors are situation type, context, and world knowl- tences. For (38a), an Accomplishment sentence,the question in (38b) is appropriatc;
edge. Consider the main clause of (35), which has an Activity verb constellation. (39) gives some possible answers.

(35) Jean chantera quand Marie entrera dans le bureau. (38) a. Mali xie gongzuo baogao.
ha n will Sing°•'whe Tt Marié will e n te r"u' the office. Mali write work report
b. TP hai zai xie ma?
The main clause has an open and a closed reading. The closed interpretation is an she still ZAl write MA?
inceptive one: Jean will start singing at the time of Marie’s entrance. On the open Is she still yriting
interpretation, Jean will already be singing when Marie enters. The inceptive possi-
bility clearly distinguishes these examples from imperfectives, which do not allow f39) a. Hai zai xie. b. Xie-le. c. Xie wan-le.
such an interpretation; recall 127a) above. Stlll ZAl write. write-LE. write-finish-LE.
According to native speakers of French, the closed interpretation is natural for (35) She is stiII writing. She stopped. She finished.
although both are possible. With different lexical items, the situation changes: in (36)
the open interpretation is more natural: Strikingly, all the answers are reasonable, showing that both open and closed read-
ings are available for (38a).
(36) Jean dormira quand Marie entrera dans le bureau. The same range of interpretation appears in LVM sentences of Navajo. I illustrate
Jean will sleepFU lyhen Marie will enter"•' the office. with two temporally related sentences in the Usitative and lterative modes. The con-
S1£UCtt Oft IS SIJII6F tO a n'/ieii-clause; there is no overt Navajo morpheme which cor-
On the open interpretation, Jean is already asleep when Marie enters. The closed inter- rcsponds to when. Thèse modes contrast with efiplicit viewpoint morphem•i in
pretation is inceptive here too. These examples demonstrate that the Curr can present /BŸBfO. g have fh0 kez/îrgl vinwwint
BO VIEW POI NT AS PECT TH E PA R A Mf TE R OU- A S PECT 8t

(40) a. dibe nanishka’go hodootal. dœs not have LVM sentences, the question arises only for Chinese. Unfortunately wt
when I herd sheep, he sing *•f4... U nit cannot conclusively test the ptediction for either marked case in Chinese. Resultativea
b. hastiin na’ad ii^go, ch'lnashdaâh. cannot be reliably distinguished from statives, because the relevant verb constellations
When my husband drinks, I leave l'•r. .UÄÎî are morphologically indeterminate. The -quo perfective is problematic in a different
way. The viewpoint requires a discontinuité between the resultant state of a situation
Most plausibly, the events of (40a) are simultaneous, whereas the most plausible and a subsequent time. But a discontinuité may be inferred in the absence of informa-
reading of (40b) presents overlapping events. The contrast is similar to that present- tion to the contrary. Thus inference from lack of information may always allow a dis-
ed above for French. continuité interpretation for LVM sentences. The prediction conceming marked view-
I now show that not all viewpoint interpretations are available for LVM sen- point spans must be studied with other languages that have the relevant structures.
tences. Consider the possible range of imperfective viewpoints. Imperfectives in 1 have shown that LYM sentences may have open and closed, but not preliminary
French and other languages often focus the preliminary stages of a situation, espe- interpretations. The temporal schema of the neutral viewpoint accounts for the range
cially situations of the Achievement type. We might predict, therefore, that LVM of interpretations that occurs. It also provides every sentence with an aspectual view-
sentences in French would have such an interpretation. But the prediction iS ROt point. This theory-internal requirement has the advantage of extending the scope of
borne out. Consider Achievement clauses in the Futur. If the preliminary reading the general account to languages like Finnish and Eskimo, which do not have explic-
were available for them, it should be reasonable to conjoin such Futur Achievement it viewpoint morphemes. Neither Finnish nor lcelandic has grammaticized view-
with assertions that the events will not actually occur. (41) illustrates; the transla- points, although there are optional lexical means for giving information about point
tions aie only approximate. of view (Heinämäki 1983, Fortescue 1984). The limitations on neutral viewpoint
interpretations hold for thèse languages. For instance, in Finnish and Eskimo, the
(41) a. La guerre éclaira. interpretation of Achievement sentences cannot invol ve preliminary stages of an
The war will break/be breaking out. Achievement without explicit lexical support. The same is true for Finnish.
b. # La guerre éclaira mais elle n’éclaira pas. The neutral viewpoint allows both open and closed readings. Its span includts the
The war will be breaking out * ' but it won’t break out^•'. initial point and at least one internal stage of a situation (where relevant). The tem-
poral schema is presented in (43):
a’ Le cheval gagnera.
The horse will win/be winning. (43) Neutral Viewpoint Temporal Schema: 1.
b’ # Le cheval gagnera le course mais i1 ne gagnera pas.
