Tense_and_aspect

768 IX.
Typology of morphological and morphosyntactic categories
proximal medial distal obvial
Portuguese DEM este esse aquele

ADV aquı́ cá aı́ alı́ (aco)lá
Quileute DEM xó{o só{o sá{a á:ça{a
Santa Ana del ADV réè rèé
Valle Zapotec
Sardinian DEM kulthu kussu kullu
(Sassarese)
Scot. English DEM this/these that/those yon
ADV here there yonder
Scot. Gaelic DEM sinn seo siad
Swahili DEM h-CL-CL h-CL-CL-o CL-le
ANAPH
Tagalog DEM ito iyan iyon
ADV dito diyan do{on
Tamil Prefix i- a-
DEM inta anta
ADV inge ange
Tolomako DEM/ADV ka(ho) tuha keni
Turkish DEM bu shu o
Wolof ART CL-i CL-a
DEM CL-ii/-ile CL-ee/-ale
ADV fi/fii/file fa/fee/fale
Yatye DEM na mi
59. Tense and aspect
1. Introduction temporal properties of situations (J Art. 42).

2. Main approaches Tense shows the temporal location of the situ-
3. Tense ation, whereas aspect indicates its temporal
4. Aspectuality in the lexicon
5. Grammatical aspect
shape or profile. Tense implies the notion of
6. Perfects (anteriors) and resultatives a reference point relative to which the situa-
7. Interaction between tense and aspect tion is located. Aspect is often concerned
8. Grammaticalization paths with the presence or absence of a bound or
9. Special abbreviations limit that ends or completes the situation in
10. References some way or another, though this is not the
only way in which the temporal shapes of sit-
uations may differ.
1. Introduction
Scholarly literature on tense and aspect is
Situation is a general name for anything that vast, and there are but few generally accepted
a sentence denotes, or an utterance refers terms. Every study of these matters, includ-
to ⫺ an event or a state, for instance. Strictly ing the present one, should therefore be seen
speaking, sentences denote situation types, as defining its own terms, explicitly or implic-
and utterances refer to situation tokens. itly. Despite these terminological problems
(Instead of “situation”, the terms “event” or there is, however, a growing body of evidence
“state-of-affairs” are also used.) Tense and suggesting that languages choose most of
aspect are grammatical expressions of the their grammaticalized temporal and aspec-
Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Authenticated | 172.16.1.226
Download Date | 3/29/12 9:29 AM
59. Tense and aspect 769
tual categories ⫺ tense and aspect grams ⫺ about, say, the perfective aspect as a cross-
from a fairly limited set of cross-linguistic linguistic phenomenon; but obviously no in-
gram types (Dahl 1985; Bybee, Perkins and teresting typological studies could be made
Pagliuca 1994), though this is not always ob- on this basis. In Comrie’s influential mono-
vious when one looks at the mere names of graph on aspect, cross-linguistic categories
these grams in the written grammars of vari- are at the centre of attention, but they are
ous languages. In what follows, a convention interpreted as semantic entities: grammatical
of writing cross-linguistic gram types with a categories of individual languages are expres-
small initial letter but capitalizing the initial sions for universal semantic categories (Com-
letters of the traditional names of grams in rie 1976: 9⫺10). Progressive Aspects of dif-
each language will be observed. It will then ferent languages thus grammatically encode
be possible to say things like “the Latin Per- the progressive meaning, and Perfects (or at
fect is not a perfect, but a perfective past least part of them) encode the cross-linguistic
tense”. perfect meaning. But Comrie’s text is actually
written in such a way that nouns such as
“progressive” and “perfect” soon begin to
2. Main approaches look like cross-linguistic grammatical proto-
types, not semantic entities as stipulated.
The traditional structuralist approach is to Comrie’s (1976: 52) claim “The perfect […]
describe tense and aspect in each language as tells us nothing directly about the situation
a system of forms and their oppositions. itself, but rather relates some state to a pre-
Each form is supposed to have a structurally ceding situation” reads as a generalization
determined, context-independent basic mean- about certain grammatical categories in dif-
ing ⫺ Roman Jakobson’s (1971a/1936) Ge- ferent languages, collectively called “the per-
samtbedeutung ⫺ which is realized through fect”, not as a description of a cross-linguistic
context-dependent “uses”, “interpretations” semantic feature such as “female” or “adult”.
or “special meanings” (Jakobson’s Sonder- Semantic categories do not “tell” or express
bedeutungen; for discussion, see Lindstedt anything, since they are the things expressed,
1985: 17⫺24). and, at any rate, the semantics of the perfect
The grammatical oppositions are com- is so complex that it would hardly be consid-
monly thought to have marked and un- ered any kind of semantic primitive at all,
marked members (J Art. 32). Thus, For- were it not for the very fact that it is often
syth’s (1970) Grammar of aspect is a detailed grammaticalized.
account of the semantic contrasts between This covert notion of a cross-linguistic
Russian Perfective and Imperfective aspect grammatical category was operationalized
forms in different contexts, with the basic as- and made explicit by Dahl (1985) and
sumption that only the Perfective positively independently by Bybee (1985). Dahl, using
“means” something, the Imperfective being typological questionnaires with sentences to
the semantically unmarked and “empty” be translated in controlled contexts, and By-
member of the opposition, used in positions bee, using reference grammars, convincingly
of neutralization. It is, however, somewhat showed that temporal and aspectual mean-
problematic to use markedness as an expla- ings which are expressed grammatically in
nation when formal marking (the presence of the world’s languages cluster around certain
a grammatical morpheme) does not run par- prototypical meanings and uses. As they state
allel with the assumed semantic marking (cf. in their later joint article (Bybee & Dahl
Dahl 1985: 19), or even runs counter to it, as 1989: 52): “In fact, our main thesis is that
is the case with most Perfective/Imperfective the meanings of grams are cross-linguistically
pairs in Russian. Moreover, since an unex- similar, making it possible to postulate a
pected use of the semantically unmarked small set of cross-linguistic gram-types, iden-
member can always be interpreted as neutral- tifiable by their semantic foci and associated
ization, only the meaning of the marked with typical means of expression” [orig. em-
member can be subject to testable hypotheses phasis]. The claim about “typical means of
in this approach. expression” means that certain meanings are
In its strictest form, structuralism con- more likely to be expressed inflectionally,
siders the oppositions and their members to others periphrastically.
be unique in each language so that there Bybee, Perkins and Pagliuca’s (1994)
would be nothing that could really be said monograph on the grammaticalization of

770 IX. Typology of morphological and morphosyntactic categories
tense, aspect and modality in the languages the present moment, the time of the speech
of the world is the next step towards the for- act itself, serves as the reference point which
mation of a post-structuralist, substantialist provides the distinction between past, present
approach to tense and aspect: and future situations. According to Reichen-
bach’s (1966/1947: 287⫺298) useful scheme,
“Like Dahl 1985, we take the universal categories
at the level of future, past, perfective, imperfective
a greater number of tenses can be described
(for example) to be the atoms of our theory and assuming that the situation (E, for “event”)
refer to them as cross-linguistic gram-types. We is located relative to a secondary reference
neither try to break their semantic foci down into point (R) which can itself be past, present or
smaller features, nor do we try to group grams into future with respect to the primary reference
higher categories such as tense, aspect, or mood. point, the point of speech (S). Since this S is
The latter represent for us cognitively significant always involved in temporal reference, di-
semantic domains, but not structurally significant rectly or indirectly, tense is a deictic category,
categories” (Bybee, Perkins and Pagliuca 1994: 3). and this property can be used to distinguish
Thus, for the structuralist, tense and aspect it from aspect (J Art. 44). However, it is not
as grammatically structured wholes are the unusual to find aspect distinctions restricted
fundamental categories, whereas for the sub- to certain temporal levels and aspect marking
stantialist approach the really significant intertwined with tense marking (see § 7 be-
categories are grams on the level of “the past low).
tense” or “the imperfective aspect”. The sub- One of the fundamental tenses in the
stantial approach has certain advantages. world’s languages is the past tense, indicating
First, it implies much stronger hypotheses that E precedes S. More often than not, its
about what is cross-linguistic. Second, it marking is inflectional, rather than peri-
makes questions like “Is the English Present phrastic (Dahl 1985: 115⫺117); the present
Perfect really a tense or an aspect?” less im- tense may be morphologically and distribu-
portant, since defining a gram as a tense or tionally unmarked. In Dahl’s (1985) 64-lan-
an aspect does not really add anything new guage sample there were, however, more past
to the description of its meaning and use. tense grams restricted to certain grammatical
And third, it liberates the linguist from the contexts than “simple past tenses” usable for
obligation of presenting an overall philo- all verbs. The past tense is often marked with
sophical or cognitive model of situations and the imperfective aspect only, or for stative
temporality every time tense and aspect verbs only, or outside narrative contexts only.
grams are discussed. There is a universal tendency to regard single,
Yet, the traditional structuralist approach dynamic and bounded events as located pro-
retains part of its value; the notion of opposi- totypically in the past.
tion especially cannot be dismissed. The per- In Dahl’s (1985: 103⫺105) sample, there
fective and imperfective aspect, for instance, were around 50 languages with a future tense,
cannot be conceived of without each other, i. e., a grammatical way of referring to future
and obviously a language with only one tense situations (E follows S). Bybee, Perkins and
gram would have no tense grams. (Of course Pagliuca’s (1994: 243) 76-language sample
one of the tenses can be formally unmarked, contained seventy languages with a gram that
and it often is.) This is why it is still useful to had reference to the future as one of its uses,
discuss tense and aspect as two grammatical and the total number of grams registered was
domains (see §§ 3⫺5 below), keeping in mind as high as 156, since most languages had
that there are numerous semantic and func- more than one gram with this function. The
tional connections between them (§§ 6⫺7). future seems to be the most common for-
mally marked tense in the languages of the
world. This is in an apparent contradiction
3. Tense with the fact that the typologically (or con-
ceptually) basic distinction is often said to be
Tense shows the location of the situation in “past vs. non-past”, not “future vs. non-fu-
time; in other words, it is the grammatical ture”. The reason may be that future time is
expression of temporal reference. No lan- modally tinged, being by its very nature epis-
guage is known to encode reference to a fixed temologically less certain than the present or
calendar time grammatically, so the temporal the past, and this uncertainty often requires
location is always indicated with respect to a formal marking even in languages that do
moving reference point. In the simplest case, not consistently mark the past. Hausa, for

