Optional Verb Raising in L2 Acquisition: Evidence from L1-Russian Learners of English Tania Ionin and Ken Wexler Massachusetts Institute of Technology
[email protected] [email protected]
Abstract This paper examines the phenomenon of optional thematic verb raising in the L2-English of adult L1-Russian speakers. The study finds that both finite verbs and participles undergo optional verb raising in L2-English, contra previous accounts linking verb raising to finiteness. It is also shown that verb-adverb order in L2-English is not a result of heavy-NP shift or of transfer. It is suggested that optional verb raising in the L2 is due to unset feature strength on a functional head lower than Tense.
adverbs) a verb may raise in L2-English. While investigating these issues, we also looked for what role transfer may play – how would the learners’ L1, Russian, a language without thematic verb raising, affect their acquisition of English? This paper is organized as follows. Section 2 provides some theoretical background on verb raising. Section 3 presents a brief (and thus necessarily incomplete) overview of analyses of verb raising in the L2. Section 4 reports the methods and results of our study with L1-Russian learners of English. The results are discussed in more detail in Section 5. Section 6 concludes the paper.
1. Introduction Research has shown that L2-English learners frequently allow optional raising of thematic verbs, a process not allowed in normal L1-English. This phenomenon has been found for L2-learners whose L1 allows verb raising (e.g., French – White 1990/91, 1992) as well as for those learners whose L1 does not allow verb raising (e.g., Chinese - Eubank, Bischof, Huffstutler, Leek and West, 1997). It has moreover been found (White 1992) that L2-English learners allow optional raising of thematic verbs past adverbs, but not past negation. Thus, L2-learners are likely to find both sentences in (1) grammatical, even though (1) is not good in English; on the other hand, they know that (2) is ungrammatical. (In these examples, * means 'unattested in the L2').
2. Linguistic Background We follow Pollock (1989) in assuming that frequency/manner adverbs adjoin to the left of the VP. We further assume, after Chomsky (1993, 1995) that a verb must check V features on INFL. If, in any given language, V features on INFL are strong, the verb raises to INFL overtly (at PF). If V features are weak, the verb raises covertly (at LF). English is considered to have weak V features in this framework. The verb therefore stays inside the VP at PF, and follows the adverb. This is illustrated in (3). An example of a language with strong V-features is French (see Pollock 1989): the verb raises to INFL and precedes the adverb at PF, as illustrated in (4).
(1)
(3)
(2)
a. b.
John slowly eats soup. John eats slowly soup.
a. b.
John does not eat soup. *John eats not soup.
The study reported here investigated thematic verb raising in the L2-English of L1-Russian speakers. We were interested in addressing issues surrounding verb raising that have not, to our knowledge, previously been discussed in L2-literature. First, we looked at whether optional raising would affect participles as well as finite verbs. Second, we strove to rule out heavy-NP shift as an alternative explanation of verb-adverb order in the L2. And third, we investigated how far (past which kinds of
a.
John often eats soup.
b.
IP John I
VP
often eats (4)
a. soup
soup
Jean mange souvent la soup. John eats often the
b)
IP Jean I VP mange souvent t
la soup
Russian, like English, has been argued to lack verb-raising (i.e., to have weak V features on INFL) (per Bailyn, 1995). The preferred order in Russian is adverb-verb, as illustrated in (5). (5)
a. b.
