Nov 7, 2008 - Roger W. Andersen and Yasuhiro Shirai (1994). Discourse ...... Following Rosch (1973), Ross (1973), Lakoff (1987), and J. R. Taylor (1989), we.
Studies in Second Language Acquisition http://journals.cambridge.org/SLA Additional services for Studies in Second
Language Acquisition: Email alerts: Click here Subscriptions: Click here Commercial reprints: Click here Terms of use : Click here
Discourse Motivations for Some Cognitive Acquisition Principles Roger W. Andersen and Yasuhiro Shirai Studies in Second Language Acquisition / Volume 16 / Issue 02 / June 1994, pp 133 156 DOI: 10.1017/S0272263100012845, Published online: 07 November 2008
Link to this article: http://journals.cambridge.org/ abstract_S0272263100012845 How to cite this article: Roger W. Andersen and Yasuhiro Shirai (1994). Discourse Motivations for Some Cognitive Acquisition Principles. Studies in Second Language Acquisition,16, pp 133156 doi:10.1017/S0272263100012845 Request Permissions : Click here
Downloaded from http://journals.cambridge.org/SLA, IP address: 195.169.108.173 on 17 Aug 2012
SSLA, 16,133-156. Printed in the United States of America.
DISCOURSE MOTIVATIONS FOR SOME COGNITIVE ACQUISITION PRINCIPLES
Roger W. Andersen University of California, Los Angeles
Yasuhiro Shirai Daito Bunka University
This paper offers an alternative interpretation for what has been called the defective tense hypothesis, the primacy of aspect hypothesis, or simply the aspect hypothesis in the literature on first and second language acquisition of tense and aspect. The aspect hypothesis states that first and second language learners will initially be influenced by the inherent semantic aspect of verbs or predicates in the acquisition of tense and aspect markers associated with or affixed to these verbs. Our account focuses on the observation that adult native speakers also appear to adhere to this primacy of inherent semantic aspect in the relative quantitative distribution of tense-aspect markers in their speech. We argue that a small set of cognitive operating principles and the notion of prototypicality account for this behavior in learners. Moreover, we argue that these principles are a consequence of how learners and native speakers alike organize information and their perspectives on it in ongoing discourse.
In recent work (e.g., Andersen, 1989b, 1989c, 1990, 1993; Andersen & Shirai, in press), we have argued that the empirical evidence available on both first and second language acquisition of verb inflections can be accounted for partly in terms of a number of cognitive operating principles of the type Slobin has elaborated in his cross-linguistic project on first language acquisition (e.g., Slobin, 1985). This account Each author contributed equally to this paper. The order of authors' names is strictly alphabetical. Grants to Andersen from the National Science Foundation (BNS-8812750) and to Shirai from a Grant-Aid for Scientific Research from the Japanese Ministry of Education (no. 05881078) contributed to the writing of this paper. • 1994 Cambridge University Press 0272-2631/94 $5.00 + .00
133
134
Roger W. Andersen and Yasuhiro Shirai
Table 1. Inherent semantic aspect A. Semantic Features
Punctual Telic Dynamic
States
Activities
Accomplishments
Achievements
-
+
+ +
+ + +
B. Examples
State have possess desire want like
Activity
run walk swim push pull
Accomplishment (Telic Event)
Achievement (Punctual Event)
paint a picture make a chair build a house write a novel grow up
recognize (someone) realize (something) lose (something) find (something) win the race
involves, more specifically, the Relevance Principle, the One to One Principle, and a Principle of Congruence—together with a prototype model of language acquisition (see Andersen and Shirai, in press). However, we have also argued that we cannot arrive at a clear understanding of how learners acquire tense-aspect systems without relating learners' underdeveloped systems to the fully developed tense-aspect systems of proficient adult native speakers of the language, as they use their native language in natural discourse. This paper builds on this earlier work and argues that some of these cognitive operating principles, which appear to be principles unique to language acquisition, are, in fact, motivated by the same discourse organizational principles that proficient native speakers follow in using language to convey information and personal perspectives in social interaction. To follow our argumentation requires at least a minimal understanding of Vendler's (1967) and Mourelatos' (1981) classification system (borrowed from Aristotle1) of the types of situations depicted by verbs and their arguments. This is summarized in Table 1 and Figure 1. As illustrated in Table IB, verbs (or rather predicates) can be categorized according to their inherent semantic aspect into four classes. States like have depict a situation that is assumed to last indefinitely. Activities like run, on the other hand, are durative but require input of energy for them to take place and have an arbitrary beginning point and end point. Accomplishments like paint a picture or run a mile, however, share with activities their durativity but have an inherent end point (when the picture is finished or the mile is up). Achievements like realize something have inherent end points like accomplishments but differ from accomplishments in that they have no duration. They are momentaneous, punctual. The terms used in Table 1 come from Vendler (1967). Mourelatos (1981) proposed the hierarchical organization among these four categories displayed in Figure 1. States contrast with dynamic actions in that states exist indefinitely without input of energy, whereas dynamic actions require input of energy. Dynamic actions are
Discourse Motivations
135
SITUATIONS
STATES
DYNAMIC ACTIONS
ACTIVITIES
EVENTS
ACCOMPLISHMENTS
ACHIEVEMENTS
Figure 1. The Vendler-Mourelatos hierarchy. subcategorized into activities, which are durative and have an arbitrary end point, and events, which have an inherent end point. Finally, events are of two types: Accomplishments have a durative element that precedes the final end point, whereas achievements are momentaneous, that is, the end point is also the beginning point. Table 1A maps Comrie's (1976) binary semantic features punctual/durative, telic/ atelic, and dynamic/stative onto the four Vendler-Mourelatos lexical classes. Thus, we can characterize states as [-punctual], [-telic], and [-dynamic], activities as [-punctual], [-telic], and [+dynamic], accomplishments as [-punctual], [+telic], and [+dynamic], and achievements as [+punctual], [+telic], and [+dynamic]. Despite the controversies in the literature on first language acquisition of morphological encoding of tense and aspect, there is a strong empirical basis for concluding that children follow the aspect hypothesis in their acquisition of tense and aspect: In the earliest stage of acquisition of their native language, children restrict their use of the verb morphology of the language they are acquiring such that each inflection is used with a separate and distinct semantic class of verbs (Andersen, 1989a; Andersen & Shirai, in press). Past and perfective inflections are restricted primarily to accomplishment and achievement verbs. Progressive inflections are restricted primarily to activities and rarely overextended to statives. Past imperfective inflections are slower to emerge and, when they do appear, are initially restricted to activities and states. Viewed from the perspective of the semantic categories, states are seldom inflected in early stages, activities most typically receive progressive and/or imperfective inflections, and achievements and accomplishments typically receive past and/or perfective inflections. These correspondences are given schematically in Table 2. A distinction is made between earlier and later inflections in Table 2, because in languages with explicit imperfective past marking, use of this inflection is a late development. When imperfective inflections appear in Spanish and Portuguese, they first appear with states and activities (in addition to the earlier inflection for progressive on activities). The last column to the right depicts the assumed distribution of these same inflections for native language use, again taking Spanish and Portuguese as examples. The assumption in the language acquisition literature appears to be that each of these formsprogressive, perfective, imperfective—occurs with approximately equal frequency
136
Roger W. Andersen and Yasuhiro Shirai
Table 2. Assumed distribution of past inflections by verb class for learners and native speakers of Spanish and Portuguese Acquisition Verb Class
Early
Late
State
Uninflected
Imperfective
Activity
Progressive
Progressive Imperfective
Accomplishment/ achievement
Perfective
Perfective
Native Adult Use Imperfective Perfective Progressive Imperfective Perfective Perfective Imperfective Progressive
across all four verb classes (reduced to three here, by combining achievements and accomplishments), with the familiar qualification that states do not usually inflect for progressive. Our position is that this assumption is wrong, as we will show in this paper. These findings are so consistent across different studies that it is no wonder that they have generated so much discussion in the literature concerning the forces that create such distributions of verbal inflections in first language acquisition. A comprehensive review of the previous studies is beyond the scope of the present paper but is treated elsewhere (Andersen, 1989a; Andersen & Shirai, in press). In this paper we will instead focus on an explanation for this behavior among language learners that also accounts for adult native speakers' use of tense-aspect markers. ASPECT IN FIRST LANGUAGE ACQUISITION In the 1970s and early 1980s, a series of studies were published claiming that children are influenced by aspect rather than tense in the acquisition of past tense morphology: Bronckart and Sinclair (1973) on French, Antinucci and Miller (1976) on English and Italian, Bloom, Lifter, and Hafitz (1980) on English, and Stephany (1981) on Greek. Along a somewhat different line, Bickerton (1981) argued for a version of what we are calling the aspect hypothesis to support his language biopro gram hypothesis. He reinterpreted Bronckart and Sinclair (1973) and Antinucci and Miller (1976) and claimed they support his punctual-nonpunctual distinction. Weist, Wysocka, Witkowska-Stadnik, Buczowska, and Konieczna (1984) argued against such a claim. They labeled the preceding studies as arguing for "the defective tense hypothesis" and claimed, based on their data, that the aspect hypothesis is not tenable. For Weist et al. (1984), the defective tense hypothesis claims that young children are cognitively incapable of conceiving of events displaced in time and can only encode reference to the here and now. Andersen (1989a) called this the absolute defective tense hypothesis and argued that the evidence actually supports a relative defective tense hypothesis, which is descriptively identical to the aspect hypothesis as stated in this paper. Weist et al. (1984) was followed by a debate