The horse will be winning the race^u' but he won’t win"•'. The neutral viewpoint complements the other viewpoints in the amount of informa-
tion it makes visible about an event. The neutral viewpoint includes one endpoint. the
The conjunctions are ill-formed. Similar examples can be constructed for Navajo, in perfective both endpoints, the imperfective neither. Thus unlike the imperfective the
which the future tense has no viewpoint morpheme. (42), for instance, does not have neutral viewpoint allows closed readings by inference.
the preliminary interpretation in which the subject is about to kick the ball. The neutral viewpoints investigated here have the same range of interpretation.
However, there might be differences in the neutral viewpoints of other languages.
(42) ‘ylitaL This is a subject for future research.
He is kicking it (e.g., ball) ^•'.

The question of such readings does not arise for Chinese, because the imperfective 4.3 J'he lnüependance où Viewpoint and Situation ’¿ype
never focuses preliminary stages.
The examples show that neutral viewpoints are not entirely flexible: they ÖO ROt In this section I will show that the components of situation type and viewpoint are
focus the preliminary stages of a situation. This finding suggests a strong and inter- independent. The discussion brings out several advantages of the two-component the-
esting prediction. The prediction is that LVM sentences do not have marked view- ory, which has thus far been assumed without argument.
point interpretations: in other words, that such sentences do not have a span beyond
the endpoints of a situation. 4.3. l Viewpoint span
This prediction might be tested for resultatives, intervals fo1lo;ving the ünal eud-
point of an event. There are viewpoints that focus such inteTvalG: the imperfective ’Fhe first argument for the independence of the components comes from the poiiible
rtsultative in Chinese and English, and the -goo perfection in Chinesc.Wc would like span of an aspectual viewpoint. There are a number of viewpoints whoae span does
to know whethqr the post-ovent focus is possible for LY i tenlflnc°8, Sinc° En81i ®h not coincide with a situation, ai we have seen. To acc0Unt for inch ¥lBWDO1nH wil•
82 VIEW POINT A SPECT
TH E PA R A M ETER OF ASPECT g$
out incoherence the viewpoint of the sentence must be independent of the situation prior situation and a post-final stage not part of the situation itself. This span cannot
schema. The relevant viewpoints have already been introduced; I bring them togeth- be stated coherently as dependent on the situation schema.
er here to make the theoretical point. Viewpoints with a span that does not coincide with the situation schema am not
Imperfective viewpoints may have a span that does not coincide with the tempo- uncommon, as the examples show. Because viewpoint is independent of situation
ral schema of a situation. For instance, the progressive may focus the preliminary type the two-composent theory can account nicely for the preliminary interpretation
stages of an Achievement, as in (44). of Achievements, the resultative interpretation of telic events, and the discontinuity
of the Chinese experiential.
(44) a. Algernon is reaching the top.
b. “Bright Star” is winning the race. 4.3.2 Visible information and the impe rfective paradox:

In these sentences, the viewpoint focuses an interval that is preliminary to the single Situation type is transparent to the receiver whatever the viewpoint of a sentence.
stage of an Achievement event. The event itself is not included in the span of the view- This constitutes another argument for the independence of the aspectual components.
point. This type or imperfective sentence appears in many languages, as we have seen. Consider a familial example, a sentence with the progressive viewpoint:
Similarly, the viewpoints of resultative sentences have a span that does not include
the event schema. Resultatives have an imperfective viewpoint that focuses a resul- (47) a. We were walking to school.
tant interval of a telic event. Such sentences are common in Mandarin Chinese, for b. Composite temporal schema: I . . / / / . . Fqp .
instance. (45) gives a Mandarin example with the stative imperfective viewpoint,
indicated by -the on the verb. Receivers of this sentence know that only part of the event is semantically visible. But
they also know what sort of event it is: [we talk to school] is telic, with a naturel final
(45) Zhangsan zai chuang shang tang-zhe endpoint. More piecesely, situation t ype information is aiways semanticaliy visible in
Zhangsan at bed on lie-ZHE a sentence. The verb constellation conveys the nature of the final endpoint, although
Zhangsan is lying on the bed the final endpoint is not visible in the sentence and may never occur. This i
sented in the composite temporal schema (47b), which contains information about
Resultatives appear in many languages. viewpoint and situation type.
Im ortant evidence for the independence of the components comes from the The composite schema correctly represents what the sentence conveys about view-
analysis of another Chinese viewpoint, the perfective gon , known as the experiential po1flt ü0d S1tMa{1On type, I beJieve. Such Jcnowledge fias been cailed paradoxical.
The experiential -quo asserts a discontinuity between the final endpoint of the prior Indeed, the understanding of progressives is said to involve an ‘inn perfective paradoz’
situation and the current state of affairs, as shown above. To see how it differs from (Dowty 1977)."
the span of the standard perfective, compare the sentences in (46), a -pro perfective In the two-componenf theory there is nothing paradoxical in the knowledge thet
and a -Ie perfecñve: the receiver has about (47). Situation type and viewpoint belong to independent com-
ponents of the aspectual system, and are signalled differently. Viewpoint is conveyed
(46) a. Tamen shang ge yue qu-Ie Xiang Gang. by a single morpheme; the constellation of a verb and its arguments conveys situation
they last CL month go-LE Hong Kong type. The linguistic forms do not contrast syntactically.
