The ASL Perfect Formed by Preverbal FINISH Chapter 3

5.4.3.2 A problem for the analysis: compatibility with habitual inflection In most crosslinguistic characterizations of aspect, habitual marking falls into the category of imperfective. Indeed, in ASL, an event—telic or atelic—signed in the habitual aspect is usually interpreted as ongoing, like the other temporal aspects (Rathmann 2005), supporting the classification of the habitual as an imperfective morpheme. It differs somewhat from the continuous aspects such as the ASL incessant or the English progressive in that it refers to the regular repetition of events, rather than the ongoing occurrence of a single event, but the regular repetition is interpreted by default as continuing unbounded. We have found that habitual inflection can occur with preverbal FINISH, as in (63).21 (63) IX-1p FINISH GOhabitual CONFERENCE (UP-TO-NOW 5-YEAR) “I have gone (regularly) to the conference (for five years).” The grammaticality of a perfect habitual sentence poses a problem for an analysis that FINISH should not be compatible with imperfective inflection because inherently unbounded events clash with FINISH. However, sentence (61) only gives a bounded reading: It can only be used to discuss a habitual experience whose repetition has ended before reference time. The sentence cannot be signed if the regular repetition of going to the conference is known to overlap with the reference time, e.g., on a plane on the way to the relevant conference. If the regular attendance is known to continue, one signs: (64) IX-1p GOhabitual CONFERENCE (UP-TO-NOW 5-YEAR) “I have/had been going to the conference (for five years).” It is not exactly clear what is happening. The current analysis survives, however, if the habitual is not explicitly unbounded like the other imperfective inflections, but rather neutral with respect to boundedness, not unlike the activities and stage-level states we have seen in Sections 5.4.1.2 and 5.4.2.1, which are interpreted as unbounded by default, but have implicit endpoints that can be utilized to contribute a bounded reading to the eventuality if it is under FINISH. The unavailability of the continuous reading for the habitual in the ASL perfect points in this direction. This habitual inflection necessarily expresses the repetition of singular, bounded events. Although it is possible to refer to a continuous state that was habitual, as in (65), this cannot be done through use of the habitual inflection under discussion here. (65) JOHN #EX LIKE CHOCOLATE (NCSLGR vol. 9 ncslgr10s U 181) “John used to like chocolate.” Furthermore, FINISH cannot be used with the construction illustrated in (65). 21 It is not a frequent construction, though. The GOhabitual example is grammatical, but ME FINISH ASKhabitual, “I had asked (regularly)” was rejected. 32 Rather than a single continuous unbounded event, the habitual inflection describes a repetition of perfective events, with a potential endpoint situated between each pair of successive discrete events. It is, however, unclear why the habitual inflection allows FINISH to impose a bound whereas the iterative inflection does not. Rathmann notes a distinction between the habitual and the iterative in that the iterative refers to a single instance containing repeated events, whereas the habitual refers to the pattern of repeated events where the events are somehow more separable. The reasons for the habitual’s compatibility with the perfect when other imperfectives are incompatible are still not completely clear. 5.4.4 Summary of distribution of preverbal FINISH Preverbal FINISH is compatible with semantically bounded events and with semantically unbounded dynamic events. Stage-level states are rare but possible in the perfect in ASL, but individual-level states are not possible under the perfect marker FINISH. The imperfective inflections iterative, durative, and incessant produce ungrammatical sentences with FINISH, but the habitual inflection can appear with perfect aspect. For the situation types that are grammatical in the perfect, regardless of whether the VP itself is unbounded, the only available reading of the event under FINISH is bounded. This explains the ungrammaticality of individual-level states and imperfective marking in the perfect in ASL: both categories are explicitly unbounded. The habitual’s grammaticality in the perfect challenges this explanation. Although it is not quite clear why the habitual is allowable under FINISH, one possible explanation is that the habitual is not inherently unbounded and has implicit endpoints that can be utilized to bound the eventuality. 