Ivan chasto est sup John often eats soup ??Ivan est chasto sup John eats often soup
And finally, we’d like to briefly mention the split-INFL framework of Pollock (1989). In the split-INFL structure in (6), negation comes between Tense and Agr, while adverbs attach to the left of the VP, thus following both Tense and Agr. White (1992) used this framework to suggest an explanation for the adverb/negation difference in optional verb raising she found for L2-English. An L2-learner could be positing a strong Agr but a weak Tense in English, thus raising the verb only as far as Agr. A verb in Agr would precede adverbs but follow negation. We will come back to the negation/adverb difference in Section 4. (6)
Split-INFL structure TP DP T
NegP
Neg
AgrP
Agr
VP
(Adv) V (DP) 3. Models of Verb Raising in L2A We now turn briefly to the major accounts of verb raising in the L2. First, the Full Transfer/Full Access (FT/FA) model of Schwartz and Sprouse (1994, 1996) argues for full transfer of categories and features from the L1 to the L2. In this framework, L2-learners would initially hypothesize verb
raising for the L2 if and only if verb raising exists in their L1. This model successfully accounts for optionality of thematic verb raising in the L2-English of L1-French learners (White 1990/91, 1992). On the other hand, under the Weak Transfer / Valueless Features (VF) hypothesis of Eubank (1993/94, 1996), categories transfer from the L1 to the L2 but the associated feature strength does not. The feature strength is underspecified or inert in L2 acquisition: it is neither weak nor strong; this results in optional verb raising. Under this model, thematic verb raising is optional in L2 acquisition regardless of the L1. The VF model receives support from findings that speakers of Chinese, which does not allow verb-raising, nevertheless allow optional verb raising in L2-English (Eubank and Grace 1996, Eubank et al. 1997). However, Yuan’s (2000) finding that L1-English speakers do not allow verb-raising in L2-Chinese is problematic for this proposal. Finally, the Minimal Trees (MT) hypothesis of Vainikka and Young-Scholten (1994, 1996) argues for transfer of only lexical projections (e.g., the VP) from the L1. Functional projections emerge gradually, and the verb raising parameter value is not set until a fully specified AgrP has been acquired. Optional verb-raising takes place while the functional projection is underspecified for feature strength. Importantly, both the VF and MT models tie verb raising to finiteness. Under both models, feature strength on INFL is not set until the relevant finiteness morphology has been acquired. On the other hand, Beck (1996, 1998) argues for very local impairment in the domain of feature strength. Beck (1998) found that even those L2-German learners who had acquired the German agreement paradigm continued to allow optional verb placement with respect to adverbs. However, since German is a V2 language, it is not clear whether the optionality for these L2-learners lies on INFL or on Comp. 4. The Study: L1-Russian Learners of English We now move to the description of our own study with L2-English learners. We asked whether L1-Russian learners of English would ever allow verb-adverb order in English, and, if so, what explanation may best account for this phenomenon. First, we looked at the role of transfer. The FT/FA model would predict that speakers of
Russian, a language without thematic verb raising, should disallow verb raising in their L2. Secondly, we looked closely at the relationship between verb raising and finiteness, which is predicted under the VF and MT models. Tests of this prediction (see, e.g., Eubank et al. 1997) have usually examined learners' use of overt agreement morphology vs. their knowledge of verb/adverb order in the L2. We posed the question differently, asking whether L2-learners would optionally raise participles as well as finite verbs. Since the MT and VF models tie verb raising to acquisition of finiteness, they would not predict optional raising of non-finite verbs, such as participles. Finally, we set out to rule out heavy-NP shift as an alternative explanation of verb-adverb order in L2-English. More details on heavy-NP shift can be found in section 4.2. An additional point that we examined was verb raising past different types of adverbs. This issue is discussed in the appendix. 4.1. Methods The study was piloted with seven native English speakers, who performed as expected (i.e., always choosing the adverb-verb order over the verb-adverb order, etc.). Only L2-learners' results are reported here. 4.1.1. Participants Participants in this study were 35 L1-Russian learners of English, 26 females and 9 males. They ranged in age from 18 to 60, with the mean age of 45. While most of the participants had received English instruction in Russia during their childhood, intensive exposure to English did not begin for any of the participants until after their arrival in the U.S. All of the participants were past 18 years of age when this intensive exposure began. Individual intensive exposure ranged from 2.5 months to 10 years. All participants resided in a large metropolitan area in the U.S. (Boston). Participants were recruited through advertisement as well as through local ESL courses. They were reimbursed for their participation. 4.1.2. Instruments The participants were asked to complete three untimed, paper-and-pencil tests. The main instrument was a grammaticality preference task, with 76 items. Each item consisted of a pair of sentences, one grammatical and one ungrammatical, as illustrated in (7). The learners were asked to say whether A or B was the more
grammatical sentence; they also had the option of marking both sentences as equally good. A participant’s response was scored as correct when only the grammatical variant was chosen (e.g., A in (7)). Choosing the ungrammatical variant, or choosing both A and B, was considered an error. (7)
a. b.