Discourse Motivations
137
between Weist and Bloom (Bloom & Harner, 1989; Rispoli & Bloom, 1985; Smith & Weist, 1987; see also Weist, 1989). Weist et al. (1984) made an important contribution to the first language acquisition literature by establishing the following empirical facts: 1. Children between 1;6 and 2;6 can make tense distinctions as well as aspectual distinctions. 2. It is not the case that children use a past form only to refer to an observable change. They can also refer to distant past. However, Weist et al.'s (1984) version of the defective tense hypothesis, which they attributed to Antinucci and Miller (1976), Stephany (1981), Aksu (1978), and others, is in fact an overly strong claim. According to their version of the defective tense hypothesis, no exception is allowed. For example, activity verbs should never receive past or perfective inflections, remote past should never be commented on with past tense morphology, and so forth. This is a very strong claim, and it would probably be unrealistic to expect such clear-cut results from child language data. As we interpret the literature, the issue to be dealt with is whether children are generally guided by inherent semantic aspect in their use of tense-aspect morphology, not whether children are capable of using past morphology as true tense markers. This point was exemplified by Andersen (1989a) and Bloom and Harner (1989). Each showed independently, by reanalyzing tables in Weist et al. (1984), that Polish children are clearly biased in their use of tense-aspect morphology in relation to inherent aspect classes of the verb to which an inflection is attached. Therefore, even though Weist et al. (1984) claimed that the defective tense hypothesis is not tenable, what Andersen (1989a) refers to as the relative defective tense hypothesis is still relevant; that is, early morphology is predominantly guided by aspectual characteristics of the verbs (or the situation they describe). It is important to point out that the relative defective tense hypothesis, which we prefer to refer to simply as the aspect hypothesis, makes no claims about a child's ability or inability to make tense distinctions, whereas the absolute defective tense hypothesis does. Our position is that the absolute defective tense hypothesis focuses so strongly on an either-or position—either aspect or tense—that the contribution of inherent semantic aspect to children's early use of verbal morphology is lost. As should become clear by the end of this paper, treating the question as one of either tense or aspect is shortsighted and actually obscures the true nature of the phenomenon under study. The Distributional Bias Hypothesis Andersen (1986) first introduced the notion of distributional bias to suggest a possible source of learners' use of verb morphology as described by the aspect hypothesis. This notion, which was formalized in Andersen (1990 [1988]), was as follows: Native speakers in normal interaction with other native speakers tend to use each verb morpheme with a specific class of verbs, also following the aspect hypothesis. When learners are then exposed to this language of native speakers, they initially interpret
138
Roger W. Andersen and Yasuhiro Shirai
this skewed distribution of forms as an absolute characteristic of the forms themselves. Andersen (1990) stated the Distributional Bias Principle as follows: There a r e . . . properties of the input that promote the incorporation of an inappropriate form : meaning relationship into the interlanguage. That is, the learner misperceives the meaning and distribution of a particular form that he discovers in the input, following the Distributional Bias Principle: If both X and Y can occur in the same environments A and B, but a bias in the distribution of X and Y makes it appear that X only occurs in environment A and Y only occurs in environment B, when you acquire X and Y, restrict X to environment A and Y to environment B. (p. 58) This formulation does not distinguish native-native discourse to which the learner has access from possible modification of native speech as directed to learners. While Stephany (1981), for child language acquisition, and Shirai (1990), for second language acquisition, would lead us to believe that language directed to learners conforms much more closely to the aspect hypothesis, this expectation is not an obligatory part of the distributional bias hypothesis. Let us take a frequently observed feature of nonnative speech: frequent use of stative verbs with past time reference in the base form in English or third singular present in Romance languages; that is, failure to mark statives for past time reference. According to the distributional bias hypothesis, we expect stative verbs to be inflected for past time reference in these languages much less frequently than dynamic verbs and especially event verbs (accomplishments and achievements). If this is the case, then learners would be more inclined to notice past marking on dynamic verbs than on state verbs and thus begin to inflect dynamic verbs for past time reference before using past inflections on stative verbs. The assumption is, therefore, that a distributional bias of this sort in native speaker speech would cause learners to miscategorize the applicability of past tense marking (or perfective marking for Romance languages). Although there is still much more work to be done on this issue, Andersen (1992, 1993), Huang (1993), and Robison (1993), for second language acquisition, and Shirai (1991), for first language acquisition, have confirmed the distributional bias hypothesis. Regardless of whether this distributional bias in native speech is the major source of the use of verb morphemes by learners according to the aspect hypothesis, the distribution in native speech itself still needs to be explained; that is, it is not sufficient to suggest or even prove beyond any doubt that learners find in native speech the model that results in their early use of verb morphology. We still need to know why native speakers use verb morphology in this fashion. The purpose of this paper is to suggest why this is the case and, in so doing, also offer an explanation for the observed use of verb morphology by learners.
Evidence for Distributional Bias Recently Shirai (1991) has studied the distributional bias hypothesis more directly, by analyzing the speech of three first language learners of English over four stages from roughly 1 year and 2 months old up to the end of the 4th year,2 as well as the
Discourse Motivations
139
Table 3. Distribution of inherent aspect with past and progressive inflections in motherese addressed to each child (total token counts for all stages) State
Activity
Accomplishment
Achievement
Adam's mother Past 17% (30) -ing 0% (0)
8% 56%
(13) (81)
11% (19) 11% (18)
64% (112) 33% (39)
Eve's mother Past -ing
13% (16) 0% (0)
7% (9) 53% (111)
21% (27) 14% (29)
59% 33%
Naomi's mother Past 12% (43) 4% (20) -ing
18% (51) 61% (315)
12% (31) 12% (58)
58% (180) 23% (119)
(74) (69)
Source: Primacy of Aspect in Language Acquisition: Simplified Input and Prototype (p. 89) by Y. Shirai, 1991, unpublished doctoral dissertation, University of California, Los Angeles. Copyright 1991 by Y. Shirai.
Table 4. Distribution of inherent aspect with past inflections in motherese and the children's speech at Stage 1 Activity
State Adam Mother Child
Accomplishment
Achievement
17% (10) 0% (0)
7% (4) 6% (1)
10% (6) 0% (0)
66% 96%
Mother Child
23% 0%
(9) (0)
5% (2) 0% (0)
13% (5) 0% (0)
100%
Naomi Mother Child
11% 0%
(4) (0)
17% (6) 0% (0)
17% (6) 0% (0)
100%
(38) (16)
Eve 59% (23) (4)
55% (20) (10)
Source: Primacy of Aspect in Language Acquisition: Simplified Input and Prototype (p. 100) by Y. Shirai, 1991, unpublished doctoral dissertation, University of California, Los Angeles. Copyright 1991 by Y. Shirai.
mother's speech to her child. Table 3 displays the distribution of past and progressive inflections in the mothers' speech to their children over all four stages of the data. These figures support the distributional bias hypothesis in that the past inflection is associated most strongly with achievements and the progressive inflection with activities. Table 4 shows that in the Stage 1 speech of the three children the past inflection only occurs with achievements (with a minor deviation from this pattern for Adam). Table 5 shows a somewhat similar pattern for the distribution of the progressive inflection at Stage 1. However, there is a much higher frequency of use of the progressive with achievements than the distributional bias hypothesis would predict. Shirai (1991) pursued further this unexpected finding that while around 70% of the uses of the progressive inflection in the children's speech are on activity or accomplishment event verbs, consistent with an "action-in-progress" meaning,3
140
Roger W. Andersen and Yasuhiro Shirai
Table 5. Distribution of inherent aspect with progressive inflections in motherese and the children's speech at Stage 1 State
Activity
0% (0) 0% (0)
51% (26) 58% (11)
14% 10%
(7) (2)
35% (18) 32% (6)
Mother Child
0% (0) 0% (0)
53% (38) 75% 0)
8% 0%
(6) (0)
39% (28) 25% (3)
Naomi Mother Child
3% (4) 4% (5)
65% (93) 68% (93)
12% (18) 4% (5)
20% (29) 24% (33)
Adam Mother Child
Accomplishment
Achievement
Eve
Source: Primacy of Aspect in Language Acquisition: Simplified Input and Prototype (p. 101) by Y. Shirai, 1991, unpublished doctoral dissertation, University of California, Los Angeles. Copyright 1991 by Y. Shirai.