Last month they went to Hong Kong (they may still be there) Note that the same sort of knowledge is assumed without qualms for
tation of progressive atelic sentences such as Mary wes «wimmin
b. Tamen shang ge yue qu-guo Xiang Gang. Knowing that an Activity has an arbitrary final endpoint relies on infonnaÑonabout
the type of situation. The case is really the same for any sentence that presente part of
they last CL month go-GUO Hong Kong
ä Situation: it is not limited to sentences of telic situation types.
Last month they went to Hong Kong (they are no longer there) The information about situation type is intensional information about an ongoing
situation. It is difficult to implement this kind of intensional knowledge in a truth-
(46a) with perfective -le, simply asserts that the event of going occurred; (46b), the - conditional account. The problem, I take it. is to recognize that a fraction of an event
quo perfective, also asserts that the result state of the event, if transitory, no longer belongs to a larger event. In fact we make such conclusions all the time on the basis
obtains. The sentences are appropriate in different situations. The perfective with -fe
of partial knowledge. If 1 see Jane walking along a certain street early in the mom-
is felicitous whether or not the subjects are still in Hong Kong; but the perfective with *Pdg, I may think that she is walking to school. If I see that she is carrying her brief-
-8•• can bt said only if they are no longer there. base royhypothesis will be strengthened. Of course I may be wron$. The point ii that
The -quo viewpoint presents a closed situation with a change of state subsequent
gq VIE W PO I NT ASP ECT TH E PA R A METER OF ASPECT

My ability to correctly assess the situation of Jane walking down the street will There is also a distributional relation between progressive sentences and sentences
depend on my knowledge of events and of Jane’s habits, and it is quite independent associated with the stative situation type. They complement each other, since the pro•
of my inability to predict the future. I cannot know that Jane will rim into Mary, who gressive is available neutrally only to non-statives. This complementarity, and the fact
is playing hookey, and decide not to go to school after all; or, in another scenario, that that both types of sentence have the sub-interval property, may suggest that progrei•
she will be run over by a truck. Undoubtedly certain events in progress are less easy sives are statives. In fact the progressive has been analyzed as a stative operator
to recognize than Mary walking to school. One may sometimes be mistaken, one may (Taylor 1977, Vlach 1981). But the analysis must be rejected on both conceptual and
sometimes be in ignorance. empirical grounds.
Similarly, we frequently categorize objects on the basis of partial knowledge. There are several empirical arguments against identifying progressives with stA•
Suppose that I see a railroad car in front of me at the railroad station, but can’t see tives. If we look at imperfective viewpoints in other languages. we find that them is
anything else. I may assume that other railroad cars are behind it, and say that I have no inherent complementarity between the imperfective and the stative. The imper-
seen part of a train. If the railroad car was actually sitting alone on the track I have fective viewpoints of French and Russian apply to stative sentences, for instance. And
made an incorrect identification, due to the fact that I had only partial information. the two imperfective viewpoints of Chinese (Lai and -the) differ precisely in the prop•
These problems belong to a different discussion, and arise for any intensional account erty of stativity, as we have seen.
of ongoing situations. Statives are linguistically distinct from progressives in Chinese and English.
Statives are more flexible than progressives: they may be taken as open or closed
4.3.3 The relation between statives and imperfectives informationally, whereas progressives are never closed. Sentences with ivfiea-claum
show this difference. Compare the interpretations of (50a-b).
I will now discuss a vexed question in the semantics of aspect, the relation between
statives and progressives. The two are similar intuitively and formally: the question (50) a. Mary was angry when John broke the glass.
is whether they belong to the same aspectual category. I will show that progressives b. Mary was singing when John broke the glass.
differ in several ways from statives. Therefore, I argue, they should not be collapsed
into one global category. The argument strengthens the case for the independence of (50a), a stative, is ambiguous: either Mary was already angry before the event of tie
the components. main clause, or she became angry at the time of the event. (50b), a progressive, W
There is an interesting similarity between stative sentences, which have the perfec- only the ongoing interpretation.
tive viewpoint, and sentences with the progressive viewpoint. They share certain prop- Although the progressive is not neutrally available for statives, progressive static
erties and have similar effects in discourse. Both present open situations, witio nt end- sentences occur, particularly in informal discourse, as we have seen. (51) give some
points. Intuitively both stative and progressive sentences present stable situations, additional examples:
although progressives have successive stages and statives do not. The interval focused
by a progressive is a process; and processes are very like states. Processes may seem (51) a. John was really liking the play.
unchanging, although they have successive stages. With no end in view the stages of b. That cake is looking done.
an event. even a telic event, may seem merely to succeed each other in time. The fac- c. Amy is resembling her great-uncle today.
tor of dynamism may seem to be the only difference between a state and a process.