Jim slowly eats hot soup. Jim eats slowly hot soup.
The participants were also asked to complete a grammaticality judgment task with 36 items. Each item was an English sentence which the participants had to judge as good or bad. The grammaticality preference and judgment tasks included many different item types testing for various aspects of verb raising. We will discuss the relevant item types in the results section. For full discussion of all item types, please see Ionin (2001). And finally, the learners were given a portion of the University of Michigan proficiency test for ESL learners. The learners completed only the written portion of the test, which consists of 30 multiple-choice items and comes with a standardized scoring scale. Since length and type of exposure to English varied so widely among the L2-learners, this test was used to gauge the learners’ overall level of English. The preference task and the judgment task were counterbalanced for order of presentation. The proficiency test was always given last. All tests were administered by an investigator fluent in both Russian and English. For the grammaticality preference and judgment tasks, participants were provided with a list of English words that could be unfamiliar to them, with the corresponding Russian translations. 4.2. Results Results of the proficiency test showed that six of the learners were beginners, 21 were intermediate, and eight were advanced. Correlations between learners' proficiency and performance on various item types are given in the appendix. In looking at the results of the other two tests, we chose to concentrate only on the non-beginner learners i . For the remainder of this paper, we report results from the 29 intermediate/advanced learners, excluding the six beginners. 4.2.1. Raising past manner adverbs First, we looked at verb placement with respect to four manner adverbs (quietly, slowly, quickly, and carefully). The relevant test items are from
the grammaticality preference task. Four items tested placement of finite verbs with respect to manner adverbs (as in (8)), and four items tested placement of participles with respect to manner adverbs (as in (9)). All items included a direct object. (8)
a. b.
(9)
a. b.
The children quietly watch television. The children watch quietly television. The girls were slowly opening a window. The girls were opening slowly a window.
The results were as follows. Recall that a correct response means choosing only the adverb-verb order as grammatical. For finite verbs, learners were, on average, 56% correct. For participles, learners were also, on average, 56% correct. Needless to say, there was no significant difference between responses to finite verb vs. participle items. Two important conclusions can be drawn from these findings. First, L1-Russian learners of English seem to allow thematic verb raising past adverbs in English. Second, there is no relationship between verb finiteness and the verb’s ability to raise: non-finite participles raise as much as finite main verbs. 4.2.2. Raising past negation Next, we looked at verb raising past negation. The relevant test items once again come from the grammaticality preference task. There were four items with finite verbs and negation (10), and four items with participles and negation (11). (10)
(11)
a. b.
Peter does not like his neighbors. Peter likes not his neighbors.
a. b.
My son is not watching TV. My son is watching not TV.
The results were as follows. The learners were correct (i.e., chose the negation-verb order), on average, 99% of the time for finite verbs as well as for participles. Thus, we can conclude that the L2-learners do not allow thematic verb raising past negationii. 4.2.3. Ruling out heavy-NP shift Why should the L2-learners raise thematic verbs past adverbs but not past negation? A possible alternative explanation is that verb raising is not
taking place at all, but that the adverb-verb order is attributable to heavy-NP shift of the direct object. (This explanation has been previously mentioned in L2-literature but has not, to our knowledge, ever been fully ruled out). It is possible that the L2-learners start out with the underlying structure such as “John eats soup slowly” (see (12)), and then shift the object, “soup”, to the right of the VP. (12)
John eats [soup] slowly John eats ____ slowly [soup]
Heavy-NP shift is a process allowed in L1-English. It is quite conceivable that the L2-learners could be overusing heavy-NP shift in English (shifting objects which are not particularly heavy, as in (12)), because of transfer from Russian. In Russian, new information focus typically falls on the rightmost element, so an NP can be shifted rightward as long as it presents new information. Thus, before we can make any claims about verb raising in the L2, we need to rule out heavy-NP shift as an explanation of verb-adverb order. We did so by using pronominal objects, which cannot undergo rightward shift in English or Russianiii. The relevant test items, from the grammaticality preference task, were eight items with manner adverbs and pronominal objects. Half contained finite verbs, and half contained participles. An example is given in (13). The two options were always preceded by a question, which provided an antecedent for the pronoun. The pronoun that we used was always it, a weak pronoun that clearly cannot undergo rightward shift. (13)
What did Polly do with the apple? a. She quickly ate it b. She ate quickly it
The results are reported in (14). There was no significant difference in responses to items with full NP complements vs. items with pronominal complements (the two-tailed p-value was .10). (14)
results: average % correct response
finite verb participle
full NP complement 56.0%
it complement 58.6%
56.0%
64.6%
It thus seems that L2-learners are quite willing to accept verb-adverb as grammatical even though this order cannot be derived by heavy-NP shift.
