The ASL Perfect Formed by Preverbal FINISH

The ASL Perfect Formed by Preverbal FINISH 1 Introduction1 It is well known that the American Sign Language (ASL) sign FINISH has various lexical, idiomatic, and grammatical functions. This paper aims to investigate in more detail one of the known grammatical uses of this sign: as a preverbal perfect marker. The ASL perfect has received scant analysis beyond the acknowledgment of its existence. Further consideration of this construction is clearly warranted; the perfect is a slippery aspectual category whose meaning, restrictions, and prevalence vary significantly across languages. The one proposed analysis (Rathmann 2005) claims that preverbal FINISH only marks perfect aspect, but does not mark perfective. The goal of this analysis is to characterize the ASL perfect formed by preverbal FINISH by providing a unified account that explains both the allowable perfect readings and the pattern of compatibility across various types of predicates. On the basis of the fact that, when used as a marker of perfect aspect, preverbal FINISH only allows a bounded reading of an event or situation, but is compatible with (some) otherwise unbounded events, we conclude that preverbal FINISH contributes boundedness to semantically unbounded predicates: FINISH marks both perfect and perfective aspect. The incompatibility of preverbal FINISH with individual-level states and (most) temporal aspectual marking is due to the incompatibility of these categories with temporal bounding. The organization of the paper is as follows: Section 2 provides a brief overview of the expression of tense, aspect, and temporal interpretation in ASL. Section 3 reviews crosslinguistic perspectives on the aspectual categories relevant to the current investigation: viewpoint, perfect, and situation aspect. The many different uses of the sign FINISH are summarized in Section 4, and Section 5 provides data and analysis of 1 The research for this project has been supported in part by funding from the National Science Foundation 1 the specific grammatical use of interest in this study, the preverbal perfect marker, based on the crosslinguistic groundwork laid in Section 3. 2 The Expression of Tense and Aspect in ASL 2.1 Controversy about the existence of tense in ASL Early analyses of American Sign Language claimed that the language lacks grammatical tense, instead relying on temporal adverbs and a rich aspectual system to express the time of an event or situation, not unlike Mandarin Chinese (e.g., Friedman 1975; Fischer and Gough 1972, 1979; Padden 1983, 1988). Friedman (1975) claimed that in ASL, the time of the speech act is not maintained when referring to another point in time. Once one temporal reference is established, Friedman claimed, no other time can be referred to until the temporal reference is reset. In the Reichenbachian (Reichenbach 1947) tense system of Speech time (S), Reference time (R), and Event time (E), where tense expresses the relationship between S and R, Friedman’s analysis implies that S is not a grammatical reality in ASL, leaving speakers with only R and E as time points expressible by ASL’s grammar. This entails that tense marking is impossible in ASL, as the relationship of S and R, which establishes tense, cannot be expressed if S does not exist. This earlier claim that ASL lacks tense has been refuted by Neidle, Kegl, MacLaughlin, Bahan, and Lee (2000, henceforth NKMBL), who identify seven lexical markers of tense and aspect that appear in the same position as modals; these markers are listed in Appendix I. They speculate that the earlier claims that ASL lacks tense marking may be attributable to the fact that six of the seven tense markers are related to temporal adverbials, with the exception of the loan sign of habitual past, #EX2. The tense/aspect markers differ from their related adverbials both in articulation and in syntactic distribution. The tense markers have a fixed path length, whereas the related adverbials can be modified to indicate temporal distance from the present. In terms of distribution, the tense markers can only occur in preverbal position or in sentence-final tags, but temporal adverbials are acceptable at the start or the end of the sentence, though not in tags. The clearest and least controversial example of a tense marker in ASL is the sign FUTURETNS, sometimes glossed as WILL. 2.2 Interpretation in the absence of lexical tense markers Tense marking as identified by NKMBL is not present in every ASL sentence. With temporal adverbials, sentences that are unmarked grammatically for tense can be interpreted as referring to past or future events, illustrated by (1) and (2). (1) LAST-WEEK JOHN GO NEW-YORK “Last week John went to New York.” 2 This sign #EX, used by some but not all signers, is a marker of both tense and aspect: it marks past tense but can only be used to mark habitual events or situations. It has an English counterpart in the periphrastic past/habitual marker “used to” 2 (2) NEXT-WEEK JOHN GO NEW-YORK “Next week John is going to New York.” This present-tense expression of non-present events also occurs in other languages that indisputably mark tense, such as English, Spanish, and French (NKMBL 2000); this use, illustrated in (3) and (4), is often called the narrative or historical present. (3) So, last night, I buy scalped tickets to the game, but when I try to get into Fenway the ticket-taker says they’re no good. (4) Next year I take off for college. I can’t wait. NKMBL argue that ASL sentences without lexical tense markers are syntactically present tense, with non-present reference, just like the English sentences above. Temporal adverbials set the time of reference. 