Formally, progressive sentences and stative sentences both have the subinterval In these examples stative verb constellations appear with the linguistic forms appto•
property and the pattern of entailment associated with it. Just as John Rios loved Mary priate for events, endowing them with the dynamism and other connotañons of events
/rom I¡ to t entails John loved Mory atj t , so John has been running from t to t¿entails Marked choice of the progressive is a live and much-used option in the language. If
John was running at iy, where t refers to any of the infinity of points or intervals progressives are identified with statives, we cannot explain the distinction betwen
standard and marked progressives.
between t, and th. The fact that imperfectives have the subinterval property may be
expressed formally with the notion of a larger interval. Dowty provides for this in Empirically, then, there is strong evidence against identifying progressives witlt
truth conditions for progressive sentences, reproduced here as (49). statives. The information conveyed by statives and progressive is similar in some
ways, as we have seen; but it arises differently. Sentences with the progressive are
(49) The progressive of a sentence S is true at interval I if there is an open informationally because the progressive does not include the endpoints of in
interval I’ properly containing I such that S is true at I. event. Stative sentences present situations which do not have endpoints in their tem•
poral schema. Strikingly, they play similar roles in narrative discourse. Neither type
Dowty shows that it follows from this definition that any sentcnce with the progres- of sentence tends to move narrative time.
sive has the subinterval property (1986:44 ct seq). Conceptually the tWO fl@ of different types. Progressives ate # type of ¡mpgrfy•
86 VIEW POI NT AS PECT TH E PA R A M ETE R OF ASPECT

tive viewpoint; they are very like statives because they have properties in common; 4.4.2 Augmented irite rpretation
but they are not statives. In the two component theory one is associated with an ide-
alized situation type, while the other is a viewpoint. Although they share the sub- The aspectual interpretation of a sentence may be direct, using the information that iz
interval property, they are not of the same category. semantically visible. Interpretation may also be augmented with additional informa-
As a final point, I note that situation type shifts also suggest the independence of tion arrived at by inference. An augmented interpretation contains information that it
viewpoint and situation type. I showed in Chapter 3 that there are several triggers for not due to the linguistic forms of the sentence. For instance, the receiver of(52) might
shifts in situation type. The triggers include viewpoints, and adverbials. The shifis are infer that the final endpoint of the event has occurred, even though no endpoints are
analyzed in a simple manner, in which viewpoints are not accorded privileged status. semantically visible.
In conclusion, I have adduced evidence of several kinds in support of the claim
that the components of viewpoint and situation type are independent. (52) John and Carolyn were painting the house.

Making such an inference, the receiver may add a final endpoint to their representa-
4.4 Viewpoints and Conventions of Use tion of the information conveyed by the sentence. Augmentation is often licensed by
information in the context, for instance, a later sentence may assume that the hou%
The aspectual meanings conveyed by a sentence include emphasis and information in question in (52) is completely painted. Receivers may have independent knowl-
arrived at by inference. These are pragmatic meanings, dependent on context and edge about a situation, or pragmatic information about the world which makes G
convention. They complement the semantic meanings associated with linguistic given interpretation plausible. The absence of information that contradicts an aug-
forms. The pragmatic meanings associated with a viewpoint are guided by conven- mented interpretation is also a factor. The notion of augmentation distinguishes int¢r•
tions of use. The conventions depend partly on general cooperative principles of pretations that depend on inference from those that arise from semantic meaning only.
inference and partly on the pattern of a particular language. The typical augmentation for sentences with an imperfective viewpoints is the
addition of endpoints to the interpretation. If situation is in progress, it follows tlict
4.4.I Pragmatic conventio ns end closed systems the situation has begun. Thus an imperfective sentence presenting a durative situafioti
entails an initial endpoint. And if context and knowledge warrant, the receiver mny
Pragmatic meanings will be stated at a separate, interpretive level in the formal infer the final endpoint of the ongoing situation. The inferences about endpoints hold
account of aspect. This allows us to maintain the analysis of semantic invariance for for sentences with a neutral viewpoint, which are also open informationally. The p t•
aspectual viewpoints. Divergent interpretations of a viewpoint are explained by dif- fective typically leads to inference that the final state continues (for telic events).
ferent inferences and conventions of use. We predict the interpretations on the basis We want the information that arises through inference to appear in the formal rAp•
of conventions associated with the viewpoints and basic patterns of inference, which resentation of a sentence, yet to be distinct from semantically licensed information.
are assumed cooperatively by speakers and receivers. There are practical reasons for keeping the two kinds of information separate. Later
Viewpoint conventions depend in part on the aspectual contrasts that a language sentences may give other facts or inferences that lead to revision of an inference
makes available. If only one viewpoint is generally available, that viewpoint is prag- about a sentence. For instance, a sentence following (52) might say that John and
matically dominant in the language. In Russian the imperfective viewpoint is domi- Carolyn never finished the painting job. If one had assumed a final endpoint for (52),
nant: The imperfective is used for all situation types, while the perfective appears one would revise the assumption after interpreting the later sentence. Further, othut
only with non-statives. The conventions associated with each viewpoint are partly types of pragmatic inferences arise in sentence interpretation and should probably be
related to this contrast. The Russian imperfective has several conventional meanings, treated together in a formal representation. Thus if there is reason to infer a final end•
each with a different emphasis. The Russian perfective has fewer possibilities. This point, continuing result, or other feature for a sentence, these features will appear In
is not surprising since the perfective is the more specific and therefore more limited an augmented Discourse Representation Structure for that sentence; see Chapter 7.
