Vous êtes sur la page 1sur 10

1

The syntax of tense

Introduction:
The generally accepted definition of the category of Tense, as a category delimiting the
part of speech verb, explains Tense as representing the chronological order of events in time as
perceived by the speaker at the moment of speaking. The notions to be accounted for in this
definition are: chronological order, Time and moment of speaking. The presence of a thing in one
place and its non-presence at the same place can be perceived by a human subject if and only if
these two contradictory properties are placed sequentially, one after another, that is in Time
(Stefanescu 1988:216). What this actually means is that Time (just like Space) is the form of
our experience of the world. (epistemic notion). If Time can be viewed as being not a
determination of outward phenomena, then it has to do with neither shape or form. Currently, this
want is supplied by analogies and the course of Time is represented by a line progressing to
infinity. This linear representation of Time preserves the sequential character (i.e. chronological
order) of our perception of the world. We perceive Time in the same way we perceive Space, i.e.
we cannot live in two times simultaneously as we cannot, at the same time, occupy two spatial
locations. It means that when Time is measured by lived-through eventualities the measurement
is unidirectional, i.e. forwards.
Time is a single unbounded dimension, conceptually analogous to Space. Just as an orientation
point is needed to locate positions in space, so an orientation point is needed to locate situations
in time.
On the Time line, times and situations are located at moments or intervals relative to the Time of
utterance. The situations may occur in order (i.e. sequentially) or they may overlap, wholly or in
part.
All sentences give us temporal information which helps us locate in Time the situation
talked about. This temporal information is given by Tense morphemes and time adverbials.
Tense is a functional category, expressed by a set of verbal inflections or other verbal
forms, that expresses a temporal relation to an orientation point.( Smith, 1991).
Tenses have consistent relational values: anteriority, posteriority or simultaneity. Tenses
may have a fixed or flexible orientation. Tenses with fixed orientation are always related to Ut-T.
Whenever tenses, or rather, Tense systems are oriented to the moment of speech (i.e. the speaker)
we say that they are used deictically (i.e. they are interpreted as pointing expressions, just like
adverbs (tomorrow, now, here, there) or pronouns (this, that, I, you)).
The traditional term for tenses that relate to Ut-T is absolute tenses. Tenses that relate to
an orientation time other than Ut-T are known as relative tenses .

2
Not all temporal reference is made by Tense. In English, the Future is indicated by
another type of morpheme, the modal auxiliary shall/will. It is also common to have a
combination of present tense (or present tense progressive in English) and future time adverbial
that indicates the future, sometimes called Futurate.
Some languages have tenses that indicate Present, Past and Future. Some others have a
tense distinction between past and non-past, still others have a distinction between present and
non-present. Some languages (e.g. Mandarin Chinese, Malay, Classical Hebrew) do not have the
functional category of Tense. For these languages temporal location is expressed directly by
adverbials and indirectly by (viewpoint) aspect. There are also languages where tenses contribute
temporal location as well as aspectual value i.e. aspectual viewpoint may also be conveyed by
Tense. The French Impairfait and the Romanian Imperfect, for instance, may also convey a
general imperfective viewpoint. In English, as we have seen, the expression of aspectual
viewpoint is independent of Tense.
The interpretation of tense is a complex matter; particular tense forms can be construed in
quite different ways, as determined by a complex interaction of factors: verbal aspectual class
(stative vs eventive); grammatical aspect (progressive vs punctual) clause type (complement
clause,relative clause; finite clause vs infinitive) and scope relations with other tenses.
The category Tense is indeed central to the computational system: it plays an important
role at both interfaces (articulatory-perceptual (PF) and conceptual-intentional (LF)). PF- triggers
agreement in many lags. LF: locus for temporal interpretation. So, in both cases T functions as a
bridge toward the context (Giorgi 2008): locates the event with respect to speech participants,
and locates events with respect to each other and with respect to UT-T.
There have been different tense theories that have relied on idiosyncratic set of semantic and or
syntactic rules to account for the distribution and interpretation of particular tense forms. The
theory we will present is designed to avoid such rules and to reduce most of the syntax and
semantics of tense to independently motivated principles of syntactic theory. The theory adopted
will be beneficial for the crosslinguistic variation of tense phenomena, in he sense that all lgs
express temporal location of events but differ widely as to the way of expressing temporal
relations. Romanian and Italian have a rich and complex morphological system encoding various
temporal relations, English have a quite simpler system explicitly encoding only a subset of the
distinctions encoded by the two lgs. Other lgs have no temporal morphemes at all. In Latin or
Russian it is possible to have main clauses without a morphological verbal form.
Caesar imperator/ Ia devotschka.
1. Syntactically, tense has been analyzed as an expansion of Aux or Infl, as a specifier of VP
and recently as a functiobal X-bar head

