Time and Tense in Language Events and Relations Event expressions; tensed verbs; has left, was captured, will resign; James Pustejovsky stative adjectives; sunken, stalled, on board; event nominals; merger, Military Operation, Gulf Brandeis University War; Dependencies between events and times: Anchoring; John left on Monday. LING 130 Orderings; The party happened after midnight. Embedding; John said Mary left. FALL, 2005
Tense Reichenbach • Tensed utterances introduce references to 3 ‘time points’ – Speech Time: S • Grammatical expression of the time of – Event Time: E the situation described, relative to some – Reference Time: R other time (e.g., moment of speech) SI had [mailed the letter]E [when John came & told me the news]R E < R < S E R S
time • The concept of ‘time point’ is an abstraction –- it can map to an interval • Three temporal relations are defined on these time points PAST PRESENT FUTURE – at, before, after • 13 different relations are possible George admires Adolf. George admired Jesus. Tense as Anaphor: Reichenbach Reichenbachian Tense Analysis • Tensed utterances introduce references to 3 ‘time points’ • Tense is determined by – Speech Time: S Relation Reichenbach’s English Tense Example relation between R and S – Event Time: E Tense Name Name – R=S, R I ate a peach going to sleep
Tense as Operator: Prior Aspect
Relation Reichenbach’s PRIOR English Tense Example • Free iteration Tense Name Name • Two Varieties E Grammatical Aspect Aktionsarten • STATIVES know, sit, be clever, be happy, • ACCOMPLISHMENTS build, cook, – can refer to state itself (ingressive) John destroy knows , or to entry into a state (inceptive) – culminate (telic) • Perfective – focus on situation as a whole John realizes – x Vs for an hour does not entail x Vs – John built a house – *John is knowing Bill, *Know the answer, for all times in that hour *What John did was know the answer – X is Ving does not entail that X has • ACTIVITIES walk, run, talk, march, paint Ved. built.a.h – if it occurs in period t, a part of it (also an – John booked a flight in an hour, John activity) must occur for every/most sub- stopped building a house periods of t • ACHIEVEMENTS notice, win, blink, – X is Ving entails that X has Ved find, reach – John ran for an hour,*John ran in an hour – instantaneous accomplishments • Imperfective – focus on internal phases of – *John dies for an hour, *John wins for an situation hour, *John stopped reaching New York – John was building a house Telic Dynamic Durative E.g. Stative - - + know, have Activity - + + walk, was building.a.h paint Accomplish + + + destroy, ment build Achieveme + + - notice, nt win Different types of tense Different types of tense systems across languages systems across languages • A much richer distinction: • Using verbal inflection: – ChiBemba (Bantu language): – Languages with a two-way contrast: For past: • English: Past (before the moment of speaking) vs. Nonpast • Remote past (before yesterday) Ba-àlí-bomb-ele ‘they worked’ past -ed: She worked hard. • Removed past (yesterday) Ba-àlíí-bomba ‘they worked’ nonpast (unmarked): We admire her. I will leave tomorrow. • Near past (earlier today) Ba-àcí-bomba ‘they worked’ • Dyirbal (Australian language): Future vs. nonfuture: • Immediate past (just happened) Ba-á-bomba ‘they future -ñ: bani-ñ ‘will come’ worked’ nofuture -ñu: bani-ñu ‘came, is coming’ For future: – Languages with a three-way distinction: • Immediate future (very soon) Ba-áláá-bomba ‘they’ll work’ • Catalan, Lithuanian: Past vs. Present vs. Future • Near future (later today) Ba-léé-bomba ‘they’ll work’ (Cat.) past: treball-à. (Lit.) Dirb-au. ‘I worked’ • Removed future (tomorrow) Ba-kà-bomba ‘they’ll work’ present: treball-a. Dirb-u. ‘I work’ • Remote future (after tomorrow)Ba-ká-bomba ‘they’ll work’ future: treball-arà. Dirb-siu. ‘I will work’ Different types of aspect Aspect systems across languages • Internal temporal organization of the situation described by an event. • Some languages use auxiliaries and particles associated • Most common: with the verb: – Perfective: Situation viewed as a bounded whole. English: – Imperfective: Looking inside the temporal boundaries of the – Perfective: have + Past Participle I have eaten. situation. – Progressive: be + Present Participle I am eating. • Habitual – Habitual: use to + Base form I used to sing. • Progressive Catalan: • Other related aspectual distinctions: – Habitual: soler + Infinitive – Iterative: The action is repeated. Sol parlar. ‘She generally talks.’ – Inceptive: The action is began. Solia cantar. ‘She used to talk’ – Inchoative: Entering into a state. – Iterative: anar(past) (‘to go’)+ Present Part Va tornant ‘She keeps coming back’ gopast coming_back Different types of aspect Tense and Aspect systems across languages • Aspect and Tense generally cross-classify: – Russian: • Other languages use a derivational component: • Present: Russian: by means of a system of verbal prefixes – Only imperfective: ˇcitáju ‘I read’ – Imperfective: simple verbs Ja ˇcitál ‘I was reading’ • Past: – Perfective: prefixed verbs Ja proˇcitál ‘I (did) read’ – Imperfective: Ja ˇcitál ‘I was reading’ – Perfective: Ja proˇcitál ‘I (did) read’ Finnish: by means of the case of the object • Future: – Imperfective: ?? – Perfective: Hän luki kirjan(acc.) ‘He read the book’ – Perfective: Ja proˇcitáju ‘I shall read’ – Imperfective: Hän luki kirjaa(part.) ‘He was reading the book’. Basic meaning: only part of the object being referred to is affected by the situation. Tense and Aspect An interesting case • Tense and Aspect in 2 different creoles, – Basque: evolved independently from each other: • Present: Hawaiian Creole Haitian Creole – Imperfect (Gerund + Present tense auxiliary) ekartzen du ‘he is Base Form He walk Li maché bringing it’ (he walks, he walked) – Perfect (Past Participle + Present tense aux.) ekarri du ‘he has Progressive He stay walk L’ap maché brought it’ he is walking, he was (Li ap maché) • Past: Pwearlfkeicntgive He bin walk Li té maché – Imperfect (Gerund + Past tense aux.) ekartzen zuen ‘he brought, used he has walked, he had to bring’ Pwearlfkeecdtive Progressive He bin stay walk Li t’ap maché – Perfect (Past Participle + Past tense aux.) ekarri zuen ‘he brought, had (he has/had been walking) (Li té ap maché) brought’ Irreal He go walk L’av maché • Future: (he would walk, he will – Simple (Future Participle + Pres. tense aux.) ekarriko du ‘he Iwrraelkal) Progressive He go stay walk L’av ap maché will bring it’ (he would/will be walking) (Li av ap maché) – Past Future (Future Participle + Past tense aux.) ekarriko zuen ‘he would Irreal Perfective He bin go walk Li t’av maché bring’ (he would/will have (Li té av maché) Iwrraelkale dp)erfective He bin go stay walk Li t’av ap maché Progressive (Li té av maché) he would/will have been walking Embedded tenses in English Constraints on interpretation Three interpretations of embedded tenses: • Tense interpretation displays both structural restrictions and Absolute: embedded tense is independent of lexical preferences main clause tense Yesterday John saw a girl who was running this morning. Relative clause interpretation: t see running c At the party John danced with the woman (previously/later) he ate dinner This morning John saw a girl who was running yesterday. with. running see tc At the party John met the woman he married Anaphoric: embedded tense is anaphoric on the main clause tense Complement clause interpretation Yesterday John saw a girl who was running. see t At the party John said that he (previously/??later) ate dinner with a certain running c woman. Relative: embedded tense is interpreted with respect to the main clause tense Tomorrow John will see a girl who was running earlier. tc running see Crosslinguistic variation Crosslinguistic variation Variation in relative clause interpretation Variation in complement clauses interpretation • Japanese Mariko-wa naiteiru otokonoko-ni hanasikaketa • Japanese Mariko-TOP cry-teiru-PRES boy-to talk-PAST Bernhard-wa Junko-ga byookida to it-ta “Mariko talked to the boy who is/was crying” B.-TOP J.-NOM sick-PRES comp say-PAST “Bernhard said that Junko was sick” • Russian • Russian Maˇsa videla ˇceloveka, kotoryj placet. Maˇsa skazala, cto Vova spit. Masha see-PAST-IMP man who cry-PRES Masha say-PAST-PERF that Voval sleep-PRES “Masha saw a/the man who is crying” “Masha said that Vova was sleeping” Embedded tenses cross-linguistically The Conceptual and Linguistic Basis Relative Complement • TimeML presupposes the following temporal entities and Clause Clause relations. English absolute relative relative anaphoric • Events are taken to be situations that occur or happen, punctual anaphoric or lasting for a period of time. They are generally expressed by Japanese absolute relative means of tensed or untensed verbs, nominalisations, adjectives, relative predicative clauses, or prepositional phrases. Russian absolute relative anaphoric • Times may be either points, intervals, or durations. They