However, we must be cautious – it is conceivable that the L2-learners were in fact shifting the pronoun rightward. We controlled for this possibility by the grammaticality judgment task. This task included six items in which the pronoun was shifted rightward, but there was no adverb (so verb raising could not be involved). This is illustrated in (15). Six control items, in which the pronoun was in its base position, were also included (16)iv. (15)
What did you do with the car? I put in the garage it
(16)
Did you hear the new symphony? Yes, we heard it on Monday
We found that the learners were quite reluctant to accept items with shifted pronouns, such as (15). Acceptance rate for this item type was only 17%, as opposed to 95% acceptance for items with non-shifted pronouns, such as (16). Moreover, only six of the 29 L2-learners ever accepted shifted pronouns at all. If we exclude these learners and look at the results of the remaining 23, we find that performance on the grammaticality preference task does not change significantly – learners are still quite willing to accept verb-adverb order with pronominal as well as full-NP complements. Thus, we can conclude two things. First, the L2-learners correctly do not allow heavy-NP shift of pronouns in English. Second, they nevertheless allow a verb to be separated from its pronominal complement by an adverb. This suggests that the verb-adverb order is in fact due to verb raising, and not to heavy-NP shift. 5. Discussion To summarize the main findings of this study: First, we have seen that L2-learners whose L1 (Russian) and L2 (English) do not allow thematic verb raising nevertheless optionally raise thematic verbs in their L2. This is not what we would expect under the FT/FA model. Secondly, optional thematic verb raising occurs past adverbs but not past negation. This suggests that the verbs are not raising all the way up to Tense. Third, we have shown that verb raising is not related to verb finiteness: participles as well as finite verbs undergo optional raising. However, both the VF and MT models, which tie verb raising to acquisition of finiteness, would predict raising of finite verbs only.
And finally, we have effectively ruled out heavy-NP shift as an explanation of the verb-adverb order. Our findings seem to agree with the Local Impairment hypothesis of Beck (1996, 1998), described in Section 3. However, Beck suggests that feature impairment lies on either INFL or Comp (for German), which is not consistent with the lack of verb raising past negation we found among L2-English learners. Our tentative conclusion is that feature strength on an intermediate functional head (below Tense) in the grammar of the L2 learners is not set. This allows optional thematic verb raising past adverbs in L2-English. This intermediate functional head might be Agr in the split-INFL framework (see the structure in (6)). However, Agr underlies main verb inflection; on the other hand, we found optional raising of uninflected participles. Thus, the head with unset feature strength may be a lower head, for instance an aspectual head or Agro (see Belletti 1990 and Cinque 1999 for discussion of possible landing sites for participial movement in Romance languages). We refrain from further discussion of the nature of the intermediate functional head. Our conclusion naturally raises the question of why feature strength should be unset on an intermediate head, but set on Tense. That is, why do the L2-learners never place verbs before negation? One possible answer is that feature strength on Tense is in fact also initially unset, but that it is set to [weak] (for English) very quickly. This “quick setting” could come about for a variety of reasons. First, do-support may serve as a cue for feature strength setting, as suggested by Schwartz and Sprouse (1996). Additionally, all of the learners received classroom instruction in English, which emphasized use of do-support with negation. It is possible that placement of not in English serves as a relevant trigger for setting Tense feature strength. The learners hear not after auxiliaries but before thematic verbs, and may use this placement to set V features on Tense to [weak]. They will then not raise verbs to Tense, past negation. However, no such relevant trigger seems to exist for setting feature strength on lower functional heads. Feature strength on lower heads would remain variable – sometimes [weak] and sometimes [strong] – hence optional verb raising past adverbs. This hypothesis makes a prediction for L2-acquisition of languages with verb raising (e.g., French). In L2-French, as soon as feature