3
2. Semantically: tense has been analyzed as a (i) sentential operator designating the time at
which the truth of the sentence is to be evaluated (ii) as a referential expression denoting
the time of a situation, and (iii) as a temporal predicate analogous to a verb.
3. A syntactically based theory of tense can and must take contextual discourse factors into
consideration (Smih). The dependency of tense and context can be handled in DRT.
The intuition that tense is a referential expression is based on the observation that different tenses
seem to refer to particular locations in time; the past in John sang a song seems to refer to some
interval of time (prior to UT) during which John sang.
The intuition that Tense has predicative content is based on the observation that it expresses a
relation of temporal ordering holding between times. In John sang a song the past tense locates
the time of singing prior to the time of utterance. Similarly in Bill reported that John sang a
song the past tense in the subordinate locates the time of singing prior to the time of reporting.
TENSE AS A REFERENTIALEXPRESSION (Enc 1987)
-tenses are referential expressions denoting intervals. The idea is not new. Partee argues that
tenses behave like pronouns. They can have antecedents in the discourse or they can have
sentence internal antecedents.
(1)
We went to a party. John got drunk. (the time of getting drunk is understood as the time
of the party)
(2)
John arrived at three (the adverb is understood as determining the time of arriving-it can
be interpreted as the antecent of tense
The reason is that it allows one to account for certain relations between temporal expressions by
appealing to general notions of linguistic theory . Temporaladverbials should be treated on a par
with NPs1)yesterday refers to an interval that precedes the day containing the time of
utterance;(2) temporal expressions may occur in argument position
(3)

Every afternoon turned out to be disastrous


I hated Monday

(Larson (1985) :temporal expressions are NPs ; they refer to times (see ex) hence should be
admitted in the domain of discourse alongside other individuals. Hence, Tense denote intervals
and provide the temporal argument of the verb. The notions past and present are relational
notions : past denotes an interval that is prior to some other interval which needs to be specified
to interpret the sentence (the reference interval). Tense is in Infl and the reference interval (the
speech time in matrix clauses) is in C (given the strong connection between Infl and C). C can
optionally carry a temporal index, and when it does, it yields the reference interval which

4
functions as RT for the Tense in Infl. This means that the local C which governs Tense functions
as an anchor (RT) i.e. C binds Tense.
Anchoring principle: Each tense must be anchored (an identification requirement for tenses)
Tenses with no specified RT are unanchored, hence cannot be interpreted. The elements
necessary for tense interpretation are present in the LF of the sentence.
Anchoring Conditions:
(i)

Tense is anchored if it is bound in a local domain (governing category) or if its local


C is anchored. Otherwise it is unanchored.
If C has a GC it is anchored iff it is bound within the GC
If C does not have a GC it is anchored iff it denotes speech time

(ii)
(iii)

Anchoring in matrix clauses


Tense is governed by C but C is not governed hence has no GC. Tense can be anchored only if C
denotes ST. C picks up a value from context which serves as RT fot T, C thus acts as binder of T.
4) John left [S C0 [S NP [I PASTi VP ]]]