5.5 Types of perfect readings with preverbal FINISH This section will review the readings of the perfect constructions involving preverbal FINISH in the context of the four crosslinguistically attested types: the experiential perfect, the perfect of recent past, the resultative perfect, and the universal perfect. The majority of the perfects we elicited when testing the range of predicates that are compatible with the ASL perfect could best be described as experiential perfects. This is mostly a function of the testing paradigm we created. Aside from the tests for distribution of predicate types, we also made a concerted effort to elicit perfects of recent past, universal perfects, experiential perfects, and resultative perfects by creating situations in which we would expect to see them. 5.5.1 Experiential perfect The experiential reading of the perfect came quite easily. We were able to elicit an entire list of experiential perfects in a situation of comparing the previous on-the-job experiences of various job applicants. One example is provided as (66). (66) IX-3p FINISH PHONE TRANSFER… “He has transferred calls…” 33 When in isolation or formulated as a question (“Have you ever…?”), there was a preference to sign PAST/BEFORE at the end of the sentence, though it is not required to get an experiential perfect reading.22 The experiential perfect in ASL refers to an event that is no longer in progress but has happened some time before reference time; the event is bounded. 5.5.2 The perfect of result The frequency of the perfect of result varies a great deal crosslinguistically, as noted in the discussion of the perfect in Section 3.2.1. ASL allows the perfect of result, as in (67), although it does not seem to occur frequently. (67) BABY FINISH WAKE-UP “The baby has woken up/the baby is awake” The perfect of result signifies that the event under the perfect resulted in a change that currently affects the state of the subject. The event is bounded, as the change of state (the event) must be completed for the new result state to be in effect. 5.5.3 The perfect of recent past The perfect of recent past is not included among the potential uses of the ASL perfect. The adverbial or tense marker RECENT-PAST or the sign PAH/SUCCEED/FINALLY in preverbal position were often substituted when informants were asked to paraphrase English sentences. They rejected preverbal FINISH, both in paraphrasing English and when they were asked to give grammaticality judgments about usage in situations where a perfect of recent past would be expected. (68) *PRESIDENT FINISH DIE “The president has died!” This perfect of recent past is on the outskirts of the meaning of the perfect. There is no real current relevance other than the newness of information regarding a recent event. It may be that ASL more stringently applies the current relevance condition than English does, and thus excludes the perfect of recent past. 5.5.4 The universal perfect ASL does not allow the universal perfect. Recall from Iatridou, et al. (2001) that the universal perfect must be unbounded. The unboundedness requirement of the universal perfect is at odds with the fact that any VP under FINISH is interpreted as bounded (43). Rathmann cites the grammatical sentence IX-1p FINISH LIVE HAMBURG 10 YEAR as a universal perfect, as it can refer to any ten year stretch, including a ten year stretch that ends on the day of utterance. His example involves a speaker who has just moved out of 22 There is a similar preference in English, using ever in experiential perfect questions and before in affirmative experiential perfects. 34 Hamburg when he signs this sentence. There is no dispute about the grammaticality of the sentence or its meaning, as we were able to elicit a similar one. Our informant said that this can be signed if the time period continues up to the present. (69) IX-1p FINISH LIVE BOSTON 5-YEAR a. “I have lived in Boston for five years” (it was in the 1980s) b. “I have lived in Boston for five years” (it was for the last five years) If this sentence were a true universal perfect, however, we would expect that (70) would be felicitous; it is not. That meaning can only be achieved grammatically without FINISH, as in (71). (70) * IX-1p FINISH LIVE BOSTON 5-YEAR IX-1p STILL LIVE IX-locBoston NOW “I have lived (have been living) in Boston for 5 years. I still live there now.” (71) IX-1p LIVE BOSTON UP-TO-NOW 5-YEAR a. “I have been living in Boston for five years.” (Lit. “I live in Boston for five years,” analogous to the Spanish and French examples (Section 3.2.4) b. “I lived/was living in Boston for five years.” On the other hand, FINISH