Table 6. Distribution of inherent aspect with progressive inflections in the speech by the three children at stage 1 (Adam 2; 1-4, Eve l;6-7; Naomi l;6-10)
Adam
Eve Naomi
State
Activity
Accomplishment
Achievement
0% (0) 0% (0) 4% (5)
58% (11) 75% 0) 68% (93)
10% (2) 0% (0) 4% (5)
32% (6) 25% (3) 24% (33)
Source; Primacy of Aspect in Language Acquisition: Simplified Input and Prototype (p. 103) by Y. Shirai, 1991, unpublished doctoral dissertation, University of California, Los Angeles. Copyright 1991 by Y. Shirai.
about 25-30% of the inflections are on achievement verbs. Table 6 displays the relevant distribution more clearly. Shirai checked all cases of the progressive used with achievement verbs, except for the last sample for Naomi, because it was extremely large and thus much larger than the samples for the other two children. This left 19 cases. Of these 19 occurrences of -ing with achievement verbs, 13 had iterative meaning (e.g., jumping, banging, turning, brushing). Shirai discusses the welldocumented fact4 that punctual activities like cough do not have an inherent end point, whereas all other accomplishment and achievement verbs do. Verbs like jump and cough are punctual in that one jump and one single cough are punctual, but they are really durative when referring to iterated situations, such as he coughed all night and the girls jump rope during recess. Shirai argues that the children in this study (and children in general) group iterative punctuals (which he calls punctual activities, following Lee, 1991) with activities when the situation referred to is one of action-in-progress with no inherent end point. Thus, the progressive inflection is not simply used redundantly to mark the same information as is available in the inherent aspect of the verb (activity or duration in this case) but, rather, marks the prototypical notion of action-in-progress, which applies to pure activities as well as to punctual activities.
Discourse Motivations
141
Table 7. Studies on the aspect hypothesis in second language acquisition (arranged by LI) N English Rothstein (1985) Kumpf(1984) Shirai and McGhee (1988) Mishina(1993) Nixon (1986) Yoshitomi (1992) Bayley(1991) Huang (1993) Flashner(1982) Cushing (1987) Kumpf(1982) Robison (1990) Robison (1993) H. Taylor (1987) Economides (1985) Bardovi-Harlig(1992) Spanish Andersen (1986,1991,1992) Ramsay (1989a, 1990) French Kaplan (1987) Bergstrom (1993) Kihlstedt (in press)
Learner Characteristics
I
LI
3 years in USA 28 years in USA 6 months in USA Uninstructed, in USA 1-6 months in USA 7 years in USA Not described Uninstructed, in USA 2, 3, and 4 years in USA 1.5 years in USA 30+years in USA Less than 3 years in USA 1st year university students 1-10 months in USA 12+months in USA Foreign students in USA
Hebrew Japanese Japanese Japanese Mandarin Mandarin Mandarin Mandarin Russian Serahuli Spanish Spanish Spanish Spanish Vietnamese Mixed
I 3()
8-14 years old Classroom SLA
English English
15 111i I
Classroom SLA Classroom SLA Classroom SLA
English English Swedish
1 2(1 5 }
2(5 I 13, 5
Shirai's (1991) study has thus supported the distributional bias hypothesis and has provided a plausible explanation for why the first occurrences of the progressive include cases of achievement verbs, that is, because these verbs are like activity verbs in that they do not have an inherent end point and prototypically refer to action-in-progress. ASPECT IN SECOND LANGUAGE ACQUISITION We have limited the review of the issues we are addressing so far to first language acquisition. In this section we review the second language acquisition studies of tense-aspect morphology in various languages. Adult second language learners show a tendency similar to what has been observed in first language acquisition. We will first review the second language acquisition studies and then relate the studies to the issues already discussed. Table 7 summarizes the studies relevant to the aspect hypothesis in SLA. Most of the studies report on adult L2 learners (except for Andersen, 1986, 1991, 1992, Economides, 1985, and H. Taylor, 1987) and naturalistic second language acquisition (except for Bardovi-Harlig, 1992, Bergstrom, 1993, and Ramsay, 1989a, who investigated classroom SLA, and Nixon, 1986, and Shirai & McGhee, 1988, whose subjects
142
Roger W. Andersen and Yasuhiro Shirai
had studied the target language extensively before coming to the United States and had been in the U.S. for only a short period at the time of the study). The results generally support the aspect hypothesis, except that L2 learners showed extensive use of progressive on stative verbs. The relevant studies are of English, Spanish, and French as a second language. The English data generally show (although with some variation) that (a) past morphology is strongly associated with perfective aspect and/or achievement or accomplishment verbs (Cushing, 1987; Economides, 1985; Flashner, 1982 [see also Wenzell, 1989]; Robison, 1990; Rothstein, 1985; Shirai & McGhee, 1988; H. Taylor, 1987),5 and (b) -ing is strongly associated with imperfective aspect and/or durative (i.e., state, activity, and accomplishment) verbs, with activity verbs receiving more -ing marking (Cushing, 1987; Economides, 1985; Kumpf, 1982; Rothstein, 1985; H. Taylor, 1987). In these studies the results regarding the use of -ing are important because they are different from what is found in first language acquisition in that progressive markers are overextended to stative verbs. This is probably because the learner has a first language on which to map L2 forms. Flashner (1982) attributes her subjects' use of past morphology for perfective contexts and the base form for imperfective contexts to transfer from Russian, the subjects' native language. Many of the native languages of the subjects in the preceding studies have imperfective aspect, which is strongly associated with durativity (Comrie, 1976; Weist et al., 1984). It is plausible that these learners associate the -ing marker with imperfective aspect in their first language, because progressive is part of imperfectivity (Comrie, 1976). One possible counterexample to the general prediction of the aspect hypothesis seen in SLA is reported by Kumpf (1984). Her subject, Tamiko, a Japanese speaker, tended to use base forms for completed actions in the foreground, while frequently using past tense markers for stative verbs, activity verbs being marked with -ing in the background (see Hopper and Thompson, 1980, for the notions of foreground and background). Clearly further research, such as more studies on Japanese subjects,6 is necessary before we treat this example as idiosyncratic variation. An apparent counterexample to the aspect hypothesis was offered by Meisel (1987). Based on his study on German as a second language, he claimed that learners do not systematically use an aspectual system. It may well be that this is a very marginal phenomenon, occurring only occasionally, which has received too much attention by researchers who base their expectations on findings in LI studies or on Creole studies, (p. 220)
However, his study had a very different framework and does not constitute a problem for the claim of the aspect hypothesis. The aspect hypothesis only concerns the acquisition of verbal morphology, whereas Meisel's focus was on how past time reference is encoded in interlanguage, including other devices such as adverbials, discourse organization, and so forth (function-to-form analysis; see Long & Sato, 1984). Echoing the caution expressed by Meisel (1987), Bardovi-Harlig (1992) concluded that her study does not constitute support for the aspect hypothesis. She claimed,
Discourse Motivations
143