instance, only grammaticalizes aspect, not away’ and ‘not more than one day away’ (ho-
tense, on the present and past time planes, diernal vs. pre-hodiernal), but Bybee et al.
but it has two grams referring to the future find a greater number of less precise distinc-
(Brauner & Ashiwaju: 1965: 68⫺70; Kraft & tions between immediate and remote tenses.
Kirk-Greene 1973: 81⫺83). Usually there are only two or three tense
Futures are more likely to be marked peri- grams that enter a remoteness opposition,
phrastically than past tenses are, and in this such distinctions being more common in the
sense they are less grammaticalized as re- past than in the future, and more common in
gards their form. Future markers are often non-narrative past contexts than in narrative
excluded from subordinate temporal or con- past contexts.
ditional clauses. There are languages (such as All continents provide examples of lan-
Finnish) that do not possess a gram that guages with remoteness distinctions, though
could be labelled a future, and languages the phenomenon is rare in European lan-
whose Futures are so infrequent as to make guages. In Catalan, Occitan and some vari-
their identification as a member of this cross- eties of Spanish the periphrastic perfect,
linguistic gram type uncertain (Dahl 1985: rather than the simple past tense, is preferred
108⫺109; 2000a: 325⫺26). when referring to situations that have oc-
The other side of the modality of the fu- cured the same day, even in connected narra-
ture tense is that those events in future time tive (Dahl 1985: 125; Schwenter 1994; Bybee,
which are somehow bound to occur are often Perkins & Pagliuca 1994: 101⫺102). This is,
expressed by the present tense: it is uncom- however, only one expression of the “current
mon to find a marked future in sentences like relevance” meaning of the perfect (§ 6) rather
Tomorrow it is Sunday. A gram expressing than a true hodiernal tense.
scheduled future of this kind is reported in In- The distinction between narrative and non-
uit by Bybee, Perkins & Pagliuca (1994: 250; narrative contexts has already been men-
their term is “expected future”), but it is the tioned as relevant to the use of tense and
only one of its kind in their database. When aspect grams. According to Dahl’s (1985: 11)
there are aspect distinctions in the present definition, a narrative discourse is “one
tense, it is often the imperfective or pro- where the speaker relates a series of real or
gressive present (see § 5) which is used for the fictive events in the order they are supposed
scheduled future. to have taken place”. Weinrich (1964) went
The problem of whether the future should so far as to claim that the distinction between
be classified as a tense or as a mood is of narrative and non-narrative forms is more
course not essential to the substantialist ap- fundamental even in well-known European
proach, but it should be noted that Dahl languages than temporal reference; he claims
(1985: 105⫺107) finds that the temporal that the basic function of certain simple past
factor in most future tenses is more dominant tenses is to be used in narratives, not to refer
than the modal factors. The modal character to the past. His monograph is still a useful
of the future does, however, account for the corrective against attempts to describe tenses
fact that a language may possess several fu- in constructed isolated sentences, but from a
ture grams. They often originate as expres- typological point of view his thesis is an exag-
sions of intention or obligation (Bybee, Per- geration, since certain African languages do
kins & Pagliuca 1994: 254⫺266), and the dis- possess separate narrative tenses (Dahl 1985:
tinction between “events that are going to 113⫺114) that do not necessarily occur as the
happen because of someone’s intention” and first predicate of the story. Typically they
“events that are simply predicted to happen” seem to be subordinate verb forms that have
(such as changes in the weather) may be re- been generalized to main clauses after the
flected in, say, the choice of the future auxil- first, stage-setting past tense form. Hausa
iary (Dahl 2000a: 309⫺13). uses the same “Relative Aspect” verb forms
A language may possess several future and in relative clauses as in narrative main clauses
past tenses expressing different remoteness after the first non-Relative verb form or a
degrees, i. e., the temporal distance between narrative-introducing adverbial; a non-Rela-
E and S (Dahl 1985: 120⫺128; Comrie 1985: tive Completive (⫽ perfective) Aspect form
83⫺101, 1994: 4561⫺4562; Bybee, Perkins & in a narrative is interpreted as a pluperfect-
Pagliuca 1994: 98⫺104, 246⫺247). Accord- like flashback, i. e., a past anterior (Brau-
ing to Dahl, the most common remoteness ner & Ashiwaju 1965: 48⫺49; Kraft & Kirk-
distinction is between ‘more than one day Greene 1979: 109, 171⫺172). (It should be

pointed out that narrative tenses are different R is located before the point of speech S. No-
from the reportative evidentials found in tice that (1) does not directly tell us what the
many languages and sometimes erroneously relative order of E and S is.
called “narratives”; these will be briefly dis- If the present perfects of various languages
cussed in § 6 below.) are analysed as absolute-relative tenses, the
All tenses that locate E directly with re- general scheme of their temporal reference
spect to the primary reference point S are is: “E precedes R, R coincides with S”. This
called absolute tenses; logically they are three is Reichenbach’s (1966: 289⫺290) original
⫺ past, present, future ⫺ but remoteness and analysis, but it has been criticized by Comrie
narrativity distinctions may increase their (1985: 78) who seems to find the idea of a
number. Relative tenses locate E with respect distinct R that nevertheless coincides with S
to a contextually given secondary reference meaningless. It is, however, one way of cap-
point R (Comrie 1985: 56⫺64, 1994: 4560). turing the intuition that She has written a let-
At least in European languages, verb forms ter seems to tell something either about the
with relative time reference are typically non- present or relevant to the present, whereas
finite (not inflected for person). An example She wrote a letter only relates to a past world.
of a relative present tense is the Latin Present In all tenses, R must be understood to repre-
Participle Active laudāns ‘praising’: it shows sent ‘the time the discourse is about’. But
that the praising E is contemporaneous with there will be more to say about various kinds
a moment R given by the context, but it does of perfects in § 6 below.
not show anything about the location of E
with respect to the point of speech S. The
Latin Future Participle Active laudatūrus 4. Aspectuality in the lexicon
‘about to praise’ shows a relative future tense,
the Past Participle Passive laudātus ‘praised’ In linguistic literature, the terms “tense” and
is a relative past tense. Not only participles, “aspect” are often used asymmetrically.
but also various adverbial verb forms (ger- “Tense” only refers to the grammatical ex-
unds, converbs) typically have relative time pression of temporal reference, not to such
reference. lexical expressions as adverbs and adverbial
Sometimes relative time reference of this phrases of the type now or in the distant past.
kind is called taxis (Jakobson 1971b: 135, In contrast, it is not unusual to see the term
140⫺142; Bondarko et al. 1987: 234⫺319). “aspect” to be used about features inherent
In some languages it can be subsumed within in the lexical meaning of the verb ⫺ about
the functions of aspect: imperfective forms the fact, for instance, that the English to blink
may express relative present tense, perfective denotes a punctual event, but to swim a more
forms relative past tense. prolonged process. One reason may be that
Forms like the English Pluperfect had writ- such lexical aspect, aspectual character or in-
ten can be called absolute-relative tenses (tra- herent aspectual meaning of the verb always
ditionally they have also been called “relative interacts with the grammatical aspect. The
tenses”, a practice rightly criticized by Com- interpretation of the Progressive is not the
rie 1985: 65, 1994: 4561). The English Pluper- same in She is swimming (a process) as in She
fect (Past Perfect) shows (in its non-modal is blinking (a series of punctual events). An-
uses) that the situation E is located prior to other reason is a gap in terminology: if tense
a contextually given reference point R that
is said to express temporal reference gram-
precedes the point of speech S. Its counter-
matically, there is no obvious term for what
parts on other temporal planes are the Pres-
aspect expresses, the more so as there are var-
ent Perfect has written and the Future Perfect
will have written; in the grammars of some ious competing theories about the semantic
languages, the future perfect is also known as essence of aspect. A wider use of the term
the Second Future or futurum exactum. An aspectuality, modelled on “mood : modality”,
example of a different kind of absolute-rela- would clarify the situation (J Art. 42). In
tive tense is the form would return in sentence Russian linguistic literature, the correspond-
(1) (Comrie 1994: 4561): ing term aspektual’nost’ is already common-
place to express the whole semantic field.
(1) Mary left at six o’clock; she would re- Between the inherent aspectual meaning of
turn an hour later. verb lexemes and grammatical aspect there is
Here, Mary’s return E is located after the ref- the phenomenon of derivational aspect, also
erence point R provided by the first sentence; called Aktionsart or actionality (if these two