is appropriate if the action is terminated prior to reference time, as in (72). (72) IX-1p FINISH LIVE BOSTON 5-YEAR NOW IX-1p LIVE NEW-YORK “I have lived in Boston for 5 years. Now I’m living in New York.” Although in the reading of (69)-b, the specified time span continues up until the present, and it is therefore tempting to label it a universal perfect, the eventuality does not overlap with the present. Only unbounded events entail that the event must overlap with R, not merely to come up to R. Bounded events cannot overlap with the present, and a universal perfect reading is impossible in ASL. Thus, the reading for (69)-b, presented schematically in (73) is significantly different from that of a true universal perfect (74). (73) Time schema for Rathmann’s purported “Universal Perfect” (reading for (69)-b) … R I ß5 yearsà F [EEEEEEEEEEEEEEEEEE] ßPerfect time spanà Now (74) Time schema for true universal perfect R …EEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE…_____ 5 yrs. ago ßPerfect Time Spanà Now 35 The sentence (or type of sentence) that Rathmann cites as universal perfect is very similar to a universal perfect. To illustrate the distinction that is critical here, let’s consider similar constructions in English. English can have both perfect imperfective (unbounded) events, which include the progressive suffix –ing (as in (75)), and perfect perfective (bounded) events (76). Recall from Section 3.2.4.2 that only perfect-level, not eventuality-level durative adverbials can be fronted, and when an unbounded perfect has the durative adverbial fronted, only the universal reading should survive. (75) I have been playing baseball for five hours (76) I have played baseball for five hours (overlaps with present) (continues up to present, might overlap) (77) For five hours, I have been playing baseball (78) #For five hours, I have played baseball (overlaps with present) (cannot overlap with present) Sentence (69) is not a universal perfect, despite the fact the sentence can be truthfully uttered if the situation continues up to utterance time, and thus has truth conditions very similar to a universal perfect. This is highlighted by the ungrammaticality of (78). The fact that ASL sentences with perfect FINISH do not overlap with reference time shows that in ASL, there are only counterparts to the English perfect perfectives (like (76) and no counterparts to English perfect progressive sentences like (75) (marked by –ing). Unboundedness is required to obtain a universal perfect; the universal perfect occurs only with unbounded states or imperfectives. As mentioned in Section 5.4.2, the ASL perfect does not occur with individual-level states, and only allows a bounded reading of stage-level states, as shown in (79). (79) BABY FINISH SICK BEFORE/PAST “The baby has been sick before” * “The baby has been sick (and still is).” As for the aspectual inflections, the story is similar: the iterative, incessant, and continuative aspects are all incompatible with FINISH under any reading. On the other hand, the habitual has been observed to co-occur with FINISH. Still, like stage-level states, only bounded (experiential) readings are possible: The event does not overlap with the present. The universal perfect does not occur in ASL. This is consistent with the observation that the perfect in ASL occurs only with events that are terminated prior to reference time (43). 5.5.5 Summary of types of perfect found in ASL Preverbal FINISH has been found to give both perfect of result and experiential perfect readings, but not perfect of recent past or universal perfect readings. The non-existence of a recent past perfect is most likely simply attributable to a more stringent condition of 36 current relevance than other languages; recentness is not enough to make the event relevant enough to the present to necessitate the perfect in ASL. The absence of the perfect of recent past does not seem to be a function of any semantic qualities of the ASL perfect, but rather an instance of cross-linguistic variation with respect to the threshold for current relevance. The exclusion of the universal perfect is more principled. The universal perfect requires an unbounded reading of an event or situation, but any event or situation under perfect FINISH has a bounded reading, regardless of the boundedness of the underlying VP. The universal perfect is not an available reading of the ASL perfect. 