2.3 Aspectual inflection on ASL verbs ASL has a large inventory of aspects that are expressed through verbal/predicate inflections “marking distinctions of temporal recurrence and duration” (Klima and Bellugi 1979:292). These exploit rate, tension, and repetition to mark the temporal structure of an event or recurrence of an event (Klima and Bellugi 1979). Klima and Bellugi provide a basic meaning for six of these inflections without addressing the exact semantic contribution of each inflection. These temporal inflections are distinct from inflections for subject and object agreement, reciprocity, distribution, or adverbial modification, which tend to exploit spatial rather than temporal patterning in their expression. Klima and Bellugi use the verb LOOK as the base form for their examples of temporal expression. 2.3.1 Protractive The protractive is characterized by no movement: A verb under this inflection is simply a long hold. The translation provided for LOOKprotractive is “to stare at (uninterruptedly)” (Klima and Bellugi 1979:292). 2.3.2 Incessant Short, tense iterated movement marks the incessant aspect, which (as the name implies), conveys that an event occurs incessantly over a period of time. 2.3.3 Durational The durational aspect focuses on an event or situation’s durational characteristics, extending the implied length of the event. Circular, reduplicated movement marks durational aspect. 3 2.3.4 Habitual The regular repetition of an event is marked by rapid, non-tense repetitions of the sign; LOOKhabitual can be translated as “to watch regularly.” 2.3.5 Continuative Large, smooth elliptical movements add an approximate meaning of “for a long time” to a single event or situation. 2.3.6 Iterative With a meaning of “over and over,” tense, end-marked movements are repeated with a slow, elliptical return. 2.3.7 Semantic contribution of the inflectional aspects Based on the form of the inflections and the associated meanings, Klima and Bellugi posit that these six aspects actually involve three basic temporal inflections, which vary according to whether or not the base verb is punctual. They group the protractive and the incessant, the durational and the habitual, and the continuative and the iterative as pairs of related inflections. As mentioned in the introduction to this section, Klima and Bellugi do not provide a semantic account of the meaning that these aspect markers contribute to the sentence. As observed by Rathmann (2005), both bounded and unbounded verb phrases (VPs), when inflected with the temporal aspect markers, are interpreted as present and ongoing in the absence of evidence to the contrary, such as a temporal adverbial or a tense marker. Thus they have an unbounded reading in ASL, which implies that these are markers of imperfective aspect. This is also supported by Klima and Bellugi’s characterization of the temporal aspect markers as providing information about the temporal nature or recurrence of an event, which fits the definition of imperfective aspect (presented in Section 3.1.2). 2.4 Default interpretation of tense in ASL Even without tense markers or temporal adverbials, some ASL sentences may be interpreted as past on the basis of aspectual properties of the verb phrase, while others are interpreted as present. Rathmann (2005) explores default temporal readings of ASL sentences at great length. He argues that—absent any other temporal information— bounded ASL sentences are interpreted as having occurred in the past, while unbounded sentences are interpreted as occurring in the present. When unmarked by tense, adverbial reference, or viewpoint aspect, telic and durationless events are interpreted as bounded (past, by default), while activities and states give an unbounded reading (default present reading). In addition to giving information on the internal nature of the event, aspectual inflections such as those described in Section 2.3 can also affect the default temporal interpretation by presenting a semantically bounded event as unbounded, thereby causing it to be interpreted as a 4 present event. Thus IX-1p3 GO CHURCH is an accomplishment, interpreted in the past (“I went to church”), but IX-1p GOhabitual CHURCH, lacking explicit bounds, is interpreted as present (“I go (regularly) to church”) (Rathmann (2005).4 A perfective gives a bounded reading to both telic and atelic VPs, locating the sentence in the past. 2.5 Summary Previous claims that ASL lacks tense, and by implication, lacks a grammatical realization of speech time, S, have been refuted by NKMBL 2000. The fact that ASL does not overtly mark tense on every sentence, combined with the similarity between the tense markers and their related temporal adverbials, may have contributed to the misconception that ASL lacks tense altogether. In addition to tense marking, ASL possesses a rich system of inflectional aspect markers, which may be best characterized as imperfective aspect morphemes. When tense is not overtly marked in ASL (which is frequently the case), temporal adverbials, situation aspect, and viewpoint aspect can all contribute to interpretation of tense. 