of the two (in Prague School terms, the perfective is marked). The conventions are In working out the inferences available for a given sentence it is essential to ascent
discussed below, and in Chapter 10. the factor of mutual knowledge. Suppose that Bill and Sue embark on the building of
The main factors that shape the conventions of use are the place of a viewpoint in a gazebo, and they finish building it before the week’s end. I can truthfully talk about
the aspectual system and the information it conveys. Use of a viewpoint in a certain the event with either of the sentences of (53). The question is which one I should me
conventional pattern depends on context and the mutual knowledge that obtains in conversation with another person, you for instance.
between speaker and receiver. When conventions conflict, contextual information
often suggests which convention overridees the other. Precisely when one interpreta- (53) a. Bill and Sue were building a gazebo last week.
tion or another is intended and conveyed is a delicate matter; in this discussion I can b. Bill and Sue built a gazebo last week.
only identify factors that play a role in aspectual interpretation.
88 VIEW PO I NT ASPECT THE PARAM EVER OF A SPECI 89

The answer to the question depends on the information that both have about the situ- (54) Mildred was shelling the peas.
ation, and our mutual knowledge. (53a) is appropriate if you know that the gazebo
was finished, but it would be misleading for me to use it otherwise. The reason: if you This sentence presents an internal interval of the situation. Taken positively, the sen-
don’t know the outcome of the situation, my choice of the imperfective might lead tence emphasizes the fact that the event [Mildred shell the peas] was in progress. I
you to think that the Accomplishment did not reach its natural final endpoint. Thus will call this the Ongoing Event Con mention. The negative interpretation emphasizes
(53b) is more appropriate if you and I do not have mutual knowledge of the situation; instead that the sentence presents an open event, inviting the inference that the ftnal
see the collection of papers in N. Smith 1982 for a discussion of mutual knowledge. endpoint of the situation has not occurred. This is the Convention of Incompleteness.
Generally, if a closed viewpoint is warranted by circumstance, the open viewpoint With a negative emphasis (54) suggests that Mildred did not finish shelling the pear.
will be too. The speaker may choose between them if the language and the state of There is also an interpretation (of 54) which emphasizes the fact that pea-shelling
mutual knowledge allow. Choice of the perfective viewpoint gives positive emphasis has occurred. There is a well-known convention for Russian, the Statement of Fact
to the final endpoint. and may suggest its continuing result. The imperfective gives Convention, which has this force. Russian speakers oflen use it in choosing the
positive emphasis to the ongoing situation. Whether the receiver knows the circum- imperfective viewpoint to talk about an event that has already occurred. Often the
stances is crucial. If the speaker has no knowledge of the situation, use of the imper- event has a known final endpoint. (55) is a famous example:
fective viewpoint may convey that the final point did not occur. I assume truthftilness
on the part of the speaker; if only the open viewpoint is warranted by the actual situ- (55) Vojnu i mir pisal Lev Tolstoj.
ation, no choice is available to the speaker. Lev Tolstoj wroteI=s' War and Pt':act!

4.3.3 Conventions of use for aspectual viewpoints This use of the impeJective does not mislead the receiver into concluding that
Tolstoy did not complete the novel or that the speaker is describing the action In
The information conveyed by aspectual viewpoint may be interpreted positively or progress. Rather, (55) emphasizes the event itself. If the perfective viewpoint werb
negatively. Positive emphasis draws attention to the information made visible by the used the emphasis would be on the completion of the event. Other Russian conven•
viewpoint. With positive emphasis, an imperfective sentence like (52) presents the tions of interpretation for the viewpoints are discussed in Chapter 10.
fact of the ongoing house-painting. Negative emphasis, in contrast, draws attention to Another use of the imperfective is noted with some interesting French examples iti
what is not visible. With negative emphasis, (52) draws attention to the absent end- Ducrot 1979. The Imparfait is a past tense that conveys the imperfective viewpoint.
points of the house-painting event. These differences are reminiscent of the Prague There is a marked, conventional use in which the speaker presents a situation with the
School insights about contrastive meaning. The conventions that I present below Imparfait although its endpoints are mentioned, or are known pragmatically. As Oucrot
draw on the possibility of different emphasis for sentence. notes, the situation is presented as continuous throughout the interval, and interpreted
Two different discourse principles underlie most cases of pragmatic emphasis. as a characterization of the interval. (56) presents two of Ducrot’s examples:
They are the principles of minimality and of maximality.1 By the principle of mini-
mality, speakers say only as much as they need to say: this leads to an understanding (56) a. C’est incroyable, le semaine dernière il pleuvait plusiers fois.
of positive emphasis. There is also a principle of maximality. According to the max- et, cette semaine, il n’y a pas eu une goutte d’eau.
imality principle, speakers say as much as they can say. This leads to an understand- It’s amazing, last week it rainedI'°s* Se V rà 1 t im s but this week.
ing of negative emphasis. there hasn’t been a drop of water.