terms are not simply used as a synonym for Progressive to show imminence (The train is
lexical aspect). Aktionsart was first clearly leaving).
distinguished from (grammatical) aspect by Various other classifications of aspectual
Sigurd Agrell (1908). Many languages possess character have been proposed, but they are
various momentary, habitual and other deriv- usually based on Vendler’s system, whose
ative affixes that change the temporal profile terms remain the best known in the field. It
of the basic verb in some way or another. The is, however, important to avoid two potential
so-called bounders, affixal elements or par- misunderstandings pertaining to them. First,
ticles that express the completion of an ac- although those beginning with an a- seem
tion, such as the English up in to eat up, are to denote actions typically performed by
a related phenomenon (Bybee et al. 1994: agentive beings, there is nothing in Vendler’s
87⫺88). The distinction between derivational definitions that would exclude non-agentive
and grammatical aspect is one of degree and situations: ‘raining’ qualifies as an activity,
generality only; obviously bounders can be ‘exploding’ as an achievement. Some would
grammaticalized as markers of the comple- actually prefer the term process instead of
tive/perfective aspect, and habituals can be- “activity”, and punctual event instead of
come imperfectives, as has happened in “achievement”. Second ⫺ and this may be
Slavic languages (Maslov 1984: 102⫺110). more seriously misleading ⫺ although Vend-
Grammatical aspect may also be morpho- ler’s terms seem to refer to different kinds of
logically parasitic upon derivational aspect. situation, they only classify linguistic entities,
This is the case with the Slavic Perfective As- viz. uninflected verb phrases. Consider (2):
pect, which is expressed with Aktionsart pre- (2) (a) She was running a mile.
fixes; there is no general Perfective marker (b) She ran a mile in ten minutes.
applicable to all or most verbs. The Bulgarian
Imperfective verb varjá ‘boil, cook’ thus pos- It would make sense to say about the situa-
sesses more than twenty prefixal derivatives, tion types denoted by these sentences that
which are all of the Perfective Aspect. Some (2a) describes an activity, whereas (2b)
are clearly lexically distinct verbs, others may describes an accomplishment. However, in
Vendler’s system run a mile is an accomplish-
be considered Aktionsarten of varjá, and one
ment irrespective of its tense and aspect; the
or two can be considered its true Perfective
English Progressive Aspect only modifies its
counterparts, viz. svarjá and the rarer uvarjá
basic meaning, but does not change its class
(Lindstedt 1985: 159⫺160). Their prefixes are
or “time schema”. Much confusion arises if
said to be “empty”, i. e., adding only an as-
these two types of classification are not dis-
pectual (general perfective), not an Aktions- tinguished.
art meaning. With other verbs the prefixes s- In one important case this distinction be-
and u- still retain their original Aktionsart tween situation types and uninflected verbs/
function. verb phrases is better supported by post-
Outside the Slavic languages, most classifi- Vendlerian terminology. This is the distinc-
cations of the aspectual character (lexical as- tion between boundedness and telicity (for a
pect) of verbs are based on Vendler’s classic good overview and further bibliography, see
study (1967; originally 1957), Vendler distin- Depraetere 1995). Vendler’s accomplishments
guished four different “time schemata” pre- are telic, i. e., they contain a reference to an
supposed by verbs. English verbs, or verb ⫹ “inherent or intended endpoint” (Depraetere
object combinations, that can be used in the 1995: 2). Thus, both (2 a) and (2 b) are telic,
Progressive Aspect denote either activities, the endpoint being the end of a run of one
such as ‘running’, or accomplishments, such mile. Activities and states are atelic: ‘running’
as ‘running a mile’ , ‘drawing a circle’. Activi- (in general), ‘raining’, ‘knowing’ and ‘loving’
ties take adverbials of the type ‘for one are not directed to any endpoint, and even
hour’ (She ran for one hour), whereas accom- though physical limitations force them to end
plishments typically take adverbials of the ‘in sooner or later, they are never “completed”
one hour’ type (She drew a circle in one min- in any sense. By contrast, boundedness refers
ute). Verbs usually lacking Progressive forms to the actual, not potential achievement of
denote either durative states, such as ‘know- the endpoint: (2 b) is bounded, but (2 a) is un-
ing’ or ‘loving’, or momentary achievements, bounded. In Russian aspectology, telicity is
such as ‘starting’, or ‘reaching the summit’. known as predel’nost’ ‘limitedness’, a prop-
Achievements can, however, be used in the erty that unites both Vendler’s accomplish-

ments and achievements. In the Western tra- tive) when an event happened (perfective). If a
dition, achievements are not always clearly form with the first of these functions excludes
classified in this respect, but it would seem to stative predicates (Vendler’s states), as the
be logical to classify them as telic, too: being English form was happening does, it is not a
essentially punctual, they are not only di- true imperfective, but a progressive. Thus, an
rected to their endpoint, but also contain imperfective is also used as the first predicate
nothing but the endpoint. in a sentence of the type “I knew what would
As will be seen in the next section, bound- happen next”, but a progressive is not. In
edness is the most central notion grammati- Dahl’s (1985: 90⫺95) sample, the progressive
calized as aspect in the world’s languages. was nearly as frequent as the imperfective,
Aktionsarten and other types of verb deriva- and it is not unusual to find both a pro-
tion, such as transitivizing affixes, often have gressive and an imperfective in the same
consequences for telicity, but being closer to language, such as Spanish, Portuguese and
lexical semantics, telicity is less apt to be Italian (for the progressive in European lan-
grammaticalized as an inflectional category. guages, see Bertinetto, Ebert & de Groot
2000). In contrast to the imperfective, the
progressive is usually not confined to the past
5. Grammatical aspect tense, nor is it used with a habitual meaning,
and it is clearly more often expressed peri-
The most important aspect distinction in the phrastically (often with locative expressions)
languages of the world is between the perfec- than the imperfective is. When the same lan-
tive and the imperfective aspect; Dahl (1985: guage has both an imperfective and pro-
70⫺71) found it in more than half of his 64- gressive aspect, I propose the tentative uni-
language sample. Some languages express the versal that the progressive is also less gram-
distinction in all verb forms, such as in maticalized in that its use is less compulsory
Russian stroit’ ‘to build (Imperfective)’ vs. in the appropriate contexts: the imperfective
postroit’ ‘to build (Perfective)’: in Russian, no may suffice to express the progressive mean-
tenses, moods or non-finite forms can neu- ing, but not vice versa.
tralize this opposition, with the exception Most theories about the basic meaning dis-
that what is morphologically a Perfective tinction between the perfective and imperfec-
Present is usually interpreted as having future tive aspect are based on the concepts of
time reference (the Imperfective does possess boundedness and totality. According to Com-
separate Present and Future tenses). Modern rie (1976: 3), the perfective “presents the to-
Greek is a language where there are separate tality of the situation referred to […] without
perfective (“Aorist”) and imperfective forms reference to its temporal constituency: the
for the past, present and future. But in many whole of the situation is presented as a single
languages this opposition is confined to the unanalysable whole” (cf. Bondarko 1990). In
past temporal domain, as in Latin and mod- a sense, a perfective form like venit ‘came, ar-
ern Romance languages: rived’ in (3) above can be compared to a
(3) Latin count noun in that a complete arrival must
Cena-ba-m forte, contain all of its parts. An imperfective form
have.dinner-past.impfv-sg1 by.chance like cenabam ‘was eating, was having dinner’
cum amic-us veni-t. is like a mass noun since even a short stretch
when friend-sg.nom come.past.pfv-sg3 of having a meal is a perfect instance of this
‘I happened to be having dinner type of activity (Carlson 1981 and forthcom-
when a/the friend came.’ ing). Notice that this is not directly related to
the duration of the situation, but to its in-
In Latin, the perfective past is traditionally ternal homogeneity. A perfective situation,
called the Perfect, and the imperfective past i. e., a situation referred to perfectively, is not
the Imperfect. “Imperfect” can indeed be necessarily more punctual than an imperfec-
used as a cross-linguistic name for imperfective situation is, but it is similar to a point in
tive past tenses, but “Perfect” is usually re- being indivisible. Moreover, an imperfective
served for other uses (see § 6 below). form is often used to describe the state of af-
Sentence (3) also illustrates a minimal nar- fairs at a definite moment, such as the pres-
rative context which is cross-linguistically ent moment, or the moment of the friend’s
diagnostic for the perfective / imperfective arrival in (3), whereas an perfective accom-
opposition: what was happening (imperfec- plishment (such as She built a house) neces-