3 Crosslinguistic Perspectives on Aspect 3.1 Viewpoint Aspect: the perfective/imperfective distinction Viewpoint aspect expresses how the temporal dynamics of an event are presented. It concerns the internal temporal constituency of a situation (Comrie 1976:5). The major contrast is between perfective and imperfective; the imperfective can be subdivided into various categories. Perfective aspect presents an event as a single, bounded whole, while the imperfective highlights the internal temporal structure of an event without referencing its endpoints. Even though any given event most likely has endpoints, the endpoints simply are not relevant, since imperfective expresses an event (or regular repetition of events) in progress—not necessarily in the present, but at some time. (5) Perfective: John ran to the store5 I F _______[EEEEEEE]_______ 3 In glossed examples, pronouns are represented by IX- (“index,” as it is normally the index finger that points to the location in space associated with the intended referent), followed by the person features. This notation reflects the indexical nature of the ASL pronominal system. Similarly, POSS- is the gloss used for possessives, and IX-loc for locative adverbials (also articulated with the index finger). 4 Some evidence supports the possibility that either manual or nonmanual inflection can convey tense information, as well as aspectual information; Jacobowitz and Stokoe (1988) claim that forward or upward extension at the wrist, elbow, shoulder and/or neck indicates future tense, and backward or downward flexion of these joints marks past tense. Carol Neidle and Michael Schlang (personal communication) have found minimal pairs of sentences contrasting in temporal reference that differ only in facial expression. The distribution of these temporal indicators in relation to the other ways to convey temporal information in ASL has yet to be explored. 5 In these schematics, I is the initial point of the event, F is the final point of the event, EEE is the part of the event presented by viewpoint aspect, and [] are the bounds of the event in the perfective viewpoint. 5 (6) Imperfective: John was running to the store I F _________…EEE…_________ Viewpoint aspect is independent of tense. Aspectual distinctions can be made in past, present, and future tenses (though not all languages make the distinctions in all three tenses). Some languages do this transparently, such as English in its use of the progressive, where tense is marked on the auxiliary be and progressive (imperfective) aspect on the verb by -ing. Other languages—Spanish, for one—conflate the marking of both tense and aspect into one unanalyzable morpheme. (7) Juan jug-ó al futbol John play-past/pfv to-the-soccer “John played soccer.” (Spanish) (8) Juan jug-aba al futbol John play-past/impfv to-the-soccer “John was playing soccer.” (Spanish) 3.1.1 Perfective aspect The perfective aspect presents an event or situation as a whole (Comrie 1976:3). English sentences like John went to the store, Pat ran in the park, and Mary ate the sandwich are perfective sentences. Individual phases within the event are not distinguished, and the start, middle, and end of the event must be included implicitly when an event or situation is presented in perfective aspect. For telic verb constellations, this requires that the event reach its natural end, and for atelic events, termination of the event or situation is necessary. The event is bounded. While perfective aspect makes no grammatical reference to the internal dynamics of an event, adverbials of duration or manner may give implicational information about the event’s internal dynamics. 3.1.2 Imperfective aspect The imperfective aspect highlights the internal structure of either an event or the repetition of an event; crucially, imperfective events are unbounded. As mentioned above, the imperfect can be broken down according to further distinctions that some (but not all) languages encode grammatically. Three aspects widely attested crosslinguistically are the Progressive, the Continuous Nonprogressive, and the Habitual (Comrie 1976). The Continuous Nonprogressive is sometimes simply termed ‘Imperfect’ (e.g., Smith 1997). Events lacking duration—achievements and semelfactives—do not have any internal structure to which the imperfective can refer. In these cases, the imperfective instead refers to the internal structure of the repetition of events. For example, hit the jackpot describes an instantaneous event. When speaking of a one-time hit of the jackpot, it would be odd to say I was hitting the jackpot (when Mary came over). Rather, that sentence implies that there was a consistent string of individual wins; the imperfective can refer to the internal dynamics of such a string, because the string itself 6 has duration, a start, a middle, and an end. The Habitual aspect refers to a situation that is a characteristic feature of an extended period, either the regular repetition of individual bounded events (I hit the jackpot every week back then) or a continuous characteristic (My uncle’s casino stood at the corner of Washington and Fourth) . 3.1.3 Summary Viewpoint aspect either shows an event as a bounded whole (perfective aspect) or highlights an event’s internal temporal dynamics without referencing the endpoints of the event (imperfective aspect). Aspectual distinctions can be made independently of tense, since tense dictates the relationship of reference time to the time of utterance, while viewpoint aspect portrays the relationship of an event with its own endpoints. 