"The learners in this study also marked tense fairly consistently across aspectual classes" (p. 274). However, a reexamination of her results clearly shows that they follow the prediction of the aspect hypothesis. For the sake of simplicity, let us discuss the 19 Level 1 (elementary) learners in her study. The general trend is that these classroom-instructed L2 learners fail to supply past marking in obligatory context much more frequently for state and activity verbs than for achievement verbs. More specifically, correct suppliances of simple past forms on achievement verbs (63.2%) are much more frequent than those on activity (35.1%) and state (31.6%) verbs. Furthermore, even when incorrect markings of pastness (i.e., past progressive, past perfect, and past perfect progressive) are included, the trend remains the same, although in a less dramatic way, with 63.2% for achievement verbs, 43.9% for activity verbs, and 42.1% for state verbs. It should be noted that Bardovi-Harlig's (1992) data elicitation procedure is quite different from most of the other studies reviewed here. It consists of responses to a cloze test in which careful monitoring is possible. In addition, her subjects are enrolled in an ESL program at an American university and are therefore exposed to formal instruction. This shows that primacy of aspect is very broad in its application; that is, not only is it observed in the naturalistic data of naturalistic L2 acquisition but also in the paper-and-pencil test data elicited from ESL learners with formal grammar instruction. More recent work by Bardovi-Harlig provides even stronger support for the aspect hypothesis (Bardovi-Harlig & Bergstrom, 1993; Bardovi-Harlig & Reynolds, 1993). The acquisition of Spanish as a second language has been a good testing ground for the aspect hypothesis, and the studies of Spanish L2 acquisition provide some of the clearest cases of the general claims of the aspect hypothesis, that is, the correlation of past perfective morphology with accomplishment and achievement verbs, and past imperfective forms with state and activity verbs. Andersen's (1986, 1991, 1992) quasi-longitudinal study clearly showed that past perfective appeared first, and the order of emergence was "achievement -• accomplishment -» activity -> state," while a slower development of imperfective past followed the course of "state -> activity -> accomplishment -» achievement." Ramsay's (1989a, 1989b, 1990) cross-sectional studies of classroom Spanish learners also showed the same tendency as those of Andersen, clearly following the general developmental pattern as predicted by the aspect hypothesis. French L2 studies on tense and aspect have been on classroom learners of French. Kaplan's (1987) study indicates that the learners tend to use perfective past forms with perfective events, while using present for imperfective processes, thus roughly replicating Bronckart and Sinclair's (1973) results. On the other hand, Bergstrom (1993), replicating Bardovi-Harlig's English L2 study in French L2 by using a cloze test, had interesting results. For example, Level 1 (elementary) learners gave past marking in obligatory-past items at 79.6% for achievement, 60.3% for accomplishment, 73.8% for activity, and 38.5% for states, thus activity showing higher marking by passe compose than accomplishment. Most evident, however, is that states are least often marked for passe compose of all four groups. This clearly
144
Roger W. Andersen and Yasuhiro Shirai
follows the aspect hypothesis. On the other hand, the aspect hypothesis predicts less success in marking passe compost for activities than accomplishments, whereas accomplishments score lower than activities.
The Role of Cognitive Development What are the implications of SLA research for the aspect hypothesis in general? Earlier studies relevant to the aspect hypothesis on LI acquisition (Antinucci & Miller, 1976; Bronckart & Sinclair, 1973; Smith, 1980) attributed the primacy of aspect to a cognitive deficit, suggesting that children did not have the concept of deictic past. However, Weist et al. (1984) argued that the primacy of aspect cannot be due to a cognitive deficit because Polish children before the age of 2 can productively use tense (and aspect) morphology. It should be noted here that most of the discussion in the first language acquisition literature regarding the relationship between conceptual development and linguistic development makes no reference to second language acquisition data,7 which would be a good testing ground because adult second language learners already have concepts on which to map linguistic labels. The area of acquisition of tense-aspect morphology is no exception. How then do we interpret the fact that adult L2 learners exhibit behavior similar to that of children acquiring their native language? As Andersen (1989a) points out, the cognitive deficit—the lack of the concept of deictic past—cannot be the sole reason for the primacy of aspect, because adult learners, who clearly have a concept of deictic past, also show the same tendency as children acquiring a first language at least with respect to the acquisition of past/perfective morphology. This rules out the possibility that primacy of aspect in French and Italian LI acquisition is exclusively due to cognitive deficit, although it is still possible that a cognitive deficit is one of the contributing factors for the primacy of aspect for children. In this connection, it is important that support for the aspect hypothesis is found for classroom learners (e.g., Bardovi-Harlig, 1992; Ramsay, 1990) as well as for LI and naturalistic L2 learners.
Zero Marking in Obligatory Contexts Most of the studies that argue for the aspect hypothesis do not report the pattern of the lack of marking in obligatory contexts; they usually report the distribution of use classified by its aspectual class. This is true for both first and second language acquisition studies. However, as already reviewed, some L2 acquisition studies clearly have shown that stative verbs receive past marking in obligatory-past contexts less frequently than accomplishment/achievement verbs (Bardovi-Harlig, 1992, Bergstrom, 1993).8 Such findings are clear evidence that primacy of aspect is not merely a distributional bias in the learner's use of morphology in the same way that native speakers exhibit such a distributional bias. In this sense, first and second
Discourse Motivations
145
language learners do things differently (i.e., show deviant performance) from adult native speakers, and the deviant behavior appears to be guided by the aspectual classes of the verbs. PRINCIPLES OF TENSE-ASPECT ACQUISITION AND USE Let us summarize here the empirical observations that need to be explained: 1. Distributional bias in adult native speech 2. More absolute bias (and deviant use) in language acquisition 3. Same behavior for native speakers and first and second language learners Andersen (1993) argues that the explanation for these three observations lies in the figure-ground distinction and four cognitive operating principles: (a) the Relevance Principle (from Bybee, 1985, and Slobin, 1985), (b) the Congruence Principle (proposed in Andersen, 1993), (c) the One to One Principle (Andersen, 1984), and (d) the Subset Principle (see Hyams 1986, p. 24, Footnote 7). The summary of that argument, given below, has been modified in two ways. First, we have replaced the Subset Principle, which is used as a syntactic principle by many linguists and therefore not appropriate here, with a prototype account for the conservative behavior of learners. Second, we have replaced the figure-ground distinction with the notion of discourse motivation, to be discussed shortly. The behavior we want to explain includes the following. When learners first start using verb morphemes (typically suffixes, but also auxiliaries), they use tense markers like English past, aspect markers like English and Spanish progressive, and tense-aspect markers like Spanish past perfective. They do not, however, use agreement markers, such as English third singular present -s or Spanish first singular present -o. Moreover, when they use something like English past, they appear to be using it not as a tense marker but to mark aspectual meaning, such as "completed" or "finished," and restrict its use to event verbs. Similarly, when they use the progressive marker, they initially restrict its use to activity verbs. And when they use Spanish past perfective, they similarly restrict it to accomplishment and achievement verbs, especially the latter. Thus, even grammatical aspect markers like the progressive and the perfective are restricted in their use to only certain semantic classes of verbs. To explain why learners initially use the past marker on event verbs, apparently as a marker of completion or end point and not simply as a tense marker, we use Bybee's (1985) Relevance Principle (see also Slobin's, 1985, use of the principle). According to our interpretation of this principle, a grammatical morpheme is first used by learners according to how relevant it is to the meaning of the verb. Because aspect is more relevant to the meaning of the verb than tense, mood, or agreement, the first uses of verb morphemes are as aspect markers. This also accounts for slower development of markers of mood, tense, and agreement. This also explains why a "tense" marker like the English "past" takes on aspectual meaning when it is first used.