sarily refers to a prolonged interval. It is Languages that express the temporal bound
therefore erroneous to label the imperfective use the perfective here (notice also the Simple
“durative” and the perfective “punctual”, as Past instead of the Progressive in English),
is often done. whereas Russian and other Slavic languages
It is also said that the perfective refers to would make use of their imperfectives since
a bounded situation, a situation that has dancing is atelic, and no material bound can
reached some kind of limit or endpoint. Since exist.
the inclusion of the limit is the most impor- In Bulgarian sentences referring to the
tant thing that makes a situation that is re- past, there are two independent aspect oppo-
ferred to perfectively indivisible, obviously sitions, onne grammaticalizing the material
this is not in contradiction with the totality bound (as in Russian), the other the temporal
definition. Perfective grams of various lan- bound (as in Romance languages). In cases
guages may, however, differ according to the like the following the uses of the two perfec-
type of bound they express. Lindstedt (1995) tives coincide; the verb form used is the so-
has suggested calling the two main types a called Perfective Aorist, a “perfective perfec-
temporal bound and a material bound. A ma- tive-past”:
terial bound presupposes telicity and entails
(8) Tja iz-pjá pesen-tá za
a temporal bound, such as in the following
she pfv-sing.aor(sg3) song-def.f prep
Russian sentence containing a Perfective
tri minút-i.
verb:
three minute-pl
(4) On na-pisá-l pis’m-ó. ‘She sang the song in three minutes.’
he pfv-write-past(m) letter-sg.acc However, when the bound is only temporal,
‘He wrote a/the letter.’ we have an Imperfective Aorist, an “imper-
The endpoint of the action has a clear mate- fective perfective-past”:
rial character: when the letter was finished, (9) Tja pja pesen-tá
the work was done. she (impfv)sing.aor(sg3) song-def.f
Now consider the Russian translation of tri minút-i
the following sentence from Dahl’s (1985: three minute-pl
74⫺75) typological questionnaire: ‘She sang the song for three minutes.’
(5) (What did your brother do after din- Notice in the translations that English only
ner yesterday?) makes use of different temporal prepositions
On pisá-l pı́s’m-a. to express the nature of the bound.
he (impfv)write-past(m) letter-pl.acc Many languages use bounders (see § 5) or
‘He wrote (some) letters.’ other kinds of morphemes ⫺ such as auxilia-
ries with the original meaning ‘finish’ ⫺ to
Russian must use an Imperfective verb here, indicate that an action totally affects or con-
because the situation is atelic: the brother sumes the object, or that it involves the whole
spent some time in letter-writing, but he is set of individuals expressed by a plural sub-
not reported to have had a definite number ject or object, or sometimes that the action
of letters to be completed. A Perfective verb has some surprise value (cf. Bybee, Perkins &
(on napisál pı́s’ma) would force the transla- Pagliuca 1994: 57⫺61; Dahl 1985: 95, 1994:
tion ‘He wrote the letters’. However, in 243; Bakker & al. 1994: 257⫺258). This kind
translating (5) most aspect languages in of completive aspect is best regarded as a rela-
Dahl’s sample make use of the perfective ⫺ tively weakly grammaticalized expression for
the Romance languages, for instance, would the material bound. Completive prefixes are
use their perfective past tenses. This is be- the diachronic source of the Slavic Perfective
cause their perfective aspect is based on the (cf. § 5), and many neighbouring languages,
notion of temporal bound: the brother spent such as Georgian, Lithuanian, Latvian, Ger-
some time in letter-writing and then did man and Hungarian possess systems of ver-
something else. bal prefixes and other bounders with various
Another type of test sentence for the tem- degrees of grammaticalization. In Finnic
poral bound is: (Balto-Finnic) languages such as Finnish, the
(6) Mary danced until midnight. nominal cases distinguishing a partially-af-
fected object from a totally affected object
(7) Mary danced for two hours. behave in some contexts like markers of

imperfectivity and perfectivity (Heinämäki tachment from other past facts, i. e., non-
1984, 1994). There is thus an area in Central narrativity (Comrie 1976: 52⫺65; Dahl 1985:
and Eastern Europe where the material 129⫺153; Maslov 1984: 32⫺47, 1988). The
bound is often marked, whereas the Western notion of current relevance particularly is
(Romance) system marks the temporal considered crucial (for its possible explica-
bound, and typically in the past only. tion, see Dahl & Hedin 2000). The English
As regards discourse structure, especially Present Perfect (as in He has read this book,
narrative discourse, it is possible to say that or He’s read this one) is a good example of
the perfective or completive aspect denotes this cross-linguistic gram type.
an event, i. e., a foregrounded situation that In recent literature the perfect is also called
moves the plot forward and changes the anterior so as to avoid confusion with the
state-of-affairs in the discourse world, where- term “perfective (aspect)”. It is indeed some-
as the imperfective aspect denotes a process what misleading that the Perfects of some
or state and therefore describes a background languages, such as Latin, can be cross-lin-
state-of-affairs in that world. The progressive guistically called perfective past tenses, but
only denotes processes, not states. This three- not perfects, the crucial difference being the
fold classification into events, processes and use of the Latin perfect as a main narrative
states necessarily pertains to the situations or tense. Thus, Bybee, Perkins & Pagliuca’s
situation types themselves, not to linguistic (1994: 54) definition of the anterior is essen-
entities à la Vendler (cf. Mourelatos 1981).
tially the same as that of the perfect discussed
A special kind of state is a situation that
here. Their definition is, however, broader in
is habitually repeated, such as:
that it also provides for past anteriors and
(10) The two children used to send me a future anteriors: “an anterior signals that the
card at Christmas time. situation occurs prior to reference time and
Imperfectives, but not progressives, are often is relevant to the situation at reference time
employed to express such habitual situations […] Anteriors may occur with past or future
(for the Slavic Imperfective and Perfective tense marking”. Past perfects (or pluperfects)
in habitual function, see Mønnesland 1984). and future perfects (futura exacta) share sev-
The English auxiliary used to [1ju:ste] is an eral properties with present perfects, but they
example of what can be called the habitual can be described as absolute-relative tenses
aspect (Comrie 1976: 26⫺30; Dahl 1985: more easily than the present perfects can (see
95⫺102; Bybee, Perkins & Pagliuca 1994: § 3 above). Notice also that in creole linguis-
151⫺160; cf. also Kučera 1981 for English tics, “anterior” usually refers to a purely rela-
and Czech). As the English used to or the so- tive past tense (Bakker & al. 1994: 250).
called Past Iterative in Lithuanian, the habit- Bybee and Dahl (1989: 67⫺68) list four
ual is often restricted to the past. In Bybee, typical diachronic sources of the perfect in
Perkins and Pagliuca’s cross-linguistic sam- the languages of the world: (i) copula ⫹ past
ple, there were 21 grams expressing habitual- participle of the main verb; (ii) possessive
ity without tense restrictions, and 10 grams constructions involving a past participle of
with the meaning ‘past habitual’. There were the main verb (cf. Maslov 1984: 224⫺248);
two cases of specifically present habituals, (iii) main verb ⫹ particle meaning ‘already’;
but both were morphologically unmarked.
(iv) constructions involving verbs like ‘finish’
Imagine that the English Progressive Present
or ‘cast aside’. The two latter sources are
(is reading) were generalized as a non-ha-
bitual present: the English Simple Present by their semantics completive constructions
(reads) would then emerge as such a morpho- (§ 5), whereas the first two, common in Euro-
logically unmarked present habitual tense. As pean languages, are originally resultatives
shown by Haspelmath (1998), in various lan- (Bybee, Perkins & Pagliuca 1994: 53⫺74). A
guages a new morphologically marked pres- perfect deriving from a possessive construc-
ent tense has pushed the old unmarked pres- tion may involve a transitive verb meaning
ent into future, habitual or subjunctive uses. ‘to have’; if this is the case, it can be called a
‘have’ perfect, or a habeo perfect. As transi-
tive verbs with the meaning ‘have’ are rare
6. Perfects (anteriors) and resultatives
outside European languages, the ‘have’ per-
The perfects of various languages typically fect is a typically European phenomenon. A
express (1) the relevance of a past situation copula-based perfect is a ‘be’ perfect, or a
from the present point of view and (2) de- sum perfect.