strength of Tense has been set to [strong], thematic verb raising should be obligatory past negation as well as past adverbs. Feature strength setting on lower heads will not be relevant, since verbs will need to raise all the way to Tense. 6. Conclusion and questions for further study Our results, coupled with previous work on verb raising in L2-English (e.g., White 1992, Eubank et al. 1997), suggest that optional thematic verb raising may be a property of L2-English regardless of the L1. It may be that L2-learners have trouble setting the feature strength parameter value for functional heads in the absence of explicit instruction and/or a specific trigger. There are some larger implications of these findings. For instance, in what other domains should we expect lack of feature strength setting and resulting optionality of movement? Would we find it for all verb movement (e.g., V2 in German, or inversion in questions)? For all head movement (e.g., N-movement)? Additionally, a few of the more advanced L2-learners did demonstrate knowledge of adverb-verb order in English. What causes eventual setting of the feature strength value? Is it a function of input? Why don’t all L2 learners set feature strength appropriately? Does age of acquisition play a role? These are all questions for further investigation. For the moment, we hope that we have added to the understanding of optional verb placement in the L2 by showing that it is not due to certain factors, such as heavy-NP shift or verb finiteness. Why optional verb raising does take place – why certain functional heads have unset feature strength – is a question requiring further study. Appendix A: Verb raising past different adverb types This section discusses possibilities of optional verb raising in Cinque's (1999) framework of multiple functional heads. Cinque proposed that the adverb hierarchy may reflect a hierarchy of functional heads. For instance, in English, epistemic adverbs precede frequency adverbs, which precede manner adverbs. This is illustrated in (17) and (18). (17)
a. b.
Mary always slowly eats her breakfast in the morning *Mary slowly always eats her breakfast in the morning
(18)
a. b.
Bill probably never sits still *Bill never probably sits still
This hierarchy of adverbs and functional heads may be relevant for our analysis of verb-raising in the L2, as follows. In Section 5, we proposed that the L2-learners may not be able to set the feature strength on a functional head below Tense. In Cinque’s framework, there are multiple functional heads below Tense. We therefore hypothesize that thematic verbs in the L2 should be more likely to raise past lower (manner) adverbs than past higher (frequency/epistemic) adverbs. The rationale for this hypothesis is as follows. Suppose, following Cinque, that multiple functional heads exist, each modified by an adverb. This is illustrated in (19). (19)
Possible structure with multiple functional heads IP DP INFL
FP1
(epistemic adverb) F1 FP2 (frequency adverb) F2 VP (manner adverb) V
DP
For simplicity, we assume that a manner adverb attaches to the left of the VP, while frequency and epistemic adverbs attach to higher functional heads, which we will call F1 and F2. Following our proposal in Section 5, both F1 and F2 have unset feature strength in L2 acquisition. Let’s say then that a thematic verb can optionally raise out of the VP either to F1 or to F2 (but not all the way up to INFL, which is set to [weak]). If the verb raises to F2, it will follow epistemic and frequency adverbs but precede manner adverbs. If it raises to F1, it will follow epistemic adverbs but precede both frequency and manner adverbs. The verb should never precede epistemic adverbs (unless, of course, there is another functional head between INFL and F1). Thus, thematic verbs should be quite likely to precede manner adverbs, less likely to precede frequency adverbs, and quite unlikely to precede epistemic adverbs.