5.6 Conclusions: Preverbal FINISH as marker of both perfect and perfective As Section 5.4 has shown, preverbal FINISH is compatible with all bounded (telic or instantaneous) eventualities, but only some unbounded eventualities and situations. Despite the fact that preverbal FINISH can combine with both bounded and some (otherwise) unbounded VPs, any perfect sentence in ASL is interpreted as bounded and not ongoing at reference time. The requirement for completion prior to reference time is intrinsic to the meaning of the perfect construction in ASL. Preverbal FINISH is, then, a marker of perfect plus perfective aspect. The perfective component of preverbal FINISH explains the distribution of unbounded eventualities that are acceptable and unacceptable with the perfect in ASL. Some, but not all, unbounded events are compatible with preverbal FINISH. Crosslinguistically, unbounded eventualities are attested as compatible with the perfect (though some languages do not allow them). On the other hand, only some unbounded eventualities tolerate bounds imposed by perfective viewpoint; others do not tolerate bounding by the perfective and therefore are incompatible with perfective viewpoint. The unbounded eventualities that are incompatible with preverbal FINISH in ASL are the same ones that tend to be incompatible with perfective viewpoint in other languages: the incompatibility of ASL preverbal FINISH with individual-level states and imperfective verbs is due to the perfectivity contributed by FINISH, not the perfectness. 5.6.1 FINISH and present time reference Preverbal FINISH, which marks both perfect and perfective aspect, contrasts with the (non-perfect) perfective VP-final FINISH in compatibility with present time reference, by virtue of the perfect aspect contributed by preverbal but not VP-final FINISH. In nonperfect constructions, event time, E, coincides with reference time, R. As briefly noted in Section 4.4.4, VP-final FINISH is incompatible with present time reference because the Bounded Event Constraint dictates that a bounded event cannot be located in the present, as the ungrammaticality of sentence (34), repeated below as (80), shows: (80) *NOW MOTHER BUY BOOK FINISH In perfect perfective sentences, however, E precedes R. ASL perfect sentences with preverbal FINISH are compatible with NOW, as seen in example (81). 37 Situation: Yesterday, John was not able to answer the teacher’s question. However, today he has read the book that was assigned and so he is on top of the material. (81) NOW JOHN FINISH READ BOOK KNOW ANSWER “Now John has read the book. He knows the answer.” The bounded reading of the event in a perfect sentence such as (80) and the adverbial NOW do not conflict with respect to the Bounded Event Constraint, because in perfect sentences, E precedes R. When NOW is specified as reference time in a perfect sentence, the event is specified as preceding, not overlapping, the present. Thus, NOW (or any other present time reference) is compatible with the perfect/perfective construction of preverbal FINISH, but not with the purely perfective VP-final FINISH construction. 5.6.2 Narrative advancement and the perfective Rathmann in part bases his claim that preverbal FINISH marks perfect, but not perfective, on the fact that perfective aspect induces narrative advancement. He states that the ordering of perfective VPs in a sentence also reflects the actual temporal ordering of the events themselves. Rathmann points out the fact that whereas FINISH23 at the end of a sentence or between sentences induces temporal sequencing, preverbal FINISH does not. Since VP ordering in sentences with preverbal FINISH does not necessarily reflect the temporal sequence of the events the VPs describe, Rathmann concludes that preverbal FINISH is not perfective. Although the default interpretation of perfective sentences may involve the sequencing of events in the same order as that of the VPs, this is certainly not a requirement of the perfective and does not prove that preverbal FINISH is not perfective. Crosslinguistically, perfective sentences do not always occur in an order that reflects the temporal order of the events. As pointed out by Alessandro Zucchi (personal communication), the Italian sentences below, for example, are perfective, but do not necessarily entail that the events occurred in the order that they are uttered: (82) Ieri John fece un sacco di cose. Camminò, corse, lavorò in palestra... “Yesterday, John did a lot of things. He walked, he ran, he worked out in the gym...” (Alessandro Zucchi, personal communication) The availability of event orderings other than the one reflecting the sequence of perfect sentences is not evidence that these sentences are not perfective, as we can see clearly from the Italian example. The perfectivity of preverbal FINISH is not disproved by the fact that perfect FINISH does not induce narrative advancement, as narrative advancement is not an absolute characteristic of perfective viewpoint. 