The distinction between resultatives and 1994: 62). In a broader definition, it only
perfects was established in linguistics only re- means that “a given situation has held at
cently, largely owing to the important collec- least once during some time in the past lead-
tive work edited by Nedjalkov (1988/1983). ing up to the present” (Comrie 1976: 58).
Resultatives “signal that a state exists as a This notion is further explicated by Dahl &
result of a past action” (Bybee, Perkins & Hedin (2000), who call it “type-focusing
Pagliuca 1994: 54). Traditionally the resulta- event reference” (as opposed to “token-fo-
tive is often subsumed under the category of cusing event reference”, as in Mary went to
perfect as a special resultative or “statal” Paris).
variant of it, or is simply called a “stative”. However, experientiality as such is neither
For the criteria for distinguishing resultatives a necessary nor a sufficient condition for a
from perfects, the reader is referred to Ned- gram to be called a perfect. Experientials
jalkov & Jaxontov (1988), Dahl (1985: proper are a distinct gram-type, described by
133⫺135), Bybee & Dahl (1989: 68) and By- Dahl (1985: 139⫺144); the Japanese -ta koto
bee, Perkins & Pagliuca (1994: 63⫺69), but ga aru construction is a well-known example
the most important single difference should (see also Dahl & Hedin 2000: 388).
be mentioned here: only resultatives combine Although the experiential meaning may
with adverbs of unlimited duration, such as become dominant in the perfect, historically
‘still’ or ‘as before’. In English, it is not pos- it is usually secondary and derives from the
sible to say *She has still gone (if still is used current relevance meaning. The perfect of a
in its temporal meaning) ⫺ in contrast to the particular language may well be compatible
resultative construction She is still gone (see with specific past time adverbials. Sentence
also Lindstedt 2000: 366⫺68). (13) is ungrammatical in English, to be sure:
The perfect is typically a periphrastic
(13) *I have woken up at 4 o’clock this
gram, being formally close to its resultative
morning.
(or completive) source. An important excep-
tion seems to be the old Indo-European Per- However, a perfect would be possible ⫺
fect, as attested in Ancient Greek and Old In- though not the only alternative ⫺ in Finnish
dic. A newer inflectional perfect in statu nas- and Bulgarian, for instance. This is because
cendi is the active resultative construction in there is an obvious current relevance reading
North Russian dialects (Trubinskij 1988: 394; ⫺ I woke up so early that I am now tired.
Tommola 2000: 465). According to Dahl (1985: 137⫺138), Swedish
The perfect of current relevance semanti- occupies an intermediate position: a specific
cally shades into what is usually called the time adverbial can combine with the perfect
experiential (or existential) perfect. In Eng- if it is part of the information focus. I assume
lish, these two types are formally differenti- that the degree of incompatibility of specific
ated only in rare cases like the following (cf. time adverbials with the perfect in a particu-
Comrie 1976: 58⫺59): lar language shows to what extent it has be-
come a dominantly experiential form.
(11) Mary has gone to Paris. In some languages the perfect has devel-
(12) Mary has been to Paris. oped evidential functions, or has become a
predominantly evidential gram. Evidential,
In (11), the fact of Mary’s having gone to to be distinguished from mood, is, according
Paris may be relevant to the present state of to Trask (1993 s. v.), “a grammatical category
affairs in various ways, but typically the sen- occurring in some languages by which all
tence at least implicates that she is not pres- statements (and sometimes other sentence
ent. The experiential perfect of (12) only ex- types) are overtly and obligatorily marked to
presses that the past situation in which Mary indicate the source of the speaker’s evidence
went to Paris is less directly part of the pres- for her/his utterance” (see also Jakobson
ent state of affairs, most notably through 1971b: 135; Chafe & Nichols (eds.) 1986; Wil-
Mary, who perhaps now knows what Paris lett 1988). There are no well-established terms
is like. for different types of evidentials. I propose
In its narrower definition, an experiential the term indirective for the most widespread
perfect presupposes an animate agent, since evidential gram type, which indicates that the
it expresses the fact that “certain qualities or speaker has not witnessed the situation he or
knowledge is attributable to the agent due to she is speaking about, but knows of it from
past experiences” (Bybee, Perkins & Pagliuca hearsay or other kinds of indirect evidence.

Other grammatical and semantic terms in use (15) Yesterday at two o’clock she was
are reportative (also called, somewhat mis- building a sauna.
leadingly, “quotative”), which should be con-
fined to the hearsay case alone, and inferen- In contrast, bounded situations (events) can
tial, referring to statements made on the basis be associated with one time point only if they
of inference, not hearsay. The Scandinavian are punctual (achievements):
perfect, for instance, has inferential functions (16) (a) ?? Yesterday at two o’clock she built
(Haugen 1972; Kinnander 1973; cf. also a sauna.
Weinrich 1964: 84⫺86 for German), though (b) Yesterday at two o’clock the sauna
it has not developed into a real indirective exploded.
evidential, as have the ‘be’ perfects of Bulgar-
ian (Lindstedt 1985: 259⫺276, 2000) and Since the present is also a point-in-time, it is
Macedonian (Friedman 1976, 1977; Graves logically impossible to have aspectually
2000). If a policeman investigates a burglary bounded forms referring to a single present
and sees footprints beneath a window, she or situation. In various languages perfective
he can say in Swedish: presents may be interpreted, for instance, as
(14) Tjuv-en ha-r komm-it in futures, or present habituals (since a habitual
thief-def.ut have-pres come-sup in series of bounded situations is in itself a
genom det här fönstr-et state-like unbounded situation; cf. Every
through def.nt here window-def.nt summer she builds a sauna). Perhaps the only
‘The thief must have entered (lit. has types of true present perfectives are perfor-
entered) the house by this window.’ matives (I promise to come) and commentar-
ies describing sports or other public events in
In Europe, there are two major areas where approximate real time.
grammatical evidentiality distinctions are If a language has a perfective/imperfective
common. One is the Baltic region, compris- opposition, but no past/non-past opposition,
ing the Baltic languages Lithuanian and Lat- perfectives are interpreted as referring to the
vian and the Finnic languages Estonian and past, imperfectives to the present, if the
Livonian. The other area can be called the context does not indicate any other temporal
Black Sea area, as it consists of languages
relation. It is also usual to find the perfective/
around this sea, though it stretches to
imperfective opposition confined to the past
Central Asia (Haarmann 1970; Dahl 1985:
tense. As pointed out by Dahl (1985: 83),
149⫺153; Friedman 1986; Johanson 1992:
there are actually two typological parame-
244⫺246). The indirectives of this area are
generally regarded as having arisen from ters involved:
Turkish influence during the Ottoman reign. (i) whether the perfective aspect is restricted
However, the tendency to develop indi- to past time reference;
rectives from various sources is typical not (ii) whether the use of a distinct past tense
only of Turkish, but of the whole Turkic marker is restricted to the imperfective
stock, and the area also extends from the aspect.
Black Sea to regions where such Uralic lan-
guages as Komi and Udmurt, farther north, These two parameters tend to correlate; Dahl
are spoken (Leinonen & Vilkuna 2000). assumes that in most cases they both receive
Most European indirectives are past parti- either a positive or a negative value. The
ciples used as finite predicates, or periphras- Slavic languages have a negative value for
tic forms involving past participles, but not both parameters, whereas Classical Arabic
all of them can be shown to have had a per- (as analysed by Comrie 1976: 78⫺81 and
fect as their diachronic source (cf. Ikola 1953 Dahl 1985: 82⫺83) is positively marked for
for Estonian and Livonian, and Schmalstieg both: the Arabic Perfective (also called “Per-
1988: 113⫺121 for Lithuanian). fect”) refers to past bounded situations, and
the Imperfective (also called “Imperfect”) re-
7. Interaction between tense and fers to the present without further context,
but with contextual indications or appropri-
aspect
ate auxiliaries it can also refer to past un-
It was pointed out in § 5 above that an imper- bounded situations and to any future situa-
fective or progressive gram can be used to tion (bounded or unbounded).
describe the state of affairs at a single point- However, as pointed out by Dahl, there
in-time: are languages which are positively marked