3.2 The perfect aspect and its subtypes across languages In contrast to the perfective/imperfective aspectual distinction, which concerns the internal temporal dynamics of an event, the perfect expresses the relationship between an event E and reference time R. The perfect aspect can combine with past, present, or future tenses, which specify the relationship between speech time, S, and R. Perfect aspect is independent of tense, but can interact with it. Thus we can form the past perfect, I had run a mile, the present perfect, I have run a mile, and the future perfect, I will have run a mile (Comrie 1976). There are three basic accounts of what the perfect is. In the Result State Theory (Giorgi and Pianesi 1998), at R, the result state of E is true, and because of this, E is implicitly established as having occurred before R. Anteriority Theory holds that the perfect specifically establishes that E precedes R (Reichenbach 1947 and others). According to the Extended Now theory, E occurs within an interval whose right boundary is R (McCoard 1978 and others). Iatridou, et al. (2001) adopt a version of the Extended Now theory: a perfect event exists within a perfect time span with a left boundary set by adverbials (since 1998, for example)—or an indefinite left boundary if there is no adverbial—and a right boundary set by tense, i.e., it coincides with R. Because the right boundary of the perfect time span is R and the event is included in the span, it occurs (or at least begins, in the case of universal perfects) before R. According to Iatridou, et al., anteriority of E relative to R is not directly expressed by the perfect, but rather implied by it. Their definition, which we will use here, is provided in (9). (9) Perfect: There is an interval (the perfect time span) in/throughout which there is a bounded/unbounded eventuality. (Iatridou, et al. 2001:212) The perfect is only used when E or the result state of E is somehow relevant to the situation or events at R. Comrie 1976 identifies four relevance conditions where the relationship between a past event and the time of reference merits the use of the perfect; these four types of perfect are listed below. Various authors have identified the same four types of perfect, but in some cases have used different terminology. The four descriptions in 3.2.1 through 3.2.4 include some of the other terms that have been used for each type. 7 3.2.1 Perfect of result In this use, also known as the resultative perfect, a present state is expressed as the result of a past event: The baby has woken up means that the baby is now awake, whereas past tense the baby woke up makes no such claim (though it is certainly possible that the baby is, in fact, awake).6 Some languages use a “perfect of result” construction where others would use a stative predicate. For example, English would allow either the baby has woken up or the baby is awake, but other languages are more specific in their selection of one over the other. Swahili is one language that uses a perfect of result where English would use a stative predicate, at least in some contexts, as shown in (10) and (11). (10) a-me-choka “He is tired.” (Lit. “He has got tired”) (11) a-me-simama “He is standing.” (Lit. “He has stood up”) (Ashton 1947:37 via Comrie 1976:57) 3.2.2 Experiential perfect The experiential, or existential, perfect expresses that the relevant situation E has occurred at some point previous to R without necessarily specifying the exact time of the event. I have been to Fenway Park, John has eaten peanuts before, and I have fixed flat tires a few times are all experiential perfects. Comrie notes that most European languages make no grammatical or morphological distinction between the perfect of result and the experiential perfect, but observes that some languages mark the experiential perfect with a specific morpheme. Mandarin Chinese is an example of such a language; the toneless suffix –guo is used exclusively for the experiential perfect (Comrie 1976:59). 3.2.3 Perfect of recent past Sometimes the perfect can be used simply on the basis of proximity of the past situation to the reference time. This perfect of recent past, or “hot news” perfect, does not actually require that the event be recent, but rather that it be new information of present relevance; in most instances, it is a recent event: The president has died! or A tree has fallen on the neighbor’s roof. Comrie (1976) imagines a clever situation in which the perfect can be used to deliver “hot news,” despite a longer temporal interval between the event and reference time: to say The Second World War has ended in the year 2007 would be infelicitous if it were directed to a passerby on the street, but perfectly appropriate if talking to someone who had been stranded on a desert island since 1944. 6 (1a) “The baby woke up; we’re ready to go now.” (1b) “The baby has woken up; we’re ready to go now.” (2a) “The baby woke up; I rocked him back to sleep.” (2b) #The baby has woken up; I rocked him back to sleep.” 8 3.2.4 Universal perfect This perfect, also known as the perfect of persistent situation or continuative perfect, has been defined in subtly but crucially different ways. What has been agreed is that it expresses an event that started before R but continues into the present. For Comrie (1976), the universal perfect is any situation that continues into the present (or R), regardless of the boundedness of the predicate. Thus events that pragmatically continue into the present but are bounded either grammatically or by an adverb of specific duration, such as I’ve shopped there for years (Comrie 1976:60), are considered universal perfects. In contrast, Iatridou, et al. (2001) state that one of the defining characteristics of the universal perfect is the perfect event must be unbounded; an eventuality is bounded if it is contained in an interval, and unbounded if it is ongoing throughout the interval. There can be bounded events that pragmatically continue up to R, as in the English perfect+perfective sentence I have run for five hours, and I’m not stopping now, but according to Iatridou, et al., these are not semantically universal perfects, despite the fact that the truth conditions are very similar to those of a universal perfect. 