TENSES AS TEMPORAL PREDICATES (Zagona 1990, Tim Stowell)


Zagona has proposed a theory of tense that assumed T to be a dyadic predicate which takes time
denoting expressions. The referential content of tense resides in one of the time denoting
arguments.The external argument of T refers to the UT while the internal refers to the time of the
event denoted by VP The T predicate establishes an ordering relation between these two times. In
this view tenses resemble prepositions like before and after . Since tense is a predicate that takes
arguments these arguments should be represented syntactically.
TP

UT

T
T

Event T

This arrangement indicates that the tense predicate rather than the verbal predicate is the main
predicate in the sentence.
Syntactic category of the time-denoting arguments: they are taken to be DPs

5
[DP thei [NP ei man]] D isa referential category that enables DP to refer. N Is a predicate and D
binds the variable.
Thetime denoting phrase is ZP and the head is Z (Zeit) Z (just like D) binds the e variable of the
V (e is the true external argument of V rather then the DP subject). (the internal ZP in Stowell
corresponds to Tense in Enc-who takes T to be a referential expression expressing ET).
The structure of the internal ZP denoting ET :

ZP
Zi

DP
VP

ei

Di
VP

John

NP
ei

Sing

man

E I is the temporal argument of the predicate sing which is bound by the referential head of ZP
(DP) a variable bound by Z (D)
The status of Z is that of a temporal quantifier (indefinite) binding a temporal argument variable
inside VP
Status of external time-argument of Tense (Zagona called it UT) but in this case it cannot be
appropriate for: Bill reported that John sang a song),where the external argument of the
complement clause PAST does not denote the UT. Hence the suggestion is to call the external
argument of T as reference time. In this system reference time can have one of its denotations UT
(In Reichenbach RT &UT are primitives encoded in all the representation of every tense).
What is the denotation of RT in complement clauses? He said that John sang a song?
To accommodated the denotation of the external ZP Stowell assumes that it is an empty ZP
PRO which can be interpreted as deictic in main clauses ref. to UT) and as anaphor in
complement clauses (when bound by the ET of the matrix clause In matrix clauses it behaves
like PROarb taking its denotation from the speech-act setting.
How does Enc account for the temporal ordering relation holding between the arguments of T?
(it is in the form of anchoring between Tense (event time) and C (UT).
Tenses in complex sentences

6
Consider the following examples:
(a)
(b)
(c)
(d)

John said that Mary finally gave up smoking


John declared that the patient died early in the morning
John said that Mary was pregnant
John said that it was raining when he left
In Encs theory the shifted reading is obtained by coindexing the complement C with the
matrix Tense which serves as a RT for the subordinate clause Tense. (the matrix tense is
the antecedent for the embedded C and hence C is anchored) . Hence the embedded C
serves as anchor for the embedded tense. The complement Tense is interpreted as anterior
to the interval designated by the embedded and the matrix Tense. The coindexation of the
matrix PAST with the embedded C makes possible the shifted reading interpretation.
The matrix Past denotes the time of saying/declaring anterior to UT while the
complement T designates the time interval when the event occurs in the embedded
clause, supposed to be anterior to the saying. Hence the shifted Past reading follows from
Binding Theory.
[C[NP[PASTi[V[Ci[NP[PASTj]]]
The simultaneous reading (pregnant example) gets a different anchoring possibility. The
embedded tense is related to the matrix tense without the mediation of the embedded C.
The tense in the dub. Clause is governed by C hence has a governing category which is
the smallest projection containing the governor (C) and an accessible subject: this
projection is the matrix clause- within it the antecedent is the matrix Tense:
[C[NP[PASTi [V[C[NP[PASTi
The complement Past and the matrix past are both anterior to the UT/RT specified by the
matric C = the simultaneous reading is obtained given the indexing trick the embedded
Past is still past rather than a disguised present, since it is past with respect to UT-T
(instead of past with respect to embedded C).
The system, though elegant faces empirical andconceptualproblem one of which
is raised by an example like :
John said to me (yesterday) that he would tell his mother (tomorrow at lunch) that they
were eating their last meal together
The Pst in the third clause shows no anteriority with respect to any interval in the
sentence but simple simultaneity with the future verb in S2: the time of eating is
simultaneous with the time telling and precedes no other time referred to in the sentence.
STOWELL:
The embedded specifier of TP hosts PRO-ZP which is coindexed with the temporal
variable e which c-commands it and is itself bound by the matrix Z head of the matrix