for the first of these parameters only. The position confined to the simple past tense
Classical Greek Aorist is an example of a per- (both in the Confirmative and Indirective evi-
fective aspect that is confined to the past dential series). As illustrated by examples (8)
tense only, at least in the Indicative Mood; and (9) above, the latter opposition is based
yet both the Aorist (perfective past) and Im- on the notion of temporal bound, whereas
perfect (imperfective past) require a past- the “Slavic-style” opposition is based on the
tense marker (the so-called augment). It is material bound. When both aspect opposi-
interesting to note that Modern Greek has tions are marked suffixally, the “Slavic-style”
generalized the aspectual opposition to other opposition, called Perfective/Imperfective, is
temporal domains and is now aligned with marked closer to the root than the past-tense
the Slavic patterns (Mackridge 1985: 104⫺ aspect opposition, called Aorist/Imperfect. In
116). Dahl (1985: 83) considers it more diffi- this example the verb is pomágam/pomógna
cult to find a clear example of a language ‘to help’:
with the opposite situation, i. e., a language
whose perfective aspect is not restricted to Imperfective Perfective
past time reference but which only uses a dis- Imperfect sg.1 pomág-a-x pomóg-n-e-x
tinct past tense marker with the imperfective. Imperfect sg.3 pomág-a-še pomóg-n-e-še
If such languages are indeed rare or non-exis- Aorist sg.1 pomag-á-x pomóg-n-a-x
tent, an implicational universal might be that Aorist sg.3 pomag-á pomóg-n-a
a morphological past tense marking confined The distinction between Aorist and Imperfect
to the imperfective aspect implies a necessary is partly marked by the theme vowel (-a-/-e-),
past tense reference for the perfective or, per- partly by different personal suffixes, partly by
haps more succinctly, that if a language has a accentual differences. As illustrated in (9), the
perfective/imperfective opposition, it always Imperfective Aorist is used when a situation
possesses grammatical means of unequivo- is bounded temporally, but not materially
cally referring to the past in both aspects. (Lindstedt 1985: 175⫺184). Since a material
When an aspect opposition is confined to bound always entails a temporal bound,
the past, it is usual to find joint marking for there does not seem to be a place for the Per-
aspect and tense. Both Latin and modern Ro- fective Imperfect. It is, however, used in cer-
mance languages have a perfective past tense tain subordinate clauses to denote an un-
(such as the Latin Perfect, or the passé simple bounded habitual series of events each of
in written French and the passé composé in which is by itself materially bounded (Lind-
spoken French) and an imperfective past stedt 1985: 189⫺198).
tense (usually called the Imperfect), but no The aspectual semantics of a sentence can
separate markers for tense and aspect can be be described as consisting of different layers.
morphologically segmented. The aspectual character of the predicate lex-
In layered marking, markers of aspect are eme is the innermost layer, whose meaning
usually closer to the root than those of tense is modified by grammatically marked aspect,
are if both are bound morphemes (cf. also different adverbials, number of the actant
Bickerton’s hypothesis in § 8). To take an ex- NP’s and so on (Verkuyl 1972; Lindstedt
ample from Modern Greek: in the verb lı́po 1984, 1985: 169⫺210; Dahl 1994: 245⫺246).
‘be absent, missing, lacking’ both the aug- As shown above, the grammatically marked
ment é- and the personal suffixes show the level itself can be layered. Bybee, Perkins &
tense; the relative ordering of tense and as- Pagliuca (1994: 21⫺22) use the term “layer-
pect is relevant to the position of the personal ing” somewhat differently to refer to the vari-
endings and the perfective marker -s-: ous diachronic sources and age of, say, com-
imperfective perfective peting future grams such as the English will,
non-past sg.1 lı́p-o lı́p-s-o shall, is going to ⬃ gonna; the difference is
non-past sg.3 lı́p-i lı́p-s-i that this second kind of layering manifests
non-past pl.2 lı́p-ete lı́p-s-ete itself paradigmatically, not syntagmatically.
past sg.1 é-lip-a é-lip-s-a
past sg.3 é-lip-e é-lip-s-e
past pl.2 lı́p-ate lı́p-s-ate 8. Grammaticalization paths
In Bulgarian, there is a Russian-like perfec- Bybee and Dahl (1989: 54⫺55) report that
tive/imperfective opposition in all tenses, but between 70 and 80 per cent of the tense and
also an additional perfective/imperfective op- aspect grams found in their typological sam-

ples (Bybee 1985; Dahl 1985) belong to the an important role in pidgin and creole lin-
six major gram-types of perfective, imperfec- guistics (J Art. 116, Art. 117). The creolist
tive, progressive, future, past and perfect (an- Derek Bickerton (1981) has put forward a
terior). Of these, the perfect, the progressive Language Bioprogram Hypothesis, accord-
and (to a lesser degree) the future tend to ing to which true creolization (which to him
have periphrastic expression, the imperfec- always means nativization, i. e., the emer-
tive, the perfective and the past morphologi- gence of first-generation native speakers, in
cally bound (inflectional) expression. In the certain well-defined sociolinguistic circum-
gram-based approach to tense and aspect, the stances) always produces the same funda-
existence of a limited set of cross-linguistic mental grammatical structures, the roots of
gram-types, as well as the typical meaning- which lie in the genetically determined hu-
form correlations (such as the propensity of man language faculty. He has argued that as
the perfect to be expressed periphrastically), regards tense-aspect-mood, the most funda-
are explained as results of recurrent processes mental markers are those of the anterior
of grammaticalization (J Art. 113), or uni- (tense), the irrealis (mood) and the non-punc-
versal grammaticalization paths, which re- tual (aspect), and that they typically occur
peatedly produce similar results in different as preverbal particles or auxiliaries in this
languages. It is an open question whether order (Bickerton 1981: 58⫺59, 73⫺99). His
these grammaticalization paths realize a “non-punctual” marker is essentially a pro-
functional teleology of some kind, producing gressive/imperfective marker for non-stative
expressions for meanings that are univer- verbs; statives are unmarkedly non-punctual.
sally needed for human communication, or The Language Bioprogram Hypothesis is
whether they simply reflect the fact that the not generally accepted by pidgin and creole
normal functioning of human communica- scholars, but it has had seminal influence on
tion and cognition automatically creates studies of creolization, and this also applies
changes that tend to be similar everywhere, to investigations of tense and aspect systems
but do respond to any kind of long-term need and their development. The results of recent
or teleology. The latter view is endorsed by studies do not generally seem to support the
Bybee, Perkins and Pagliuca (1994: 297⫺ hypothesis that emerging creoles must pass
300); for a compatible theory of language through a stage at which the three grams pos-
change, see Nyman (1994) and the literature tulated by Bickerton are employed (see Ro-
quoted there. maine 1988: 264⫺274 and the papers in Sin-
The Evolution of Grammar by Bybee, Per- gler (ed.) 1990). For instance, the configura-
kins and Pagliuca (1994) is a major contributions of past, anterior (perfect) and perfective
tion to the study of the grammaticalization markings may differ from creole to creole.
paths of tense, aspect and modality in the Bakker et al. (1994: 258) present a more cau-
languages of the world. They find, for in- tious generalization than Bickerton: all cre-
stance, that resultatives and completives (see oles possess at least three markers, one for
§ 6 above) often develop into anteriors (per- tense, one for mood and one for aspect, and
fects), which tend to develop further into they are most often placed before the verb in
perfective pasts or simple pasts; an alterna- this order.
tive path leads from resultatives to eviden- The main methodological problem is that
tials (pp. 104⫺105). Progressives, often origi- there may simply not be any creoles that have
nating as locative expressions (pp. 128⫺129), been created from scratch, without the strong
tend to develop into presents or imperfectives influence of one or more existing languages,
(pp. 140⫺149). Unmarked perfectives can as Bickerton’s theory would require; there-
come into existence by becoming opposed to fore, we can not know what a tense and as-
a newly developing imperfective gram; the pect system without a history would be like.
same applies to the relation between an In his paper originally circulated since
unmarked present tense and a past gram 1971, Labov (1990) traces the development of
(pp. 90⫺91). Futures arise from such modal tense markers in some creoles, and concludes
notions as desire and obligation, but also that they do not appear as a response to
from movement verbs (either ‘going’ or ‘com- some fundamental cognitive and communica-
ing’) and (rarely) temporal adverbs (pp. 251⫺ tive need:
271). “One might say that a developing grammar serves
The question of fundamental tense-aspect- the need of stylistic variation. But it would be more
mood distinctions and their origins has had accurate to say that grammar is style. […] On the