3.2.4.1 Identification of this construction In the current analysis we will adopt Iatridou, et al.’s definition of the universal perfect, as the unboundedness requirement of their definition provides a distinction that we will see later is relevant to the ASL perfect. Thus, the universal perfect refers to a situation that began before reference time but necessarily overlaps with reference time, such as I have known Bill since college; For three years, John has been living in Boston; or I have been building the house since May. Some languages, including French and Spanish, have perfect aspect available but express persistent situations in the simple present instead of in the perfect, as in (12) and (13). (12) Jean travaille depuis trois ans/1989. (French) Jean work.3p.sg.pres since 3 years/1989 “Jean has been working for 3 years/since 1989.” (Lit. John works since three years/1989.”) (13) Juan vive en Madrid desde hace muchos años. John live.3p.sg.pres in Madrid from since many years “John has been living in Madrid for many years.” (Lit. John lives in Madrid for many years.”) (Spanish) In the universal perfect, the temporal dynamics differ slightly from the three types of perfect discussed so far, in that E overlaps with R, whereas in the other perfects, E precedes R. In Iatridou, et al.’s (2001:191) terms, the eventuality of a universal perfect is unbounded and holds throughout the interval specified by the adverbial and through its endpoints (italics mine), entailing that the eventuality holds at reference time, which is the right boundary of the perfect time span specified by the adverbial. In any of the other 9 three perfects, this is not true; the event time is included within the perfect time span, but not throughout it. Examples (14) and (15) illustrate this schematically. (14) Experiential perfect (or resultative or recent past) I have been in Boston since Tuesday (Experiential reading: I stopped there on Wednesday for lunch on my way to Cape Cod.) R Tuesday I F [EEE]__________ ßPerfect Time Spanà Now (15) Universal perfect I have been in Boston since Tuesday (Universal reading: I arrived on or before Tuesday, and have been here ever since; Tuesday, the left boundary of the perfect time span is required to be included within E.) R …EEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE…_____ Tuesday ßPerfect Time Spanà Now In order for a universal perfect to occur, the perfect event or situation must be unbounded, i.e., incomplete telic events, unterminated atelic ones, or states. So a stative predicate (verb or adjective) or an event verb in imperfective viewpoint is required to form a universal perfect (Iatridou, et al. 2001). The fact that the universal perfect is never available without adverbial modification or contextual support,7 along with the fact that some languages that have a perfect do not allow any universal perfect readings (Modern Greek is an example, Hedin 1987) has led some to claim that the universal perfect is not a core meaning or use of the perfect (Jesperson 1924 and Comrie 1976, via Iatridou, et al. 2001:191). The universal reading has been said to be just a subcase of the experiential perfect, attributed to vagueness of the duration of the unbounded predicate. Iatridou, et al. refute this claim by pointing out constructions in English where only a universal reading survives, meaning that the universal cannot be a subcase of the existential, but rather is truly a semantically-based distinction. They also demonstrate differences in the tense of embedded clauses under existential perfects (morphological past is allowed even if the embedded event is contemporaneous with matrix clause event) as compared to universal perfect (morphological past requires that the embedded clause 7 Mary has been sick, unmodified by adverbials or context, can have an experiential reading, but it can also have a reading where she is still sick at utterance time. However, the utterance does not entail this. It could be continued with either …and still is or …but is fine now. Iatridou, et al. conclude that this is not a true universal perfect, but rather closest to a perfect of recent past. With the right contextual support, though, a sentence without adverbials can yield a universal perfect: Stop telling me how to do everything! I’ve been working here and I know how to do that! 10 event be prior to the matrix clause event). Iatridou, et al. argue that the crosslinguistic variation among languages with respect to the existence or nonexistence of the universal perfect is due to the morphosyntactic features of the perfect event: the universal perfect is available only if the event under the perfect can be unbounded. 3.2.4.2 Adverbials and the universal perfect As mentioned above, an adverbial is usually necessary to obtain a universal perfect reading. Some adverbials, such as always, allow only universal readings with the perfect. Others, such as since and for, allow either a universal reading or an existential one. Still others, e.g., four times, allow only existential readings. Adverbials can occur on two levels that differ in scope: perfect-level adverbs are higher in the syntactic structure than eventuality-level adverbs (Iatridou, et al. 2001 following Dowty 1979 and Vlach 1993). Perfect-level adverbs specify the length of the perfect time span, whereas eventualitylevel adverbials specify the length of the event or situation itself. In addition to this scope distinction, adverbials referring to intervals can be either durative or inclusive. Perfect-level durative adverbials give a universal perfect, because they require universal quantification over the interval (Iatridou, et al. 2001). Perfect-level inclusive adverbials assert inclusion in the interval, yielding an existential perfect. Some adverbials, like since, can be either durative or inclusive, and therefore yield either universal or existential perfects. For- adverbials, on the other hand, are always durative, but