7
event time (complement of matrix past) (see19p.15). The past shifted reading follows
from control theory as consequence of the assumption that the externalargument of Tense
is a ZP-PRO.
Differences Enc Stowell
1) In Ss theory the predicative content of Tense is given an an overt X-bar theoretic
expression. This aspect of tense is given no syntactic expression in Enc: the semantic
content of the anchoring relation is derived by stipulative interpretive rules.
2) In S. the postulation of PRO-ZP (corresponding roughly to Es use of C) reduces the
shifted reading to an effect of control theory, making it unnecessary to postulate an
idiosyncratic system of binding for tenses per se.
3) Ss theory draws a sharp distinction between 2 types of relations between timedenoting categories: (i) the relation btw the internal & external arguments of a tense
predicate is a relation between co-arguments of a predicate (ii) the relation between
PRO-ZP f complement clause and the ZP of a higher VP is acontrol relation
analogous to the one between a PRO-DP and its antecedent
4) Ens theory recognizes an analogous interpretive distinction btw the anchoring of
tense to C and the anchoring of C to a higher Tense but neither typeof relation has a
direct structural analog elsewhere in the theory of grammar. Moreover we have
different indexing strategies for the same syntactic configuration

Jacqueline GUERON (1993

8
On the syntax of time
-Tense interpretation =relations between the nodes C, T and V within a Tense-Chain- this is
universal the structural relations btw the 3 heads does not change from a language to another,
along the lines suggested by Reichenbach.
-employs mechanisms likeL Ouantificayion, movement &Binding. Bi-uniqueness principle
(see 3.1.3. 25, 26)
-Present is construed as anaphor; past as pronominal Future as Name
-Differences btw E &F in the interpretation of Pr, Pr.Perf,Progressive & Imp follow from a single
parameter presence vs absence of a morphological person feature.
1. the first paragraph

V, INFL, C participate in the temporal interpretation of the sentence. Synt.


Representation feeds semantic interpretation at LF.
C abstract temporal morpheme R which denotes the temporal reference of the clause
(Enc 1987) If R=S (time of utterance) then R is interpreted as deictic; I includes T
morpheme the value of which is calculated relative to the Time in C (i.e. R). VP is
associated with E (state or event).
At LF E functions like a variable saturated by T in Infl The event deoted by E is
predicated of an instant/moment in time
In (1) the event =a property of a moment anterior to the moment of utterance
C,T,V=T-chain each having a temporal index
In (2) Ti cannot bind E2 since C and Tj block it = constraint of locality and c-command
and shortest link.

2. Differences btw F and E

2.1

In E the present simple cannot have an existential reading with events: see (3); states
allow this reading see (7 vs 8). See 9,10 punctual events are ecvluded.
Pres denotes: generic /narrative/future both in E &F
Existential reading in rendered by means of the progressive
Present behaves as if it had inherent durative aspect (see punctuality constraint
Georgi and Pianesi)
Tense and Aspect
Tense denote a point; the event is predicated globally
Difference btw tense and aspect (see 14) comment on the necessity of introducing T2

3.THE GRAMMAR OF TENSES


1. Different representations for states and events (stages)branching structure see ((20,21)
2. R in C and T in Infl are nominal nodes (ENC, Zagona, Stowell). In SpecCP (+D) a deictic
operator= R in C which is identified as a deictic pronoun R=Speech time.