whole, grammar is not a tool of logical analysis: tini & Bertinetto 2000), or the change of the
grammar is busy with emphasis, focus, down-shift- German perfect into an aspect-neutral past
ing and up-grading; it is a way of organizing infor- tense (Latzel 1974; Thieroff 1992, 1994,
mation and taking alternative points of view” (La-
2000). Of course, when a development of this
bov 1990: 45).
kind has been completed ⫺ as is the case with
Bybee, Perkins and Pagliuca (1994: 299⫺300) the Russian Past Tense, historically a present
add to this that grammar may also facilitate perfect ⫺ no cline can be observed any
production through automation. Automated, longer; but new changes are then already go-
obligatory grams help the speaker concen- ing on elsewhere in the grammar. An ade-
trate on the propositional content of the ut- quate description of the tense and aspect sys-
terance. This kind of argument is functional, tem of any language must thus refer not only
but it does not presuppose that grammaticali- to the meaning of the individual grams and
zation has a long-term teleology. When lexi- to the structure of their overall system, but
cal words become tense and aspect markers, also to the on-going grammaticalization pro-
the final result of grammaticalization is no cesses as reflected in various kinds of syn-
more intended by anyone than inflation is a chronic variation.
goal of those economic agents whose deci-
sions produce it, jointly and inadvertently.
Notice, incidentally, that the claim that gram- 9. Special abbreviations
maticalized (especially inflectional) items fa- aor aorist (⫽ perfective past tense)
cilitate production is indirectly an assertion impf imperfect (⫽ imperfective past tense)
about what kind of language we are biolo- impfv imperfective
gically best equipped to process, though it is nt neuter (gender)
less specific a claim than Bickerton’s biopro- past past tense (⫽ preterit)
gram hypothesis (J Art. 7, Art. 29). prep preposition
It is important to note that grammaticali- sg1 1st person singular
zation theories are relevant not only to the sg3 3rd person singular
question of the origins and diachronic devel- sup supine
opment of tense and aspect grams; they are ut utrum (gender)
also a necessary part of their synchronic de-
scription or, rather, they make the very dis-
tinction between diachrony and synchrony 10. References
relative rather than absolute (J Art. 111; cf.
Agrell, Sigurd. 1908. Aspektänderung und Aktions-
Dahl 2000b: 8⫺18). There are no synchronic
artbildung beim polnischen Zeitworte. Ein Beitrag
steady states to be described. Each gram may zum Studium der indogermanischen Präverbia und
have uses that still preserve its earlier mean- ihrer Bedeutungsfunktionen. Lund.
ing; for instance, a future auxiliary that origi-
Arends, Jacques & Muysken, Pieter & Smith, Nor-
nally meant ‘want’ may still exhibit this sense val (eds.) 1995. Pidgins and creoles: An introduction.
in special contexts (as in English He will have (Creole Language Library, 15.) Amsterdam & Phil-
his own way, with stress on will). New and adelphia: John Benjamins.
expanding uses of a gram point to its future Asher, R. E. & Simpson, J. M. Y. (eds.) 1994. The
development, but they are also a synchronic encyclopedia of language and linguistics 1⫺8. Ox-
fact. ford et al.: Pergamon Press.
Grammaticalization processes are syn- Bache, Carl & Basbøll, Hans & Lindberg, Carl-
chronically reflected as grammaticalization Erik (eds.) 1994. Tense, aspect and action: Empiri-
clines (Dahl 2000b: 14⫺15): the propensity cal and theoretical contributions to language typol-
towards the use of a gram decreases mono- ogy. Berlin & New York: Mouton de Gruyter.
tonically along some dimensions, such as the Bakker, Peter & Post, Marike & van der Voort,
semantic distance from certain prototypical Hein. 1995. “TMA particles and auxiliaries.” In:
(focal) contexts, the geographical distance Arends et al. (eds.) 1995, 247⫺258.
from the centre of innovation, or the age Bertinetto, Pier Marco & Ebert, Karen H. & de
scale from younger to older generations. A Groot, Casper. 2000. “The progressive in Europe.”
multidimensional synchronic cline is a suit- In: Dahl (ed.) 2000, 517⫺558.
able model for describing the gradual change Bertinetto, Pier Marco & Bianchi, Valentina &
of the Romance perfect (such as the French Dahl, Östen & Squartini, Mario (eds.) 1994. Tem-
passé composé or the Italian passato prossimo) poral reference, aspect and actionality, vol. 2: Typo-
into a general perfective past tense (Squar- logical perspectives. Torino: Rosenber & Sellier.

Bickerton, Derek. 1981. Roots of language. Ann Depraetere, Ilse. 1995. “On the necessity of distin-
Arbor: Karoma. guishing between (un)boundedness and (a)telicity.”
Bondarko, A[leksandr] V[ladimirovič] 1990. “O Linguistics and Philosophy 18: 1⫺19.
značenijax vidov russkogo glagola.” Voprosy jazy- Forsyth, J. 1970. A grammar of aspect. Cambridge:
koznanija 1990/4: 5⫺24. Cambridge University Press.
Bondarko, A[leksandr] V[ladimirovič] et al. 1987. Friedman, Victor A. 1976. “Dialectal synchrony
Teorija funkcional’noj grammatiki: Vvedenie. Aspek- and diachronic syntax: The Macedonian perfect.”
tual’nost’. Vremennaja lokalizovannost’. Taksis. Le- In: Sanford B. Steever et al. (eds.), Papers from the
ningrad: Nauka. parasession on diachronic syntax. Chicago: Chicago
Brauner, Siegmund & Ashiwaju, Michael. 1965. Linguistic Society, 96⫺104.
Lehrbuch der Hausa-Sprache. (Lehrbücher für das Friedman, Victor A. 1977. The grammatical cate-
Studium der orientalischen und afrikanischen gories of the Macedonian indicative. Columbus,
Sprachen, 10.) Leipzig: VEB Verlag Enzyklopädie. Ohio: Slavica.
Bybee, Joan L. 1985. Morphology: A study of the Friedman, Victor A. 1986. “Evidentiality in the
relation between meaning and form. Amsterdam & Balkans: Bulgarian, Macedonian, and Albanian.”
Philadelphia: John Benjamins. In: Chafe & Nichols (eds.) 1986, 168⫺187.
Bybee, Joan & Dahl, Östen. 1989. “The creation of Givón, Talmy. 1982. Tense-aspect-modality: The
tense and aspect systems in the languages of the creole prototype and beyond. In: Hopper (ed.)
world.” Studies in Language 13: 51⫺103. 1982, 115⫺163.
Bybee, Joan & Perkins, Revere & Pagliuca, Wil- Graves, Nina. 2000. “Macedonian ⫺ a language
liam. 1994. The evolution of grammar: Tense, as- with three perfects?” In: Dahl (ed.) 2000, 479⫺494.
pect, and modality in the languages of the world. de Groot, Casper & Tommola, Hannu (eds.) 1984.
Chicago & London: University of Chicago Press. Aspect bound: A voyage into the realm of Germanic,
Carlson, Lauri. 1981. “Aspect and quantification.” Slavonic and Finno-Ugrian aspectology. Dor-
In: Tedeschi & Zaenen (eds.) 1981, 31⫺64. drecht & Cinnaminson: Foris.
Carlson, Lauri (forthcoming). Tense, mood, aspect, Haarmann, Harald. 1970. Die indirekte Erlebnis-
diathesis: Their logic and typology. form als grammatische Kategorie. Eine eurasische
Isoglosse. (Veröffentlichungen der Societas Uralo-
Chafe, Wallace & Nichols, Johanna (eds.) 1986. Altaica, 2.) Wiesbaden: Otto Harrassowitz.
Evidentiality: The linguistic coding of epistemology.
(Advances in Discourse Processes, 20.) Norwood, Haspelmath, Martin. 1998. “The semantic develop-
ment of old presents: new futures and subjunctives
New Jersey: Ablex.
without grammaticalization.” Diachronica 15.1:
Comrie, Bernard. 1976. Aspect: An introduction to 29⫺62.
the study of verbal aspect and related problems.
Haugen, Einar. 1972. “The inferential perfect in
Cambridge: Cambridge University Press.
Scandinavian: a problem for contrastive linguis-
Comrie, Bernard. 1981. Language universals and tics.” The Canadian Journal of Linguistics 17:
linguistic typology. Oxford: Basil Blackwell. 132⫺139.
Comrie, Bernard. 1985. Tense. Cambridge: Cam- Heinämäki, Orvokki. 1984. “Aspect in Finnish.”
bridge University Press. In: de Groot & Tommola (eds.) 1984, 153⫺177.
Comrie, Bernard. 1994. “Tense.” In: Asher & Heinämäki, Orvokki. 1994. “Aspect as bounded-
Simpson (eds.) 1994, 4558⫺4563. ness in Finnish.” In: Bache et al. (eds.) 1994,
Dahl, Östen. 1985. Tense and aspect systems. Ox- 207⫺233.
ford & New York: Basil Blackwell. Hopper, Paul J. (ed.) 1982. Tense-aspect: Between
Dahl, Östen. 1994. “Aspect.” In: Asher & Simpson semantics and pragmatics. (Typological Studies in
(eds.) 1994, 240⫺247. Language, 1.) Amsterdam & Philadelphia: John
Benjamins.
Dahl, Östen. 2000a. “The grammar of future time
Ikola, Osmo. 1953. Viron ja liivin modus obliquuk-
reference in European languages.” In: Dahl (ed.).
sen historiaa. Helsinki: Suomalaisen Kirjallisuuden
2000, 309⫺328.
Seura. (Suomi 106, 4.) [With a summary in Ger-
Dahl, Östen. 2000b. “The tense-aspect systems of man: “Zur Geschichte des estnischen und livischen
European languages in a typological perspective.” Modus obliquus.”]
In: Dahl (ed.) 2000, 3⫺25.
Jakobson, Roman. 1971a [1936]. “Beitrag zur all-
Dahl, Östen (ed.) 2000. Tense and aspect in the lan- gemeinen Kasuslehre: Gesamtbedeutungen der
guages of Europe. (Empirical approaches to lan- Russischen Kasus.” In: Selected writings II. The
guage typology/EUROTYP 20⫺6.) Berlin: Mou- Hague & Paris: Mouton, 23⫺71. [Originally in
ton de Gruyter. Travaux du Cercle linguistique de Prague, VI.]
Dahl, Östen & Hedin, Eva. 2000. “Current rele- Jakobson, Roman. 1971b. “Shifters, verbal cate-
vance and event reference.” In: Dahl (ed.) 2000, gories, and the Russian verb.” In: Selected writings
385⫺401. II. The Hague & Paris: Mouton, 130⫺147.