23 Though remember that he conflates two FINISHes, both the VP-final perfective marker and the sentence-external discourse marker which we here liken to “and then” or “when that was over,” which certainly induces narrative advancement but does not mark perfective. The discourse marker is the basis for the claims of narrative advancement. 38 5.6.3 Restrictions due to perfectivity We have three categories relevant to the pattern of distribution of the perfect/perfective FINISH: a) bounded events, which are compatible with FINISH; b) unbounded events that tolerate FINISH and can be bounded by it; and c) unbounded events that do not tolerate FINISH. Of those, (b) and (c) pattern together in that they are interpreted as unbounded by default, while (a) and (b) are the two categories that the data show are acceptable with preverbal FINISH. What differentiates category (b) from category (c) is the existence of implicit endpoints that can be utilized to bound the eventualities. a) Bounded eventualities: telics and instantaneous events24 Achievements Instantaneous events b) I---F I,F Unmarked interpretation I---F I,F Interpretation under FINISH I---F I,F Unbounded eventualities that tolerate FINISH: activities, stage-level states, (possibly habitual aspect) Activities and Stage-level states c) Schema Schema (I)…---…(F) Unmarked interpretation …---… Interpretation under FINISH I---F Unbounded eventualities that do not tolerate FINISH: individual-level states and events in imperfective viewpoint Individual-level states Imperfectives Schema …---… I]…---…[F Imperfective viewpoint explicitly excludes the endpoints Unmarked interpretation …---… Interpretation under FINISH Incompatible with perfective viewpoint: I and F are not available for bounding The difference between categories (b) and (c) is that the individual-level states and imperfectives of category (c) are explicitly unbounded and therefore can never be bounded, whereas the activities and stage-level states are unbounded but since they are temporary eventualities, they have implicit endpoints that are accessible by the perfective. Since the perfective viewpoint can be applied to category (b), preverbal FINISH is compatible with eventualities of this category. Preverbal FINISH is ungrammatical with eventualities of category (c) because perfective viewpoint cannot be applied to category (c). 24 For these schemata, I=initial point, F=final point, --- = the eventuality, (I) and (F) = implicit endpoints, bold I, F = initial and final points bounded by perfective viewpoint, and … = an indefinite continuation of the eventuality 39 5.7 Summary Both the types of readings available in sentences with preverbal FINISH and the cooccurrence restrictions with certain types of predicates follow from the observation that preverbal FINISH contributes both perfect and perfective viewpoint to the event. Preverbal FINISH establishes culmination of the event or situation prior to reference time. It is thus compatible with bounded events, contributes bounds to unbounded eventualities that can (because they have implicit endpoints) be bounded, and is ungrammatical with intrinsically unbounded events because the explicit lack of bounds of individual-level states and imperfectives clashes with FINISH’s need to bound the eventuality. 40 Appendix I Lexical markers of tense/aspect, from NKMBL 2000 Marker PASTtns RECENT-PASTtns FORMERLYtns #EX Meaning Past tense Past tense, restricted to recent past Formerly Habitual past UP-TO-NOW (often glossed SINCE) Articulation Bent-B handshape taps (or moves towards) shoulder Restrained PASTtns or like the sign RECENT; in either case co-occurring with the ‘cs’ facial expression 5 handshape moving in small circular motion with thumb side brushing near collar bone Identical to the fingerspelled loan sign used as a nominal prefix From some time Two index fingers, bent at first knuckle, palm in the past to the downward, fingertips contacting upper shoulder present and moving in an arcing motion to a palm-up orientation at about chest height in front of the body IMMEDPRESENTtns Immediate present Similar to the sign NOW, but articulated closer to the body, co-occurring with nonmanual marking in which teeth are clenched, lips are stretched, and mouth corners are pulled downward FUTUREtns Future tense B handshape oriented palm sideward, starting near the dominant side of the face and moving forward along a fixed pathlength Table of lexical tense and aspect markers identified by NKMBL (2000) 41