We tested this hypothesis using items from the grammaticality judgment task. We have already discussed the eight items with manner adverbs (only items with full NP complements are included in the analysis below). Additionally, the task contained eight items with frequency adverbs (always, often, never ) – four with finite verbs and four with participles (20). There were also eight items with epistemic adverbs (probably, definitely, certainly ) – four with finite verbs and four with participles (21). (20)
a. b.
I never eat beans I eat never beans
(21)
a.
My daughter is definitely eating dinner right now My daughter is eating definitely dinner right now
As these results show, there is more optional verb-raising past lower (manner) adverbs than past frequency or epistemic adverbs, supporting our hypothesis. However, there is no significant difference between raising past frequency vs. past epistemic adverbs, contrary to our expectations. It is possible, however, that the learners are treating epistemic verbs as parentheticals, as illustrated in (24). As Cinque (1999, p. 45) also notes “another potential source of unexpected orders is the ‘parenthetical’ usage of (typically “higher”) adverbs; something for which I have no interesting account to propose.” (24)
b.
The results (% correct) are given in (22). The statistical comparisons of performance on the three relevant items types are given in (23). (22)
Raising past different adverb types: average % correct (adverb-verb order)
appropriate verb placement with respect to…
finite verb
manner adverbs
56.0%
56.0%
frequency adverbs
87.9%
76.7%
epistemic adverbs
75.9%
75.0%
negation
99.1%
99.1%
(23)
participle
Comparisons in performance on items with different adverb types: t-tests (paired two-sample for means). N=29.
comparison
average % correct
t-statisti c
manner vs. frequency adverbs manner vs. epistemic adverbs frequency vs. epistemic adverbs
56.0% vs. 82.3% 56.0% vs. 75.4% 82.3% vs. 75.4%
-4.95
-4.23
1.41
p (two-taile d) p < .0001
p < .001
p = .17
a. b.
John ate (probably) soup I am selling (definitely) my car
While the sentences in (24) are marginal, at best, for native English speakers, the equivalent sentences in Russian are quite grammatical, especially in the case of “probably”. (25)
a. b.
Ivan est (navernoe) sup John eats probably soup ?ja prodam (objazatel’no) I sell-FUT definitely moju mashinu my car
Thus, parenthetical postverbal usage of higher adverbs may be obscuring the adverb/functional head hierarchy. The difference in thematic verb raising past frequency vs. manner adverbs, on the other hand, suggests that the L2 learners are sensitive to this hierarchy. Additional control items on the grammaticality preference task suggested the learners are aware that manner adverbs typically follow frequency adverbs in English. Appendix B: Learner proficiency and optional verb raising (N=35) correlation between placement test scores and… scores on items with manner adverbs (+full NP complement) (e.g., (8), (9)) scores on items with manner adverbs (+pronominal complement) (e.g., (13)) scores on items with frequency adverbs (e.g, (20)) scores on items with epistemic adverbs (e.g., (21)) scores on items with negation (e.g., (10), (11))
correlation r = .40 (p < .05) r = .60 (p < .001) r = .42 (p < .05) r = .62 (p < .001) r = .38 (p < .05)
Thanks to (in alphabetical order by first name) David Pesetsky, Ora Matushansky, Steve Pinker and Suzanne Flynn for very helpful discussion and suggestions. i This decision was motivated by the need to ensure a minimal level English knowledge among the participants. ii An additional item type was used (following White, 1990/91, 1992) to ensure that L2-learners understood properties of do-support with negation in English. That is, we wanted to be certain that the learners were not simply using a strategy of "always use a form of do with negation, regardless of the following verb." To control for this possibility, we included four items of the form in (i):
a) Laura is not eating an apple b) Laura does not eating an apple The learners were, on average, 86% correct on this item type. Only six out of the 29 learners failed this do-support check (i.e., chose the correct answer in fewer than three of the four contexts); excluding these six learners from the other analyses does not change any results reported here significantly. We therefore analyze data from all 29 learners together. iii Thanks to David Pesetsky for suggesting this test. iv Other items on the grammaticality judgment task included control items with full NPs and various distractors. See Ionin (2001) for full discussion of all item types.