Johanson, Lars. 1992. Strukturelle Faktoren in Nyman, Martti. 1994. “Language change and the
türkischen Sprachkontakten. (Sitzungsberichte der ‘invisible hand’.” Diachronica 11: 231⫺258.
Wissenschaftlichen Gesellschaft an der Johann Reichenbach, Hans 1966 [1947]. Elements of sym-
Wolfgang Goethe-Universität Frankfurt am Main bolic logic. New York: The Free Press; London:
XXIX, 5.) Stuttgart: Franz Steiner. Collier, Macmillan.
Kinnander, Bengt. 1973. “Perfektum i ‘sekundär’ Romaine, Suzanne. 1988. Pidgin and creole lan-
användning.” Nysvenska studier 53: 127⫺172. guages. London & New York: Longman.
Kraft, G. H. & Kirk-Greene, A. H. M. 1979 [1973]. Schmalstieg, William R. 1988. A Lithuanian histori-
Hausa. Sevenoaks, Kent: Teach Yourself Books. cal syntax. Columbus, Ohio: Slavica.
Kučera, Henry. 1981. “Aspect, markedness, and Schwenter, Scott A. 1994. “The grammaticalization
t0.” In: Tedeschi & Zaenen (eds.) 1981, 177⫺189. of an anterior in progress: Evidence from a pen-
insular Spanish dialect.” Studies in Language 18:
Labov, William. 1990. “On the adequacy of natural 71⫺111.
languages: I. The development of tense.” [Origi-
nally written in 1971.] In: Singler (ed.) 1990, 1⫺58. Singler, John Victor (ed.) 1990. Pidgin and creole
tense-mood-aspect systems. (Creole Language Li-
Latzel, Sigbert. 1974. “Zum Gebrauch der deut- brary, 6.) Amsterdam & Philadelphia: John Benja-
schen Vergangenheitstempora: Zwei Studien.” In: mins.
Gelhaus, Hermann & Latzel, Sigbert. Studien zum
Squartini, Mario & Bertinetto, Pier Marco. 2000.
Tempusgebrauch im Deutschen. (Institut für deut-
“The simple and compound past in Romance lan-
sche Sprache, Forschungsberichte, 15.) Mann-
guages.” In: Dahl (ed.) 2000, 403⫺439.
heim, 169⫺348.
Tedeschi, Philip J. & Zaenen, Annie (eds.) 1981.
Leinonen, Marja & Vilkuna, Maria. 2000. “Past Syntax and semantics 14: Tense and aspect. New
tenses in Permian languages.” In: Dahl (ed.) 2000, York: Academic Press.
495⫺514.
Thieroff, Rolf. 1992. Das finite Verb im Deutschen.
Lindstedt, Jouko. 1984. “Nested aspects.” In: de Tempus ⫺ Modus ⫺ Distanz. (Studien zur deut-
Groot & Tommola (eds.) 1984, 23⫺38. schen Grammatik, 40.) Tübingen: Narr.
Lindstedt, Jouko. 1985. On the semantics of tense Thieroff, Rolf. 1994. “Das Tempussystem des
and aspect in Bulgarian. (Slavica Helsingiensia, 4.) Deutschen.” In: Thieroff & Ballweg (eds.) 1994,
Helsinki. 119⫺134.
Lindstedt, Jouko. 1995. “Understanding perfectiv- Thieroff, Rolf. 2000. “On the areal distribution of
ity ⫺ understanding bounds.” In: Bertinetto et al. tense-aspect categories in Europe.” In: Dahl (ed.)
(eds.) 1995, 95⫺103. 2000, 265⫺305.
Thieroff, Rolf & Ballweg, Joachim (eds.) 1994.
Lindstedt, Jouko. 2000. “The perfect ⫺ aspectual,
Tense systems in European languages. (Linguistische
temporal and evidential.” In: Dahl (ed.) 2000,
Arbeiten, 308.) Tübingen: Niemeyer.
365⫺383.
Tommola, Hannu. 2000. “On the perfect in North
Mackridge, Peter. 1985. The Modern Greek lan- Slavic.” In: Dahl (ed.) 2000, 441⫺478.
guage. Oxford: Clarendon Press.
Trask, R. L. 1993. A dictionary of grammatical
Maslov, Jurij S[ergeevič] 1984. Očerki po aspekto- terms in linguistics. London & New York: Rout-
logii. Leningrad: Izdatel’stvo Leningradskogo uni- ledge.
versiteta. Trubinskij, Valentin I. 1988. “Resultative, passive,
Maslov, Jurij S[ergeevič] 1988. “Resultative, per- and perfect in Russian dialects.” In: Nedjalkov
fect, and aspect.” In: Nedjalkov (ed.). 1988, 63⫺85. (ed.) 1988, 389⫺409.
Mønnesland, Svein. 1984. “The Slavonic frequen- Vendler, Zeno. 1967. “Verbs and times.” In: Z.
tative habitual.” In: de Groot & Tommola (eds.) Vendler: Linguistic in Philosophy. Ithaca, New
1984, 53⫺76. York: Cornell University Press, 97⫺121. [Revised
version of “Verbs and times.” The Philosophical
Mourelatos, Alexander P. D. 1981. “Events, pro- Review 66 (1957): 143⫺160.]
cesses and states.” In: Tedeschi & Zaenen (eds.) Verkuyl, Henk J. 1972. On the compositional nature
1981, 191⫺212. of the aspects. (Foundations of Language, Supple-
Nedjalkov, Vladimir P. (ed.) 1988. Typology of re- mentary Series, 15.) Dordrecht: D. Reidel.
sultative constructions. (Typological Studies in Lan- Weinrich, Harald. 1964. Tempus: Besprochene und
guage, 12.) Amsterdam & Philadelphia: John Ben- erzählte Welt. Stuttgart: W. Kohlhammer.
jamins. [An enlarged translation of Tipologija re- Willett, Thomas. 1988. “A cross-linguistic survey
zul’tativnyx konstrukcij (rezul’tativ, stativ, passiv, of grammaticalization of evidentiality.” Studies in
perfekt), Leningrad 1983: Nauka.] Language 12: 51⫺97.
Nedjalkov, Vladimir P. & Sergej Je. Jaxontov. 1988.
“The typology of resultative constructions.” In: Jouko Lindstedt, University of Helsinki
Nedjalkov (ed.) 1988, 3⫺62. (Finland)


Tense_and_aspect

Uploaded by

Copyright:

Available Formats

Tense_and_aspect

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tense_and_aspect

Uploaded by

Copyright:

Available Formats

768 IX.

Typology of morphological and morphosyntactic categories

proximal medial distal obvial

Portuguese DEM este esse aquele

59. Tense and aspect

1. Introduction temporal properties of situations (J Art. 42).

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

Brought to you by | University of Helsinki (University of Helsinki - empty ID - use)

You might also like