There are smikingly consistent uses of the aspectual viewpoints across languages.
The same conventions appear in languages of very different structures and in many b. L’année dernière je démenagais.
types of discourse. The consistency is due to the similar information that the view- Last year I moved*=r*.
points present and the contrastive nature of viewpoint choices. In this section I men-
tion several conventions that are commonly associated with the viewpoints. What is In (56a) the several rainfalls are presented as covering the entire week; in (56b) the
conveyed in a given case depends also on mutual knowledge and the structure of the event moving is presented as taking up the whole year. The force of these sentences
language. is to characterize the time intervals by the events talked about. The convention
implicitly invoked may be called the Characterizing Convention. As Duerot rioter,
Imperfective viewpoints: Sentences with an imperfective viewpoint may have a posi- this is an extension of a standard meaning: the Imparfait typically ascribes a proper•
tive or a negative cmphasis, as noted above. Consider another example of the English ty to a situation for a continuing period.
progressive: The imperfective is frequently used to present a situation as a temporal frame for
90 VIEW PO I NT ASPECT TH E PA RA METER 0 F ASPECT 91

an event. The open, ongoing situation provides a background against which the sec- Other contexts in which imperfective viewpoints are associated with the irrealis mod0
ond takes place. Otto Jespersen discussed this use lucidly for the English progressive. are discussed in Fleischman 1995.
(57) is his example: he argues that it requires a context to be fully interpreted. This is by no means an exhaustive account of the pragmatic meanings associated
with imperfective viewpoints; see the discussions in Part II for some other cases.
(57) He is hunting.
Perfective viewpoints: Sentences with the perfective viewpoint may have a positive
Jespersen explains the temporal meaning of the progressive in this sentence as fol- or a neutral emphasis. Positive emphasis for perfectives draws attention to the actual
lows: “The hunting is felt to be a kind of time frame around something else; it is rep- completion of an event. This understanding is salient in Russian, with a convention
resented as lasting some time before and possibly (or probably) also some time after of Final Emphasis. (60) illustrates, with a telic and an atelic example:
something else, which may or may not be expressly indicated, but which is always in (60) a. On vstal ran”se vsex.
the mind of the speaker ... if we say (57) we mean that the hunting (which may be He got up •" before everybody else.
completed by now) had begun, but was not completed at the time mentioned or
implied in the sentence, and this element of incompleteness is very important” b. On mne pomog vo vremja moej uceby.
(1931:179). In other words, the main event is located temporally by its occurrence in He helped e" md dtl rin g my StUdieS.
the middle of some other, protracted, process.
Imperfective sentences often seem incomplete in isolation, although they are not The perfective viewpoint also conveys neutral emphasis when the event itself is of
ungrammatical. Native speakers sometimes reject them when they are presented out interest, as in the context of a sequence of events.
of context. The sense of incompleteness is due to the fact that imperfectives give only The perfective may invite the inference that the final or resulting situation con•
partial information. (57) is not complete informationally unless the “something else” tinues. The convention of Continuing Result is common. The example illustrated with
which Jespersen invokes is explicit elsewhere, or is independently known. a Russian sentence:
Information is needed that can anchor the situation. allowing the receiver a more
complete temporal understanding. Since we cannot identify the referent of the pro- (61) a. K vam kto-to prisel.
noun, this example is incomplete in another way as well. Someone has come e for you ( An d iS S til l h re ).
Present imperfectives are anchored to the moment of speech, so that a context for
understanding them can often easily be provided by the receiver. Past and future Aspectual viewpoints are salient in Russian, and the conventions of use play en
imperfectives are more dimcult to interpret in isolation. They often appear with tem- important role in the language.
poral adverbials which provide such an anchor, as in (58): The inference of Continuing Result is commonly used with inceptives. When am
inceptive sentence has the perfective viewpoint it is plausible to infer that the situa•
(58) a. Michael was sleeping. tion continues. This is particularly natural with super-lexical morphemes such in
b. Michael was sleeping when the gong rang. begin. (62) illustrates for Navajo and English:
c. At noon. Michael was sleeping.
(62) a. bi’niilpaii.