are universal when perfect-level, experiential when eventuality-level. This can lead to ambiguity: I have been playing baseball for three hours can be either experiential or universal, as illustrated in (16) and (17). (16) I have [been playing baseball for three hours] … (eventuality-level, experiential) R I ß3 hoursà F [Playing baseball]___ ßPerfect Time Spanà Now (17) I have [been playing baseball] for three hours (perfect-level, universal) R …Playing baseball… 3 hrs ago Now [Perfect Time Span] The ambiguity disappears when the for-adverbial is sentence-initial, as in (18). Only the perfect-level adverbial for- is possible in this position. (18) For three hours, I have [been playing baseball] (perfect-level, universal) 11 As mentioned before, universal perfects require unboundedness, which entails that the eventuality holds throughout the entire interval and is subject to the subinterval property.8 When a for- adverbial is eventuality-level as in (16), the event does not have the subinterval property: [playing baseball for three hours] is not true of every subinterval of the interval; one cannot say he has been [playing baseball for three hours] in the second hour of doing so. In contrast, the perfect-level for- allows the interval to maintain the subinterval property, because [playing baseball] is true at every subinterval throughout the interval. 3.2.5 Combination of the perfect with other aspects In languages that possess both a perfect/non-perfect distinction and other aspectual distinctions, such as perfective/imperfective, the two types of aspect may be combined so that perfect+perfective and perfect+imperfective can be expressed. This is demonstrated by the compatibility of the English perfect with the progressive, which is an imperfective aspect (Comrie 1976), I have been studying since Tuesday, and the perfect with the perfective, John has studied chapter three. The perfect+imperfective combination is frequently used for universal perfects such as John has been watching football since nine this morning, but can also be used for experiential or recent past/hot news perfects as well, such as Have you ever been watching TV when the tube exploded? (Comrie 1976) or Johnny has been stealing cookies from the pantry! 3.2.6 Summary The perfect aspect places an event E in or throughout a perfect time span whose right boundary is R (set by tense) and whose left boundary is infinite unless specified by a perfect-level temporal adverbial. Crosslinguistically, either bounded or unbounded eventualities can be expressed in the perfect, though some languages only allow bounded events in the perfect. There are four types of perfect: the experiential, the perfect of recent past, the resultative perfect, and the universal perfect. The first three typically involve bounded events (but can tolerate unbounded ones). However, on the universal perfect reading, the perfect event is necessarily unbounded, and the sentence includes a perfect-level durative adverbial or is spoken in a context where the event clearly overlaps with reference time. 3.3 Situation Aspect Situation aspect refers to the temporal properties intrinsic to a verb constellation—a verb and its arguments—in contrast to viewpoint aspect, which portrays that event or situation as a part or a whole. There are five basic situation types, differentiated by the combinations of three temporal features: dynamism, telicity, and duration. The dynamic/stative distinction separates events, which require some energy input to happen, from states, which are undifferentiated situations that do not require any energy input to persist. Telic events have a natural endpoint of some change of state and cannot continue once that endpoint is reached. Once someone drinks a gallon of milk, for example, he 8 The subinterval property holds of an interval iff the eventuality holds at every subinterval of that interval. 12 cannot continue to drink that gallon of milk. Telic events are countable: If they take an object, there will be a specified quantity of that object. Atelic events, on the other hand, are processes that lack an outcome, so the endpoint is arbitrary, simply the time when the process is stopped. Events that are durative occur over time and have a beginning, middle, and end, whereas instantaneous events have no internal structure or duration; they happen in an idealized instant (Smith 1997). Figure 3-1 illustrates the temporal features of the basic event types. State Activity Accomplishment Semelfactive Achievement Dynamic Durative Telic + + + + + + + + + Figure 3-1: Temporal features of situation types (Smith 1997) 3.3.1 Dynamic Event Types 3.3.1.1 Activities Activities are dynamic, durative, atelic events. They have an arbitrary endpoint, so activities stop but do not finish (Smith 1997). Example activities are chat, laugh, run in the park, and drink water. There are three types of activities: continuous processes that are at least potentially unlimited, like run in the park; activities built by many internal stages, e.g., a telic verb with an object of unspecified quantity (eat cherries); and coerced activities: instantaneous events that are iterated (the series of iterations has duration). 3.3.1.2 Accomplishments Accomplishments are dynamic, durative, telic events; they involve a process and an outcome, which is to say that they have natural temporal bounds. Eat a sandwich, walk to the store, and run thirty miles are accomplishments. Smith also includes activities with an adverbial of specific duration, such as run for an hour, in the category of coerced accomplishments. Although the verb constellation does not entail an outcome, these events are bounded by the durative adverbial because now the event has specific start and end points. 