146
Roger W. Andersen and Yasuhiro Shirai
This does not explain, however, why a past or perfective marker is initially used primarily with event verbs and a progressive marker with activity verbs. All three markers are used to convey aspectual notions. Andersen (1993) argues that learners will choose from the various aspectual morphemes that they have noticed in the input the morpheme whose aspectual meaning is most congruent with the aspectual meaning of the verb (see also Bybee's, 1985, pp. 76-77, discussion of local markedness). Andersen (1993) calls this the Congruence Principle. The One to One Principle (Andersen, 1984), which has independent motivation for its use in language acquisition, promotes an expectation on the part of learners that each new verb morpheme that they notice will have one and only one meaning and function. This reinforces the learner's perception that the past morpheme signals completion or an end point, the progressive morpheme ongoing activity, and so forth. Following Rosch (1973), Ross (1973), Lakoff (1987), and J. R. Taylor (1989), we adopt a prototype explanation for the conservative behavior of learners in their assignment of morphological tense-aspect marking to verbs. We argue that learners initially infer from the input directed to them the most prototypical meaning of each inflection and associate the inflection with the most prototypical members of each semantic aspect class of verbs. Gradually they relax this restriction and extend the inflections to less prototypical verbs and then to other verb classes. Space does not permit a fuller discussion of this proposal, but see Shirai (1991) and Andersen and Shirai (in press). DISCOURSE MOTIVATIONS FOR TENSE-ASPECT ACQUISITION AND USE Native Speaker and Learner Usage
X,
The preceding account for learners' use of verbal morphology appeals to cognitive principles that appear to be specific to the acquisition of the verbal morphemes. When we are confronted with the observation that native speakers also appear to follow the aspect hypothesis, although as a strong statistical tendency rather than a strict absolute adherence, we are forced to account for this behavior. We believe that the same principles that guide learners also guide native speaker users of the verb morphology, especially the Relevance Principle and the Congruence Principle. Native speakers display a more diverse repertoire of forms and functions for these forms than learners because (a) the native speakers have either acquired or created more marked functions for the tense-aspect forms that learners initially internalize in more conservative unmarked uses, based on the prototypical meaning for each form, and (b) the native speakers have added to their repertoire additional forms and constructions beyond the reach of less experienced language learners. It is for this reason that native speakers exhibit a strong statistical tendency in the direction of the aspect hypothesis rather than an absolute adherence. In addition, fully proficient, experienced native speakers talk about more abstract, specialized, and complex topics than learners, whether young children acquir-
Discourse Motivations
147
ing their native languages or second language learners. This more complex discourse of native speakers includes many departures from the pairings of verb with inflection dictated by the aspect hypothesis, but usually not enough to constitute counterevidence to the distributional bias hypothesis. We believe that both learners and proficient native speakers follow especially the Relevance Principle and the Congruence Principle and even the One to One Principle because of the communicative function in live ongoing discourse. Not all events, activities, and states referred to by a speaker are of equal importance to the speaker's and the listener's goals and purposes. Some inflectable verbs are central to the speaker's topic, position, or ultimate goal, and others provide background information of various types. The Relevance Principle and the Congruence Principle, we believe, are especially sensitive to discourse-pragmatic function. Verbs and inflections are chosen to fulfill the needs of the speaker in ongoing discourse, and the Relevance Principle and Congruence Principle together govern the speaker's choices in such a way that both learners and proficient speakers exhibit the type of distributional bias we have discussed. We believe that the Congruence Principle accounts especially well for the observed similarities in the distribution of inflections in the speech of learners and of native speakers. When speakers say something about an event they participated in or experienced, they choose a verb appropriate for reporting the event and, almost simultaneously, the appropriate verb inflection for that verb. They will probably choose an accomplishment or achievement verb to report the event and a past or perfective inflection for the verb, depending on the language they are speaking. Both the choice of the verb and the choice of the inflection are motivated by the information the speaker wants to convey. This is probably how inflections most congruent with the meaning of a verb get attached to verbs in both learners' and proficient native speakers' speech. In a large number of such pairings of lexical verb and verb inflection, the inflection will be congruent with the aspectual meaning of the verb and the inflection adds little new information, other than highlighting the aspectual meaning already inferrable from the meaning of the verb. It is where the particular verb inflection is least expected that it will have the greatest information load: a progressive inflection on an achievement verb, because without the progressive inflection we would have to assume a nondurative interpretation, and a past inflection on a stative verb, because, as states normally continue without interruption, without the past inflection we would not otherwise understand that the state only existed in a past time frame. Thus, a relatively small number of instances of verb inflections in the speech of fully proficient native speakers would be extremely important in conveying the speaker's perspective—those uses where the meaning of the verb inflection is not at all congruent with the meaning of the verb and not at all predictable from the context. This then brings us to the frequency counts, which provide evidence for the distributional bias hypothesis. An overall frequency count would reveal that each verbal morpheme is used most frequently with verbs whose aspectual meaning is congruent with the meaning of the morpheme. Therefore, the distribution of the verb morphemes in adult native speaker speech and in child language learners and
148
Roger W. Andersen and Yasuhiro Shirai
second language learners all have the same source: the Relevance Principle and the Congruence Principle (used along with the One to One Principle for learners), with the unmarked meaning being provided by the prototype for each inflection and each verb. The real development in language learning and the real virtuosity in native speaker speech occurs when speakers gain expressive control over the use of each morpheme so that they can intentionally (although presumably subconsciously) impose their own perspective on each proposition through the use of the morphology. For learners, this will include adding the marked uses of existing forms to the more conservative uses with which they began their acquisition of the forms. They will also acquire marked forms that were not accessible to them in the earlier stages of their learning. Extension of Prototypical Meaning What mechanism must we assume for learners to be able to extend the domain of application of a particular verb inflection beyond the earliest conservative form: meaning categories? First, we must assume that learners initially infer9 from the input the most prototypical meaning of each inflection: "action in progress at that moment" for progressives, "completed action" for past and perfective marking (and perhaps "change of state" as a result of the completion), and "continued existence" for present marking. Learners will eventually look for ways to impose their own perspective on situations that can be viewed from different perspectives. In searching for the linguistic means to do this, they are able to impart to each inflection the prototypical meaning they first associated with it and then use that meaning to impose that type of perspective on a situation that does not normally have that attribute.10 For example, they can infer from cases like broke and fell down the notion of "completed," which applies to both the individual situation and the inflection (from their standpoint), and at some point free the meaning of the inflection from the array of meanings of the prototypical situations that allowed them to infer that meaning and impose on a less prototypical situation that same perspective. For example, they can treat the activity run, which has no inherent end point, as having completion by using the past form ran and thus refer to a particular situation that is no longer part of the present moment. They can also generalize further to a set of similar situations, thus extending the past inflection to include past habituals." The notion of prototype nicely explains conflicting claims in the literature regarding whether aspect or tense is coded by early morphology. Stephany (1981), for example, claimed that children's early use of past tense morphology is nondeictic and aspectual, whereas Weist et al. (1984) claimed that it codes tense. In fact, it may be both aspectual and deictic. J. R. Taylor (1989), in discussing prototype-based development of past morphology, suggested that in view of the fact that children's "earlier uses [of past tense] are restricted to items like fall, drop, slip, crash, break, which designate highly punctual events," the central meaning of past tense may be "completion in the immediate past of a punctual event, the consequences of which are perceptually salient at the moment of speaking" (p. 243). If children acquire the semantic category "past" by starting with the prototype of the category, it is natural
Discourse Motivations
149
Semantic Category PERFECTIVE
Semantic Category PAST
\
PROTOTYPE PAST/PERFECTIVE [+unitary] [+result state] [^punctual] [+past] Figure 2. The relationship between perfective and past. that they will start with situations that are punctual, completive, and involve change of state. This in fact is also argued to be the prototype of perfective aspect. Based on their extensive cross-linguistic survey, Dahl (1985) claimed that the prototypical case of perfective is as follows: It will typically denote a single event, seen as an unanalyzed whole, with a well-defined result or end-state, located in the past. More often than not, the event will be punctual, or at least, it will be seen as a single transition from one state to its opposite, the duration of which can be disregarded, (p. 78) It appears that his description also fits the description of the prototype past. This convergence of the prototype past and the prototype perfective is, we argue, the reason why early past marking is taken to be perfective marking. The overlap (shown in Figure 2) of perfective and past semantic categories is the notion that children give marking to, using early past/perfective morphology.12 It is not clear how much of the mechanism that we must assume allows learners to extend the scope of these inflections is conceptual and universal and how much is dependent on the learners finding exemplars in the language addressed to them to allow them to extend their use of the inflections beyond the initially conservative prototypical uses. We would speculate that there will always be a tendency to extend the domain of application of an inflection in certain directions but that some of these tendencies are curtailed when other tendencies are more directly modeled in the input. These other tendencies are thus discovered faster precisely because they are clearly modeled. We will not at this time pursue further this interesting and important question.