(58 b, c) have temporal anchors and are better in isolation than (58a). More general- I finished••" starting to heat it.
ly, sentences with imperfective viewpoints are often dependent on other information b. Tania began to talk.
in the manner typical of anaphora; see Partee 1973, 1984; and Westfall 1995 on tem- c. Sonia started to walk away.
poral anchoring and the progressive, The partial information given by imperfectives,
and the fact that they are frequently used for backgrounding, conspire to produce this Continuing Result is a strong convention in Navajo. It is particularly frequent for
effect. inceptives. (62a), for instance, would be used to convey that the speaker is now in the
There is often a strong connection between the imperfective viewpoint and the process of heating it; see Chapter 12.
irrealis modality. The imperfective-irrealis connection appears in counterfactual con- Since there are con ventions for positive, negative, and neutral emphasis, it is
ditionals, for instance, as in the following French example from Fleischman 1995. often necessary to choose between competing conventions when one interprets a
given sentence. Competing conventions are discussed in connection with Russian, in
(59) Si j’avais le temps, je t’écrirais. Chapter 10.
If I had*=r' time, I would writeC••^''" to you.
92 V I MW PO t NT A SP ECT T HE PAR AM ETER OF ASPX CT 93

4.4.4 Narrative conventions arid the viewpoints In the first sentence, a perfective, we are told through the implicature of résolut
(decided) that the protagonist of the story, Félicité, will go to the nearby town of
Perfectives often function in narrative discourse to move things forward, because the Honfleur. The next sentence is imperfective, a stative that describes the highway. The
endpoints of events are presented explicitly. “The expression of a sequence of actions reader understands that the highway goes to Honfleur. and that the narrative her
is one of the most characteristic functions of perfective verbs in an extended context” moved forward to the middle of Félicité’s journey. Here the descriptive, imperfecti*0
(Forsyth 1970: 9-10). Advancing the plot is perhaps the most basic use of perfectives sentence serves to advance the narrative (Whitaker & Smith 1985: 263). There ate
in narrative. It has been documented extensively for many languages, including other ways in which the imperfective used for foregrounded information. Por
French, Navajo, Inuktit, Indonesian, Malay, Greek (see the preface and articles in instance, in some Navajo narratives the progressive is the viewpoint of choice to pm-
Hopper 1972). sent foregrounded events that are simultaneous with other events (Midgettc
In narratives, imperfective sentences may present situations that provide a back- 1987: 104-150).
ground for the main events. They present open situations, not closed events which Imperfectives may also be used contrastively to present foregrounded events.
move the narrative forward; Labov & Wiletzky 1967, Heinrich 1973, Hopper 1982. There is a well-known convention for such a use of the French imperfective past
More specifically, imperfective sentences tend to provide descriptions and other tense, the Imparfait. By this convention, known as the Imparfoit de Rupture, an
information, and they present situations that are simultaneous with the main events imperfective sentence terminates a discourse or an episode told in a perfective tense.
(Midgette 1987: 107). Backgrounded information “does not immediately and crucial- For instance, in the following example all sentences but the last have the perfective
ly contribute to the speaker’s goal, but . assists, amplifies, or comments on it” viewpoint:
(Hopper & Thompson, 1980:280). Imperfectives are said to have a backgrounding
function. (64) Elle s'arreta un instant. Elle consulta un petit carnet de notes,
However, an analysis in terms of backgrounding threatens to be too simple: dis- leva la tête, regarda et continua son trajet. Arrivée à la dernière
course does not consist of a foreground and background, deployed in an entirely rangée, elle s’engagea sur l’herbe mouille. Elle se penchait
straightforward manner. To do justice to the features of discourse, including use of sur chaque croix et lisait les noms.
aspectual viewpoints, a more complex view is necessary. Waugh & Monville-Btirston
discuss the complexity of discourse with a metaphor of density: “ ... imagine a text She stopped for a moment. She consulted a small notebook,
as a dense construction.(it has) levels of different depth: foreground of narration vs raised her head, looked around. and went on. When she arrived
background of description and commentary; level of the general vs level of the par- at the last row, she got down on the wet grass. She bent' P*
ticular; surface of the text vs. projected elements (in high or low relief) which stand toward each of the crosses and read'=s' the names.
out; basic configuration of the text in the recesses of which one finds subordinated (Arnothu; Tasmowski-De Ryck 1985:73).
elements; dominant lines of development vs secondary or even subsidiary shunted
elements” (1986: 873). This correctly suggests that imperfectives may function dif- The last sentence is in the Imparfait. In this conventional use of the imperfective
ferently in different narrative contexts. viewpoint, the contrast between aspectual viewpoints plays a significant role.
Conventions of narrative, like other conventions, can be flouted. In a narrative the I conclude with a cautionary example which shows the importance of conventions
imperfective can be used to move events forward, since it entails the beginning of the of use. If speakers do not know the relevant convention of use, they may fail to con•
situation in question. This is a marked use that occurs with some frequency. It flouts vey the meaning they intend. I give an example of such a failure for a Russian iinp¢r•
the convention that imperfectives do not move narrative time forward. The following fective sentence with an Achievement verb constellation. This type of sentencc ii
example illustrates in French, from a short story by Gustave Flaubert, t/n Coeur often indeterminate, with a single-event reading and an habitual reading. In Russian
Simple: the habitual reading is basic; convention requires adverbial support to convey the sin•
gle event reading. (65) was produced by a student of the Russian language. It war
(63) Elle résolut de le porter elle-même ä Honfleur. Les pommiers intended to present the preliminary stages of a single event, an Achievement.
sans feuilles se succédaient au bord de la route.