3.3.1.3 Semelfactives Semelfactives are dynamic, instantaneous, atelic events. They are single-stage events with no result, such as knock, flicker, or cough. Because they have no internal structure, when semelfactives are presented in the imperfective viewpoint, they are usually interpreted as occurring repeatedly and become coerced activities. 3.3.1.4 Achievements Achievements are dynamic, instantaneous, telic events such as win the game or explode. These events have no internal structure, but result in some change. As with 13 semelfactives, referring to them in imperfective aspect usually coerces an activity reading of repeated events.9 3.3.2 States States lack dynamism, differentiating them from all four types of events above. They do not describe a dynamic event but rather a property of the subject. When a state holds of an interval, it also holds of every subinterval of that interval, and there are no natural endpoints to a state. There are two types of states: individual-level (inherent) and stagelevel (temporary). 3.3.2.1 Individual-level States Individual-level states are relatively intrinsic, stable properties of the subject, such as be extinct or be tall: they hold true of an individual throughout the individual’s existence, i.e., they are considered to be permanent properties. 3.3.2.2 Stage-level states Stage-level states describe a temporary stage of a subject, relatively transitory properties like be hungry or be in the next room. Like individual-level states, there is no inherent endpoint to the property, but the fact that stage-level states are by definition temporary means that implicitly, endpoints to the state exist. 3.3.2.3 Differences in the behavior of individual-level and stage-level states Crosslinguistic evidence shows that patterns of syntactic behavior and semantic interpretation can be predicted by the individual/stage-level distinction. It must be noted that there is not a one-to-one correlation between a state and a classification of state type. For example, have blue eyes is an inherent, individual-level state for most, but a stagelevel state for someone who changes colored contacts on a daily basis, since for that person having blue eyes is not an inherent property of the person, but rather a temporary stage (Kratzer 1995). Still, the state has blue eyes would behave differently on the basis of which type of state it was expressing. Some of the varying behaviors that point to the temporary nature of stage-level states are discussed in this section. Lifetime effect. Individual-level states in the past tense in English are interpreted to have held throughout the existence of the subject, while stage-level states are interpreted as no longer holding true. (19) John was tall/liked chocolate/had blue eyes. (Individual-level) (20) John was hungry/was in the next room/ was sick. (Stage-level) 9 The Red Sox were winning, an apparent example of an imperfective achievement that is not iterated, really has the meaning “they were ahead,” not “they were in the transition moment from the game happening (when no team has won) to the game being over, when a winner has been established. The imperfective can be used with achievements to slow down the event in order to express simultaneity: As the bomb was exploding, the lieutenant threw on his helmet. It saved his life. 14 The reading of the individual-level states is unbounded—it continues indefinitely— because an inherent state lacks endpoints. In contrast, the stage-level state is interpreted as a bounded situation, one that is not true anymore, because the state is known to be temporary. Compatibility in Mandarin with the perfective marker –le. Stage-level states are compatible with the Mandarin perfective marker –le, but individual-level states are not (Chang 2003). (21) Stage-level states a. Ta e le. He hungry LE “He got hungry.” b. Ta lei le. He tired LE “He got tired.” (22) Individual-level states a. *Ta haoke le. *He hospitable LE b. *Ta pa she le. *He afraid snake LE (from Chang 2003) This analysis does not aim to delve into the semantics of –le, but we will characterize it as a morpheme of perfective aspect. In Chang’s analysis, -le is able to evoke a boundary when occurring with stage-level states but cannot do so with individual-level states because the permanence of the state entails that there are no endpoints to it. There are still other diverging behaviors of individual-level and stage-level states including ser/estar selection for “to be” in Spanish, grammaticality of There- insertion in English,10 and specification of the time and place of situation. The differences in behavior in these cases relate to the notion that stage-level states are temporary and therefore have implicit endpoints that can be evoked to bound a stage-level state (despite the fact that the state has no natural endpoint, unlike telics), whereas individual-level states are permanent features of the subject and thus have no endpoints, implicit or otherwise. 