may be referred to by fully specified or underspecified temporal Via cross-linguistic investigation a picture of embedded tenses expressions, or intensionally specified expressions. emerges: • Relations can hold between events and events and times. They • Absolute tense is limited to relative clauses can be temporal, subordinate, or aspectual relations. • Relative tense is predominant in complement clauses Allen (1984) Allen’s 13 Temporal Relations Temporal Logic A • Time primitives are temporal intervals. B A is EQUAL to B B is EQUAL to A • No branching into the future or the past A A is BEFORE B B B is AFTER A • 13 basic (binary) interval relations •[b,a,eq,o,oi,s,si,f,fi,d,di,m,mi], A A MEETS B (six are inverses of the other six) B B is MET by A • Supported by a transitivity table that defines the conjunction of A A OVERLAPS B any two relations. B B is OVERLAPPED by A A A STARTS B • All 13 relations can be expressed using meet: B B is STARTED by A •Before (X, Y) # $Z , (meets(X, Z) % (meets (Z, Y)) A A FINISHES B B B is FINISHED by A A A DURING B B B CONTAINS A Allen’s Temporal Ontology Situation Type: Formal Constraints • Homogeneity – All subevents of P are also of P (downward entailment) • Properties hold over every subinterval of an interval • though only down to a minimal size – The sum of all subevents of P are also of P (upward —> Holds(p, T) e.g., ”John was sick for a day." entailment) • Subinterval Property • Events hold only over an interval and not over any subinterval of it. – Activity: x Ps for t => x P’s for all subintervals of t. excluding —> Occurs(e, T) e.g., ”Mary wrote a letter this afternoon." those below a minimal size and excluding certain gaps • For is downward entailing, but the maximal interval is more • Processes hold over some subintervals of the interval they occur in. felicitous —> Occuring(p, T) e.g., ”Mary is writing a letter today." – Accomplishment: x Ps in t => there is a subinterval t’ of t in which Become(x, P) is true • In is upward entailing, but the minimal interval is more felicitous • In-adverbials apply to quantized event predicates – A predicate is quantized iff whenever it applies to e it doesn’t apply to subparts of e Event Structure Theories of Event Structure • Quantification over events as individuals: I.e., Davidson (1967): Proposes individuation over events. events as first-order objects. Kamp (1968): Formal Model for tensed events, extending • Finer-grain representation than Prior’s tense Prior’s Tense Logic to predicates. logic. • Allows representation of word-based Moens and Steedman (1988): Finite-state model of event causality. phases. • Simplifies reasoning with identity and Pustejovsky (1991): Phrase structure model for subevent overlap relations. semantics for word meaning. McCarthy and Hayes (1969) Hayes 1985 The Situation Calculus Histories in Naïve Physics • Represents actions and their effects on the world • A history is an entity that incorporates time and space • An object O in a situation s is the intersection of the • The world is represented as a set of states. situation with the object’s history • Fluents are time-varying properties of individuals. • Permanent locations are bound spatially, but are restricted • Actions are functions that map states to states. temporally • Situations are unbound spatially, but are limited • Used for multiple tasks, especially planning temporally by surrounding events • Major problems: • Most objects are between these two extremes – Concurrent actions cannot be represented • Events are instantaneous – No duration of actions or delayed effects • Episodes have a duration • The history of an object is described over time Kowalski & Sergot (1986) Properties of Events Event Calculus • Developed for updating databases and for narrative understanding Events have parts: • Based on the notion of an event and its descriptions (relationships) The rock broke the window. $e1$e2[action(e1,rock,window) & • Relationships are ultimately over time points broken(e2,window) & e1 • Udates can only add; deletions add new information about Actions have consequences: the end of the period of time over which the old relationship holds Mary arrived in Boston. • Uses nonmonotonic, default reasoning since relations change as new $e1$e2[action(e1,mary,boston) & information arrives (a new event can signal the end of an old one) in(e2,mary,boston) & • Allows partial description of events, using semantic cases e1 • Defined and interpreted as Horn clauses in Prolog