(65) la privykal k vaiemu klimatu. (Imperfective)
She decided to take it herself to Honfieur. There were apple trees I was in the process of adjusting to the climate here.
without leaves lining' r' the highway.
94 VIEW PO I NT ASP ECT TH E PA R AM ETER 0 F ASPECT 9D

But for Russian speakers, (65) has the only habitual reading . It does not convey that Notes
preliminaries are in progress: convention requires adverbials for such an interpreta-
tion. “The foreign student may ose (64) intending to express action in progress but ’ Chomsky explains tbe acquisition of parameters as fo11ows. “The pfincip1es of UG bave cezzain para•
eliminating all adverbial modifiers of that action. The utterance is incomplete, inade- meters, which can be fixed by experience in one or another way. We may think of tf+c language faculty as
quate for conveying the intended meaning. Some contextual element is needed a complex and intricate network of some sort associcced with a swifch box that can be in one of two po#i•
tions. Unless the switches are set one way or another, the system does not function. When they are eet in
(Rassudova 1977:141, cited in Brecht 1984). Unfortunately, the foreign student who one of the permissible ways, then the sysrem functions in accordance with its nature, but differently, depen-
constricted (65) did not follow the convention requiring contextual support in such denting on how the switches are set. The fixed network is ice system of principles of universal grammar;
cases. the switches are the parameters to be fixed by experience. Acquisition of a language is in part a process of
The web of conventions that suppon discourse of all kinds must be taken into setting the switches one way or another on the basis of the presented data. a process of fixing the values of
the parameters “ tChomsky 1988:63).
account in constructing a sentence, and interpreting aspectual meaning. Contextual 2 As I note in the Introduction. the relation between viewpoint and situation type, while familiar to schol-
and other pragmatic cues help the user to select the appropriate convention. The con- ars, has rarely been studied in a systematic manner. In several recent formal semantics treatments of
ventions are required because of the many-one relation between forms and meanings: English the progressive viewpoint is analyzed as an operator on situation type (Dowty l979,V1ach l980)i
a limited number of linguistic forms and semantic meanings are used by speakers to it is not clear whether they regard the perfective viewpoint as basic or Hso as an operator. Hoepelman 1978,
Brecht 1984, Timberlake 1982, analyze the perfective and imperfective viewpoints of Russian as opera•
convey many different pragmatic meanings.
tors. There are other similar approaches in the literature.
^ Lyons 1977 presents an inuoduction to the contrasts among viewpoints. The contrasts are sometimes
expmssed as sets oF Features in formal opposition. e.g. punctual fctosed)-Curative (open}, complete-incom•
plete, either one of which may subcaiegofize the other. and stative-non-stative tPñedrich 1974:535). In
each par r the first property is generally characteristic of the perfective viewpoint, I he second is chamcter•
istic of the imperfective. Thus the aspectual system of a language may be charactefized by the main oppo•
nitrous of its viewpoint system. Friedrich proposes that the basic opposition of Slavic is complete-incom•
plete, while thai of Homeric Greek is Curative-nondcrative. For other poinis of view on Slavic, see Forsyth
1970 and the references of Chapter 10.
* Dense time contrasts with discrete time, w hich may consists of maximally small moments. lv deny
time, no smallest unit occurs: for any iwo moments. there is in principle always another moment 1›etwun
them. Ttie notion that time is dense has been adopted by many scholars, including Taylor 11977), Eamp
( 1989). There is a procedure for conveying discrete to dense time, suggesting that the distinction betwmn
them may not be of great significance tDowty 1979. 76).
^ The Neutral viewpoint is a default in the languages studied here, because ii arises only in the absence
of an oven viewpoint morpheme. However, in principle a language might have a neutral viewpoint that
cos toasted with perfectives and imperfectives. I woold like to thank Haihua Pan for helpful discussion of

^ The French tense-viewpoint system has a present tense and a future tense. which do not convey aspec•
tual viewpoint. There is a set of past tenses which differ in viewpoint: see Chapter 9.
The Chinese neural viewpoint represents a choice when it appears with non-stative sentences; stative
require the neutral viewpoint.
8 I would like to thank Orvokki Heinamak i and Tony Woodbury for heIpfuI aiscussions of Pinnieh and
icelandic, respectively.
Por discussion of the so-called imperfective paradox see Dowty 1977. Vlach 1981. The approach pw•
sented here is an intensional one: it depends on the notion of the telic property as intensional when it holds
of a situation type. Parsons 1988 presents a similar view. Recently Dowty has argued that the imperfective
paradox can only be resolved with an intensional treatment of some kind (at a conference on Events at the
Univenity of Texas, 1988).
' Minimality and maximality are discussed in connection with the conversational principles of Grice In
Levinson 1987.

You might also like