150
Roger W. Andersen and Yasuhiro Shirai
States, Iteratives, and Habituals Events in our lives are bounded (see Chung & Timberlake, 1985; Hopper & Thompson, 1980), but the circumstances surrounding those memorable events tend to be unbounded. In simple terms, bounded means that a physical entity has boundaries, a point beyond which it does not exist, and an event has an end point (and perhaps a discrete beginning point). Events thus can be related to a specific time period, whereas unbounded situations and conditions exist and persist across and, therefore, beyond the particular time period in which a verbally reported event occurred. From a discourse-semantic perspective, the unbounded "circumstances" that provide the background against which the event unfolds are stative in almost the same sense that a person or object has a certain quality. We typically reserve "stative" for the local lexically expressed notion—he has an odd birthmark, that man really likes to drink beer—and use terms like "habitual," "customary," "repeated," "common," "usual," and so forth for consistent, continuing behavior—he always talks too loud, she got mad because he snored (every night), they typically overcharge^) their customers. One common observation in second language speech is that speakers frequently mark real unitary realized punctual and telic events with past or perfective morphology while in the same past-anchored episode use stative verbs as well as verbs that refer to habitual, customary, typical, or usual actions and behavior in the base form (or, for Romance languages, third person singular present, which Bybee, 1985, argued is the least marked form, roughly equivalent to the base form for English). Thus, it appears that speakers frequently treat states (e.g., a stative verb used to describe someone's appearance, mood, or temperament) and actions and situations that similarly seem to be typical of the person, environment, or time period the same way—as characteristics that go beyond the reported time period, thus unbounded. Thus, one achievement verb used in a past time frame may be inflected with a past or perfective marker because it encodes information about a single realized bounded event, and another achievement verb (even the same lexical item) would go unmarked because it refers to customary, habitual, typical, or usual behavior. We believe that it is this apparent13 association of explicit marking with real, realized events (past/perfective) and, perhaps, depending on the language, simultaneous but temporary actions or processes (progressive) that make states as well as habitual/typical and repeated actions the unmarked default. This is consistent with a discourse account for the observed patterns of morphological marking on verbs. Thus, it is not that speakers mark verbs with some zero marking (base form in languages like English or third person singular present—typically also zero—in highly inflected languages) to explicitly encode their status as states/habituals. Rather, it appears that they simply refer to their existence as possibly permanent, possibly temporary qualities, conditions, and characteristics of the situation, people, or things being talked about. As Sankoff (1990) argued, such explicit markers are used as facultative encoding of meaning; when speakers choose to explicitly encode a particular meaning, they choose the linguistic device they associate with that meaning, and lack of that device (zero) does not automatically imply that the speaker
Discourse Motivations
151
intends to encode the opposing meaning. They are simply not explicitly encoding the meaning associated with the linguistic device (inflection, auxiliary, etc.). In this case, then, not all achievement verbs in past episodes receive past/perfective markers, only those the speakers choose at the moment to mark explicitly. Stative verbs and "stative" scenarios (habitual/iterative/customary/usual) "receive" zero marking (i.e., remain uninflected in English, receive third person singular present marking in Spanish) not because states somehow repel inflections or because the learner's interlanguage grammar has restrictions on inflecting states. Rather, they do not -J.. identify states or "stative" scenarios as something needing explicit marking. This is a dynamic online discourse decision. The measurable result is the low (often nonexistent) past inflection of states. But we should not confuse the learner's motivation, that is, the discourse motivation we are claiming guides the speaker, with some syntactic or semantic constraint against marking states. With time, in languages that use past/perfective markers on states as well as events and activities, learners will increase their frequency of past/perfective marking and extend the use of the appropriate inflection to more achievements/accomplishments as well as activities and states. When this happens, we interpret this change as reflecting a change in what the speakers are choosing to mark explicitly: They have gradually moved up the prototypical hierarchy to meanings implicationally derived from the initial prototypical meaning. Thus, a real, realized unitary bounded event gradually expands to include typical or habitual or repeated events (therefore, not unitary events) and even nonreal (hypothetical) events (if you CAME more often, this wouldn't happen) as well as states (which are not bounded; she HAD an odd birthmark on her shoulder), reported as part of a past time frame. We thus believe that the semantic attributes of verbs that appear to correlate with verb morphology are really a consequence of the discourse motivation to highlight some, but not other, events and situations with newly perceived morphology. And we interpret principles like the Relevance Principle and the Congruence Principle as being motivated similarly by decisions to explicitly encode a notion like "has end point" or "ongoing at reference time" by choosing both a verb that already has part of that meaning and an inflection that is assigned that meaning in the learner's developing linguistic system. CONCLUSION In conclusion, we have argued that a unified account for the observed phenomena in first and second language acquisition and in native speaker use of tense-aspect morphology is to be found in the discourse function of tense-aspect morphology. We will summarize that argument here. As we have shown, there is considerable agreement across studies on the acquisition of tense-aspect marking. Learners restrict use of verb morphology such as past/ perfective, progressive, and imperfective to a small subset of the verbs to which the morphology could be attached in fluent adult native speakers' language use. We attribute this early conservative use of verb morphology to adherance to (a) the Relevance Principle (which guides learners to look for morphological marking rele-
152
Roger W. Andersen and Yasuhiro Shirai
vant to the meaning of the verb), (b) the Congruence Principle (which guides learners to associate verb morphology with verb types most congruent with the aspectual meaning of the verb inflection), and (c) the One to One Principle (which causes learners to expect each newly discovered form to have one and only one meaning, function, and distribution). We also believe, however, that a learner must have access to prototypicality in linguistic form: meaning relations. This is necessary to account for how learners discover which meaning to assign a one-to-one form: meaning correspondence to. In earlier work (e.g., Andersen, in press), reference was made to adherence to the Subset Principle. Here we have replaced the Subset Principle with prototypicality. Prototypicality is a semantic notion, consistent with the other semantic-based principles discussed. The Subset Principle, which makes a similar prediction, was proposed to handle syntactic relations and, for this reason, we have dropped it from our argumentation. Prototypicality and the One to One Principle account for the observed phenomena without recourse to the Subset Principle. These three cognitive principles (Relevance, Congruence, One to One) plus access to a notion of prototypicality (also a cognitive construct) go a long way toward accounting for language learners' behavior in the realm of tense-aspect marking. If it were not for the findings in support of the distributional bias hypothesis, which applies to native speakers, we would stop here. Fully proficient native speakers exhibit a relative distributional bias in their use of the same verb inflections similar (but definitely not identical) to that found in more absolute terms among learners. We have argued that the learners and adult native speakers alike are subject to the Relevance and Congruence Principles. Learners are much more heavily constrained by the One to One Principle and the restriction of form: meaning relations to the least marked member of a prototype hierarchy. From here we argue that all of these principles follow naturally from the speakers' (both learners and nonlearners) communicative need to distinguish reference to the main point/goal of talk from supporting information, within the tradition of research on grounding and the functions of tense-aspect marking in narratives (Givon, 1982; Hopper, 1979; Hopper & Thompson, 1980; Labov, 1972; Schiffrin, 1981; among others). Thus, speakers, whether learners or not, impose a hierarchical organization on the information they are attempting to convey in terms of the role each bit of information conveys in their overall goals. Some of the information they convey or intend to convey is more directly related to their goal or purpose and other information is intended as supporting information. We argue that in ordinary discourse this causes the observed association between past/perfective marking and achievements and accomplishments, progressive marking and activities, and past imperfective marking and states and activities. Achievements and accomplishments typically fill the central role of laying out events in narration and are logical recipients for past/perfective marking. States and activities typically serve supporting roles, and if they receive inflectional marking they tend to be inflected with progressives (on activities) and past imperfectives. Where learners and nonlearners differ is in the freedom with which the latter, but not the former, can free these morphological markers from the more expected