References
department of Brain and Cognitive Science, MIT. Available on www.mit.edu/~tionin Pollock, Jean-Yves. 1989. Verb movement, universal grammar, and the structure of IP. Linguistic Inquiry, 20, 365-424. Schwartz, Bonnie D. and Rex A. Sprouse. 1994. Word order and nominative case in non-native language acquisition: A longitudinal study of (L1 Turkish) German interlanguage. In Language Acquisition Studies in Generative Grammar: Papers in Honor of Kenneth Wexler from the 1991 GLOW Workshops, ed. Teun Hoekstra and Bonnie D. Schwartz, 317-368. Amsterdam: John Benjamins. Schwartz, Bonnie D. and Rex A. Sprouse. 1996. L2 cognitive states and the Full Transfer/Full Access model. Second Language Research 12, 40-72. Vainikka, Anne and Martha Young-Scholten. 1994. Direct access to X’-Theory: Evidence from Korean and Turkish adults learning German. In Language Acquisition Studies in Generative Grammar: Papers in Honor of Kenneth Wexler from the 1991 GLOW Workshops, ed. Teun Hoekstra and Bonnie D. Schwartz, 265-316. Amsterdam: John Benjamins. Vainikka, Anne and Martha Young-Scholten. 1996. Gradual development of L2 phrase structure. Second Language Research 12, 7-39. White, Lydia. 1990/91. The verb-movement parameter in second language acquisition. Language Acquisition 1, 337-360. White, Lydia. 1992. Long and short movement in second language acquisition. Canadian Journal of Linguistics/Revue Canadienne de Linguistique 37, 273-286. Yuan, Boping. 2000. Is thematic verb raising inevitable in the acquisition of a nonnative language? In C. Howell, S. Fish & T. Keith-Lucas (eds.), Proceedings of the 24th BUCLD, 797-807. Somerville, MA: Cascadilla Press. Contact information: Tania Ionin Ken Wexler NE20-415 NE20-409 77 Massachusetts Ave. Cambridge, MA 02139 www.mit.edu/~tionin
Bailyn, John. 1995. A configurational approach to Russian 'free' word order. Doctoral dissertation, Cornell University. Beck, Maria-Luise. 1996. Verb raising in English-German Interlanguage: First you see it, then you don’t. Paper presented at the 17th Second Language Research Forum, Tucson, AZ. Beck, Maria-Luise. 1998. L2 acquisition and obligatory head movement: English-speaking learners of German and the Local Impairment Hypothesis. Studies in Second Language Acquisition 20, 311-348. Belletti, Adriana. 1990. Generalized Verb Movement. Torino, Italy: Rosenberg & Sellier. Chomsky, Noam. 1993. A minimalist program for linguistic theory. In The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, ed. Kenneth Hale and Samuel Jay Keyser, 1-52. Cambridge, MA: MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Cinque, Guglielmo. 1999. Adverbs and functional heads: a cross-linguistic perspective. New York: Oxford University Press. Eubank, Lynn. 1993/1994. On the transfer of parametric values in L2 development. Language Acquisition 3, 183-208. Eubank, Lynn. 1996. Negation in early German-English interlanguage: more Valueless Features in the L2 initial state. Second Language Research 12, 73-106. Eubank, Lynn, Janine Bischof, April Huffstutler, Patricia Leek, and Clint West. 1997. “Tom eats slowly cooked eggs”: Thematic-verb raising in L2 knowledge. Language Acquisition 6, 171-199. Eubank, Lynn and Sabine Grace. 1996. Where’s the mature language? Where’s the native language? In A. Stringfellow, D. Cahana-Amitay, E. Hughes, and A. Zukowski (eds.) Proceedings of the 20th BUCLD, 189-200. Somerville, MA: Cascadilla Press. Ionin, Tania. 2001. Optional verb raising in L2 acquisition: Evidence from adult L1-Russian learners of English. Second-year paper for the
i.