3.4 (Un)boundedness In Section 3 we have seen two different types of bounded and unbounded eventualities. At the level of situation aspect, each VP has intrinsic characteristics with regard to 10 (1) (2) There are firemen available. (stage) *There are firemen altruistic. (individual) (Kratzer 1995) 15 endpoints. Telic and durationless VPs have intrinsic bounds at which the eventuality ends: once one [eats four eggs], a (telic) accomplishment, he cannot continue to eat the same four eggs. On the other hand, one can [run in the park], and then continue to run in the park; an activity is an example of an atelic VP, which does not reach a natural endpoint but rather only an arbitrary, external one (whenever the runner gets tired or bored or hit by a bus). States also have no natural endpoints. Although a sentence may or may not have intrinsic endpoints, viewpoint aspect can provide extrinsic endpoints where only arbitrary ones exist, in the perfective, or hide intrinsic endpoints from view, in the imperfective. One situation class does not tolerate extrinsic bounds: individual-level states. As discussed in Section 3.3.2.3, individual-level states hold throughout the existence of the subject. Not even arbitrary, implicit endpoints are available to which the extrinsic bounds can apply. Individual-level states are inherently unbounded, in contrast with the implicitly unbounded stage-level states and activities that can have extrinsic bounds provided to them. In the perfect, Comrie (1976) notes an asymmetry: some languages allow both unbounded and bounded eventualities in the perfect; and some only allow bounded eventualities in the perfect—Modern Greek, for one; but it seems no languages allow exclusively unbounded eventualities in the perfect. The languages that only allow bounded eventualities in the perfect do not have a universal perfect. The following sections will suggest that ASL patterns with Greek, and the ability of the VP to be bounded is relevant to this classification. 3.5 Summary Aspect, in general, concerns the temporal characteristics of an event or situation. Viewpoint aspect presents the temporal constituency of a situation, and it either presents an event as a single, bounded whole (perfective aspect) or it highlights an event’s internal structure without referencing its endpoints (imperfective aspect), giving an unbounded interpretation. The imperfective aspect can be subdivided into various categories (the inventories vary crosslinguistically) that can express the continuousness of a single event, the progression of an event, or the internal structure of a series of iterations of events. The perfect aspect does not express the structure of the event itself but rather its relationship to the time of reference. The perfect asserts that there is a perfect time span continuing up to reference time, and that an event either occurs in the perfect time span (if the event is bounded) or throughout it (if the event has an unbounded reading). The event occurring in the perfect time span must somehow be relevant to reference time; there are four relevance conditions for use of the perfect. The four types are the experiential perfect, the perfect of result, the perfect of recent past, and the universal perfect. The universal perfect asserts that E persists throughout the perfect time span; it is only available if the event is unbounded. In languages such as English where a perfective/imperfective distinction is separable from the perfect/nonperfect distinction, these two types of aspect can be combined. Other languages, such as Modern Greek, restrict the perfect to use with perfective verbs only. 16 Situation aspect refers to the temporal properties of a sentence before viewpoint is applied. The distinguishing features are dynamism, telicity, and duration. The situation types are achievements, accomplishments, semelfactives, and activities—the four dynamic event types—as well as states. States have no natural endpoints, but the category is further divided into stage-level states, which are temporary, and individuallevel states, which are inherent properties of the subject. The transience of stage-level states allows boundaries to be evoked for these, but not for individual-level states. 4 FINISH in ASL The sign FINISH has many different uses in ASL. Fischer and Gough (1972, 1999) found seven meanings and four grammatical uses of the sign FINISH, most of them semantically related, and their list is not exhaustive. Crosslinguistically, verbs of termination, with a meaning of “to finish” or similar have a tendency to become grammaticalized as markers of perfect or perfective aspect. For example, the Swahili perfect prefix me- is thought to have come from the main verb meele, “to finish” (Heine and Reh 1984); the Mandarin perfective suffix –le may have been derived from liao, “to finish” (Li and Thompson 1981); and “done” marks perfect in African American Vernacular English (Dayton 1996). So it is not particularly surprising that this sign has become a target of grammaticalization in ASL, as it has been in other signed languages as well, including Israeli Sign Language (Meir 1999), Italian Sign Language (Zucchi 2003), and German and Dutch Sign Languages (Pfau & Steinbach 2006). A newly created grammatical morpheme and its lexical cousin are disambiguated by diverging behavior in certain contexts, and sometimes later in linear order. English have exemplifies this: it is a main verb meaning “to possess,” but also a grammatical auxiliary marking perfect aspect. The syntactic behavior of have differs when used for those two functions, at least in American English, in the requirement of do-support for questions and negation and in patterns of contraction. In their identification of lexical markers of tense and aspect in ASL, NKMBL (2000) identify lexical markers of tense and aspect that have been grammaticalized. The grammatical markers differ in articulatory pathlength and in syntactic position and behavior from the adverbials from which they presumably derived. Similarly, some of the different uses of FINISH can be distinguished by articulatory or prosodic characteristics, syntactic position, and interaction with negation. Section 4 discusses the various lexical, idiomatic, and grammatical uses of FINISH previously identified in the literature or found in the current study. Section 5 provides an analysis of one of the grammatical uses of FINISH, a marker of perfect aspect. 17