Discourse Motivations
153
normal use described here to serve their communicative purposes. Fully fluent adult native speakers, thus, exhibit only a strong tendency toward the distributional bias hypothesis, not an absolute mechanical adherence. (If they did, such behavior would be dysfunctional.) It is almost common sensical to observe that learners cannot learn everything simultaneously and instantly. In the realm of tense and aspect marking, it appears that what they first learn are the basic tools they will continue to use later as more fluent speakers.14 We contend that learners are motivated by the same communicative need to distinguish central events (with past/perfective marking) from simultaneous situations (with progressive) and both from static background (initially with present verb forms, much later with past imperfectives, or, for languages like English, with past habitual would/used to). Furthermore, as learners acquire more of the linguistic repertoire they need to use the language they are acquiring, they elaborate on this basic framework rather than abandoning it. Fluent adult native speakers have the same communicative need to distinguish completed events (past/perfective inflections on achievements and accomplishments) from simultaneous processes (progressive marking on activities) and both from more static background (present as well as past imperfective marking and past habituals). This, we argue, is what causes native speakers and learners to exhibit similar distributional biases in their speech. This also goes a long way to account for why learners also differ considerably from fluent native speakers in their actual use of the forms that constitute evidence for the distributional bias. But learners learn from more proficient native (and, often, nonnative) speakers. So we should not minimize the role played by the model provided by the more proficient speakers in causing the learner to construct a linguistic system along the same lines, although in miniature, so to speak, as the proficient native speaker. What is the relative contribution of (a) cognitive predisposition to distinguish semantically the state, process, and event, (b) the claimed communicative need to distinguish (i) static background and (ii) simultaneous events and situations from (iii) central storyline and goal-oriented events, and (c) the model provided by more fluent speakers? Future research needs to explore these competing forces in greater detail. NOTES
1. Kenny (1963, p. 171) attributes this classification system to Aristotle's Metaphysics, Physics, and Nicomachean Ethics. Vendler and Mourelatos also give Aristotle as the original source of these distinctions. See, for example, Mourelatos (1981, p. 193, Footnote 4). 2. The data were provided by Brian MacWhinney through the CHILDES database system (MacWhinney & Snow, 1985) and originally came from Brown's (1973) data on Adam and Eve and Sachs' (1983) data on Naomi. See Shirai (1991) for more details. 3. Activity and accomplishment verbs, which denote action with duration, have an "action-in-progress" meaning when combined with the progressive inflection. 4. See, for example, Comrie (1976) on this point. 5. Nixon's (1986) Chinese subject expressed perfective aspect mostly by using have + verb (base form) instead of past morphology. 6. Shirai and McGhee (1988) studied a Japanese subject and reported more frequent use of past marking on nonstative verbs, which is consistent with the aspect hypothesis. 7. See Weist, Wysocka, and Lyytinen (1991) for an exception. 8. Andersen (1986, 1992) has also shown this for second language acquisition of Spanish and many of
154
Roger W. Andersen and Yasuhiro Shirai
the studies in Table 7 (unpublished studies conducted at UCLA) have done the same for second language acquisition of English. 9. We leave unsettled the important question of how much children come prepared to notice and thus conceptualize distinctions such as "in progress," "completed," and "there is/exists" (state) and how much the particular way the language they are exposed to prepares them to conceptualize in this way. See Slobin's (1982,1985) interesting proposals in this regard. 10. Smith (1983) makes a similar argument. 11. For languages like Spanish, this would involve extending the prototypically durative meaning of the imperfective to repeated and thus habitual situations that can also be viewed as having duration in a more abstract sense. 12. This is consistent with Slobin's (1985) claim that early morphology codes the Result perspective. 13. We must use the term "apparent" here, because time and space prohibit careful documentation of this claim, which is based on years of examination of second language speech data on diverse languages, including primarily English, French, Spanish, German, Papiamentu, and Quechua. 14. If these are children acquiring their native language, we are sure they will attain this status. If these are adult nonnative speakers, there is less certainty that they will get beyond the restrictions we have reported. REFERENCES
Aksu, A. A. (1978). Aspect and modality in the child's acquisition of the Turkish past tense. Unpublished doctoral dissertation, University of California, Berkeley. Andersen, R. W. (1984). The One to One Principle of interlanguage construction. Language Learning, 34, 7795. Andersen, R. W. (1986). The need for native language comparison data in interpreting second language data. Forum Lecture, 1986 TESOL Summer Institute, University of Hawaii at Manoa. Andersen, R. W. (1989a). La adquisicion de la morfologia verbal [The acquisition of verb morphology]. Linguistica, 1,90-142. (English version available from the author) Andersen, R. W. (1989b). The theoretical status of variation in interlanguage development. In S. Gass, C. Madden, D. Preston, & L. Selinker (Eds.), Variation in second language acquisition: Psycholinguistic issues (pp. 46-64). Clevedon, UK: Multilingual Matters. Andersen, R. W. (1989c). The "up" and "down" staircase in secondary language development. In N. Dorian (Ed.), Investigating obsolescence. Studies in language contraction and death (pp. 385-394). Cambridge: Cambridge University Press. Andersen, R. W. (1990). Models, processes, principles, and strategies: Second language acquisition in and out of the classroom. In B. VanPatten & J. F. Lee (Eds.), Second language acquisition—Foreign language learning (pp. 45-78). Clevedon, UK: Multilingual Matters. (Reprinted from IDEAL, 1988,3,111-138) Andersen, R. W. (1991). Developmental sequences: The emergence of aspect marking in second language acquisition. In T. Huebner & C. A. Ferguson (Eds.), Crosscurrents in second language acquisition and linguistic theories (pp. 305-324). Amsterdam: John Benjamins. Andersen, R. W. (1992, November). The insider's advantage. Paper presented at the annual meeting of the Linguistic Society of Italy, Siena. Andersen, R. W. (1993). Four operating principles and input distribution as explanations for underdeveloped and mature morphological systems. In K. Hyltenstam & A. Viborg (Eds.), Progression and regression in language (pp. 309-339). Cambridge: Cambridge University Press. Andersen, R. W., & Shirai, Y. fin press). Primacy of aspect in first and second language acquisition: The pidgin/creole connection. In W. Ritchie & T. Bhatia (Eds.), Handbook of language acquisition. New York: Academic Press. Antinucci, F., & Miller, R. (1976). How children talk about what happened. Journal of Child Language, 3, 169-189. Bardovi-Harlig, K. (1992). The relationship of form and meaning: A cross-sectional study of tense and aspect in the interlanguage of learners of English as a second language. AppliedPsycholinguistics, 13,253-278. Bardovi-Harlig, K., & Bergstrom, A. (1993, February). Tense and aspect in SLA and FLL Learner narratives in English (SL) and French (FL). Paper presented at the 3rd SLA-FLL Conference (Second language Acquisition—Foreign Language Learning), West Lafayette, IN. Bardovi-Harlig, K., & Reynolds, D. W. (1993, March). The acquisition of tense, lexical aspect, and the effect of positive evidence. Paper presented at the 27th Annual TESOL convention, Atlanta, GA. Bayley, R. (1991, December). Variation in interlanguage tense marking. Paper presented at the annual meeting of the Linguistic Society of America, Chicago. Bergstrom, A. (1993, March). The expression of temporal reference by English speaking learners of French:
Discourse Motivations
155