9
T in Infl = pronoun, name, anaphora (zagona) (see 22)
Past tense = pronoun like l in l-am vazut since it can get its reference from discourse;
i.e. it is free within the domain CP since past is not identical with present in C.
Future is interpreted as a name since T can not be identified with any R in the world. (see
23, 24)
Present corresponds to a anaphoric interpretation of T since it is bound by R in C which
governs it. Anaphoric T copies the content of R in C.
If the V does not permit the division of the eventuality into stages (sub-events) we have
several possibilies: iteration with punctual predicates
The durative, continuous aspect of Present Tense is not inherent to the grammatical
morpheme. It depends on the identification of R. If it is marked as deictic it denotes a
series of moments of times. .
If SpecCP contains a generic operator (i.e. universal) R functions as a restrictor of
that operator while T is a variable bound by R (see 29)
For the habitual reading SpecCp has a multiplying operator (see30)
For the narrative SPecCP has a narrative operator (see31).
In (33,34) the operator in SpecCP is Action (a distributive operator) which is
close to a domain which contains a series of CPs Each T is identified as a pronoun
bound by a different R. Action = Narrative operator and it accounts for
performatives and achievements as well (35,36).

3.2.

The difference between E and F (9ab) vs (10ab) are the result of the constraints in
(37), (38).i.e. the difference lies at the level of morphosyntax.
The lack of tense morphemes in E goes along with the present tense in F where
(37) is applicable too. In F, though, the V is marked for person : , -ons, -ez.
In E the morpheme s (Kayne 1989) does not indicate person but NUMBER in E
(both for N and V). Hence T is not identifiable acc. to (38) and cannot function as an
anaphor.
With state verbs T is not active if E is not branching T is inert (see39) At LF T
adjoins to R/S and the value of T is that of R/S
Future value (see 43, 44)

Complex tenses

A complex sentence contains 2 T nodes (see 18) one of which is assumed to be Aspect given
the assumption in 45, 46 :
(45)

(i)T governed by R in C is Tense

10
(ii)All Ts not falling under (i) is assumed to be Aspect
(46)

(i) T binds E viewed as closed


(ii) Asp. Binds a subevent of E namely ei

The examples in (47 are unacceptable assuming that (i) it does not have a CP node or (ii) that
CP does not have an operator to license R. Given (45i) in the absence of R we cannot have T.
The perfective morpheme in (47) saturates a subevent of E children go to bed the time of this
subevent depends on the time of the main clause (Past a, future (b) or present in (c)). In 48 a
the event that T licenses completely has a complex structure and can be modified by he
adverb. In (b) the Aspect node only licenses a subevent with no internal stages hence cannot
be modified.
3.2.2 The structure of the Perfect
In 49 the perfective aspect pp binds a subevent of E (ej) The auxiliary verb have is not
associated with any E , T1 being inert- it raises to C at LF and its temporal value is S; HAVE is
taken to be an inclusion operator hence (52). The inclusion of Ej in S places ej in the past
relative to the now of S but other instants follow ej since S is a sequence of moments,.
The perfect denotes a state not an event. In Eng. it does not allow punctual adverbs and does
not progress the narrative. (53 vs 54). In French it does.
Explanation: In F (given 45) it is only the context that distinguishes T from Aspect.The principle
allows that a morpheme in Aspect may be interpreted as Tense if it raises to T in syntax or LF. In
French Aspect raises to T in syntax (see 55) becoming a past Tense and blocking the present to
move to C. Avoir is interpreted as carrier of agreement not of tense, no longer as an inclusion
operator. V2 also moves to Infl 1 in order for E to be licenced. The interpretation of the perfect
as past is possible only for lgs where V moves to Infl. (possible due to rich inflection Pollock =
person features Gueron)

Vous aimerez peut-être aussi