Report on the cloze. Paper presented at the 13th meeting of Second Language Research Forum, University of Pittsburgh, Pittsburgh, PA. Bickerton, D. (1981). Roots of language. Ann Arbor, MI: Karoma. Bloom, L, & Harner, L. (1989). On the developmental contour of child language: A reply to Smith and Weist. Journal of Child Language, 16,207-216. Bloom, L, Lifter, K., & Hafitz, J. (1980). Semantics of verbs and the development of verb inflection in child language. Language, 56,386-412. Bronckart, J. P., & Sinclair, H. (1973). Time, tense, and aspect. Cognition, 2,107-130. Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press. Bybee, J. L. (1985). Morphology: A study of the relation between meaning and form. Amsterdam: John Benjamins. Chung, S., & Timberlake, A. (1985). Tense, aspect, and mood. In T. Shopen (Ed.), Language typology and syntactic description: Vol. 3. Grammatical categories and the lexicon (pp. 202-258). Cambridge: Cambridge University Press. Comrie, B. (1976). Aspect. Cambridge: Cambridge University Press. Cushing, S. T. (1987). Use of verb morphology in the English interlanguage of a native speaker of Serahuli. Unpublished manuscript, Applied Linguistics, University of California, Los Angeles. Dahl, O. (1985). Tense and aspect systems. Oxford: Basil Blackwell. Economides, P. J. (1985). The expression of tense and aspect in the English interlanguage of a Vietnamese child. Unpublished master's thesis, University of California, Los Angeles. Flashner, V. (1982). The English interlanguage of three native speakers of Russian: Two perspectives. Unpublished master's thesis, University of California, Los Angeles. Giv6n, T. (1982). Tense-aspect-modality: The Creole prototype and beyond. In P. Hopper (ed.), Tense-aspect: Between semantics and pragmatics (pp. 115-163). Amsterdam: Benjamins. Hopper, P. J. (1979). Aspect and foregrounding in discourse. In T. Givon (Ed.), Syntax and semantics: Vol. 14. Discourse and syntax (pp. 213-241). New York: Academic Press. Hopper, P. J., & Thompson, S. A. (1980). Transitivity in grammar and discourse. Language, 56,251-299. Huang, C. (1993). Distributional biases of verb morphology in native and non-native English discourse. Unpublished M.A. thesis, University of California, Los Angeles. Hyams, N. (1986). Language acquisition and the theory of parameters. Dordrecht: Reidel. Kaplan, M. (1987). Developmental patterns of past tense acquisition among foreign language learners of French. In B. VanPatten (Ed.), Foreign language learning: A research perspective (pp. 52-60). Rowley, MA: Newbury House. Kenny, A. (1963). Action, emotion, and will. London: Routledge and Kegan Paul. Kihlstedt, M. (in press). A longitudinal study of two university students' use of verb forms and functions in their French interlanguage. In B. Hammarberg (Ed.), Problem, process, product in language learning: Papers from the Stockholm-Abo Conference, 21-22 October, 1992. Stockholm: Institute of Linguistics, Stockholm University. Kumpf, L. (1982). Tense, aspect, and modality in interlanguage: A discourse-functional perspective. Paper presented at the 16th Annual TESOL convention, Honolulu, HA. Kumpf, L. (1984). Temporal systems and universality in interlanguage: A case study. In F. R. Eckman, L. H. Bell, & D. Nelson (Eds.), Universal ofsecond language acquisition (pp. 132-143). Rowley, MA: Newbury House. Labov, W. (1972). The transformation of experience in narrative syntax. In W. Labov, Language in the inner city: Studies in the Black English vernacular (pp. 354-405). Philadelphia: University of Pennsylvania Press. Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind. Chicago: Chicago University Press. Lee, H. S. (1991). Tense, aspect, and modality: A discourse-pragmatic analysis of verbal affixes in Korean from a typological perspective. Unpublished doctoral dissertation, University of California, Los Angeles. Long, M. H., & Sato, C. J. (1984). Methodological issues in interlanguage studies: An interactionist perspective. In A. Davies, C. Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp. 253-279). Edinburgh: Edinburgh University Press. MacWhinney, B., & Snow, C. E. (1985). The Child Language Data Exchange System. Journal of Child Language, 12,271-296. Meisel, J. M. (1987). Reference to past events and actions in the development of natural second language acquisition. In C. W. Pfaff (Ed.), First and second language acquisition (pp. 206-224). New York: Newbury House. Mishina, S. (1993). Second language acquisition of verb morphology by three Japanese speakers. Unpublished manuscript, Applied Linguistics, University of California, Los Angeles. Mourelatos, A. P. (1981). Events, processes, and states. In P. J. Tedeschi & A. Zaenen (Eds.), Syntax and semantics: Vol. 14. Tense and aspect (pp. 191-212). New York: Academic Press. Nixon, N. (1986). Tense/aspect in the English interlanguage of a native Mandarin speaker. Unpublished master's thesis, University of California, Los Angeles.
156
Roger W. Andersen and Yasuhiro Shirai
Ramsay, V. (1989a). The acquisition of the perfective/imperfective aspectual distinction by classroom learners. In R. Carlson, S. DeLancy, S. Gildea, D. Payne, & A. Saxena (Eds.), Proceedings of the fourth meeting of the Pacific Linguistics conference (pp. 374-404). Eugene: University of Oregon, Department of Linguistics. Ramsay, V. (1989b, December). On the debate over the acquisition of aspect before tense: Setting the record straight. Paper presented at the annual conference of the American Association for Applied Linguistics, Washington, DC. Ramsay, V. (1990). Developmental stages in the acquisition of the perfective and the imperfective aspects by classroom 12 learners of Spanish. Unpublished doctoral dissertation, University of Oregon, Eugene. Rispoli, M., & Bloom, L. (1985). Incomplete and continuing: Theoretical issues in the acquisition of tense and aspect. Journal of Child Language, 12,471-474. Robison, R. E. (1990). The primacy of aspect: Aspectual marking in English interlanguage. Studies in Second Language Acquisition, 12,315-330. Robison, R. E. (1993). Aspectual marking in English interlanguage. A cross-sectional study. Unpublished doctoral dissertation, University of California, Los Angeles. Rosch, E. H. (1973). On the internal structure of perceptual and semantic categories. In T. E. Moore (Ed.), Cognitive development and the acquisition of language (pp. 111-144). New York: Academic Press. Ross, J. R. (1973). A fake NP squich. In C. J. N. Bailey & R. Shuy (Eds.), New ways of analyzing variation in English (pp. 96-140). Washington, DC: Georgetown University Press. Rothstein, G. (1985). The expression of temporality in the English interlanguage of a native Hebrew speaker. Unpublished master's thesis, University of California, Los Angeles. Sachs, J. (1983). Talking about the there and then: The emergence of displaced reference in parent-child discourse. In K. E. Nelson (Ed.), Children's language (Vol. 4, pp. 1-28). Hillsdale, NJ: Erlbaum. Sankoff, G. (1990). The grammaticalization of tense and aspect in Tok Pisin and Sranan. Language Variation and Change, 2,295-312. Schiffrin, D. (1981). Tense variation in narrative. Language, 57,45-62. Shirai, Y. (1990). The defective tense hypothesis: Is there a distributional bias in the input? Unpublished manuscript, Applied Linguistics, University of California, Los Angeles. Shirai, Y. (1991). Primacy of aspect in language acquisition: Simplified input and prototype. Unpublished doctoral dissertation, University of California, Los Angeles. Shirai, Y., & McGhee, R. (1988). An interlanguage analysis of a Japanese ESL learner. Unpublished manuscript, Applied Linguistics, University of California, Los Angeles. Slobin, D. I. (1982). Universal and particular in the acquisition of language. In L. R. Gleitman & E. Wanner (Eds.), Language acquisition: The state of the art (pp. 128-172). Cambridge: Cambridge University Press. Slobin, D. I. (1985). Crosslinguistic evidence for the language-making capacity. In D. I. Slobin (Ed.), The crosslinguistic study of language acquisition: Vol. 2. Theoretical issues (pp. 1157-1249). Hillsdale, NJ: Erlbaum. Smith, C. S. (1980). The acquisition of time talk: Relations between child and adult grammars. Journal of Child Language, 7,263-278. Smith, C. S. (1983). A theory of aspectual choice. Language, 59,479-501. Smith, C. S., & Weist, R. M. (1987). On the temporal contour of child language: A reply to Rispoli and Bloom. Journal of Child Language, 14,387-392. Stephany, U. (1981). Verbal grammar in modern Greek early child language. In P. S. Dale & D. Ingram (Eds.), Child language: An international perspective (pp. 45-57). Baltimore, MD: University Park Press. Taylor, H. (1987). Tense/aspect: A longitudinal study of a native Spanish speaker. Unpublished manuscript, Applied Linguistics, University of California, Los Angeles. Taylor, J. R. (1989). Linguistic categorization: Prototypes in linguistic theory. Oxford: Oxford University Press. Vendler, Z. (1967). Linguistics in philosophy. Ithaca, NY: Cornell University Press. Weist, R. M. (1989). Time concepts in language and thought: Filling the Piagetian void from two to five years. In I. Levin & D. Zakay (Ed.), Time and human cognition: A life-span perspective (pp. 63-118). Amsterdam: Elsevier. Weist, R. M., Wysocka, H., & Lyytinen, P. (1991). A cross-linguistic perspective on the development of temporal systems. Journal of Child Language, 18,67-92. Weist, R. M., Wysocka, R, Witkowska-Stadnik, K., Buczowska, E., & Konieczna, E. (1984). The defective tense hypothesis: On the emergence of tense and aspect in child Polish. Journal of Child Language, 11, 347-374. Wenzell, V. E. (1989). Transfer of aspect in the English oral narratives of native Russian speaker. In H. W. Dechert & M. Raupach (Eds.), Transfer in language production (pp. 71-97). Norwood, NJ: Ablex. Yoshitomi, A. (1992). Primacy of aspect: Additional support for the hypothesis based on an interlanguage analysis of a Chinese speaker's ESL. Unpublished manuscript, Applied Linguistics, University of California, Los Angeles.