Expletives and Unaccusative Predicates in L2A
开篇：润墨网以专业的文秘视角，为您筛选了一篇Expletives and Unaccusative Predicates in L2A范文，如需获取更多写作素材，在线客服老师一对一协助。欢迎您的阅读与分享！
Received 23 March 2012; accepted 10 April 2012
Abstract
Common errors are examined and compared in the use of subject expletive it and the Spanish pronoun se with unaccusative predicates in the written production of advanced Spanish and Americanstudents of English and Spanish as a Foreign Language, respectively. They have in common not to respond to L1 surface transfer in the sense of not corresponding to a mere L2 relexification of L1 syntax. They rather involve a process of construction of the L2 grammar which results in similar structures in either interlanguage, different from both the L1 and the L2. The interlanguage of both groups, as reflected, in respective production data, is compared with respect to these structures. The hypothesis is proposed and developed that they originate in the interaction of Universal Grammar principles as well as both L1 and L2 influence in a restructuring process of the L1.
Key words: Expletive pronoun; Unaccusative predicates; English/Spanish as a foreign language; Interlanguage; Universal Grammar; L1 transfer
Marciano Escutia Lopez (2012). expletives and unaccusative predicates in l2a. Higher Education of Social Science, 2(3), -0. Available from: URL: /index.php/hess/article/view/j.hess.1927024020120203.3022
DOI: /10.3968/j.hess.1927024020120203.3022
Introduction
Escutia (2008, 2010) addresses how English as a foreign language (EFL) and Spanish as a foreign language (SFL) advanced students seem to construct their foreign language (L2) grammar, as seen in consistent written production data, with respect to so-called unaccusative predicates. These are intransitive predicates which favour subjectverb inversion (both in Spanish and in English) and (overt) expletive anticipation of the notional post-verbal subject (in English). The data provided there are very interesting in that one can see clear parallel structures being produced in both L2s.
In this paper we undertake a qualitative study by comparing some of those data referred to above, trying to show that both types of advanced learners are giving in their L2 grammars the same value to expletive or pleonastic it and the Spanish pronoun se, respectively. We will examine pairs of parallel sentences, with similar unaccusative predicates, used by both groups and attempt to provide an explanation for their similar underlying syntax. As part of it, the syntactic feature values of both linguistic elements will also be examined, as revealed in the same written production.
We will proceed as follows: first, we will put forward our assumptions about certain aspects of how the learners’ transitional competence in the L2 (interlanguage, IL) is formed; secondly, the subjects and the IL written data will be introduced and the divergence between native and non-native language (NL vs. NNL) production explicated; thirdly, a linguistic analysis will be presented of the type of predicates under study as well as the syntactic features of the pronouns non-natively used; next, the presence of both inter-lingual and intra-lingual factors will be shown to characterize the formation of these students’ IL. Finally, conclusions will be drawn regarding the non-native acquisition of these items.
1. THE FORMATION OF INTERLANGUAGE COMPETENCE
In the last thirty years, there have been several important attempts at formulating a coherent theory of L2 acquisition (L2A) compounding the two processes that have traditionally been distinguished as underlying it: on the one hand, native language (L1) influence, or transfer, and, on the other, the learner’s own creative linguistic power. Some of them which will be mentioned as we proceed- have taken as reference framework the chomskyan theory of Universal Grammar (UG, or innate constraints in the development of natural grammars), first within the Principles and Parameters framework (Chomsky, 1981) and more recently within the Minimalist Program (Chomsky, 1995, 2000, 2001a, 2001b), which would make both components, transfer and construction, compatible and derivable from the same linguistic theory.
The Principles and Parameters approach postulated that UG included universal linguistic principles (e.g. structure dependence, projection principle or X-bar theory, binding, etc.) and built-in options called parameters with binary settings allowing for crosslinguistic variation (e. g., head first or last with respect to the different types of phrases in a language). Parameter options determined a cluster of derived syntactic properties that a native speaker would display and learners would theoretically learn once they had set correctly the parametric option of the L1 or L2, respectively. The Minimalist Program associates those parametric differences between grammars and their consequent syntactic effects with the feature values of functional categories, such as complementizers, agreement, negation, determiner, etc., all of them part of the UG linguistic inventory. As White (2003) explains, “the lexicons of different languages vary as to which functional categories and features are instantiated and what the strength of the various features may be (p.10) ”.
Here the position is taken that L2students project those feature values present in their L1s as an initial hypothesis in their handling of the L2, that is, what has been called the Full Transfer/Full Access Hypothesis (Schwartz & Sprouse, 1994, 1996). Within this hypothesis, convergence on the relevant L2 properties may occur through restructuring of the L1 grammar in response to L2 input. However, this is not guaranteed and divergent outcomes are possible though all of them supposedly constrained by UG. In addition, we agree with other researchers (Liceras et al., 1997; Smith & Tsimpli, 1995) that L2 functional feature values are not activated as in L1A and restructuring of the L1 grammar takes place in a piece-meal fashion, trying to accommodate the different surface structures or manifestations which in the L2 are the syntactic consequence of those values.
More recently, it has been pointed out that L2A does not necessarily involve native production of certain linguistic forms, as the latter might depend on processes at the interface between the syntax and the linguistic production mechanisms. At this level, there may be conflict with the L1, which comes in to lighten up linguistic processing (Lardiere, 2000, 2008; Lozano & Mendikoetxea, 2010; Sorace & Serratrice, 2009).
With this view in mind, which ascribes formal deficits either to the non-native acquisition of syntactic features of functional categories or to, we will examine the type of written error present both in EFL and SFL advancedlearners to try to characterize their IL competence and to determine what the role of L1 influence may be in either case. Although the origin of those deficits might indeed derive from processing difficulties at the level of the interface between the syntax and its corresponding production mechanisms, we will not pursue that line since we are not analyzing the context in which these production data took place.
These are certainly performance data obtained from the students’ academic writing. Although extrapolating to grammatical competence may be perceived as a big jump, there is no doubt that they need to rely on their grammatical competence to write in their L2. The frequency and systematicity of their IL forms1 (errors from theL1 viewpoint) may be evidence of the existence of some stable underlying grammatical specifications or, perhaps, a processing deficit at the level of the interface between syntax and the linguistic production mechanisms.
2. PARTICIPANTS AND DATA
The data to be examined contain a common type of error found in both the written and oral production of advanced Spanish EFL and SFLstudents. It consists of the insertion of the non-standard preverbal pronouns it or se, respectively, and post-verbal placing of the non-clausal subject in sentences with so-called unaccusative verbs, including ergative, passive and be+adj predicates. The items exemplifying it and used here for comparison correspond, respectively, to EFL and SFL written sentences.
The seven EFL items have been extracted from the homework compositions of seven high intermediate students, as measured by their having passed the Cambridge First Certificate in English examination the previous year. They were all second year Spanish university students majoring in English at the Universidad Complutense of Madrid, Spain. As this is a qualitative study, only one sentence has been chosen from each participant, each item except for one, as we will see- reflecting the error under study with a different kind of predicate. They represent the whole gamut of possibilities for this type of error within the students’ group, as can be observed by looking up the other eighteen examples produced by their classmates and given in Appendix I. The students selected ―sixteen altogether― were enrolled in a four-month Composition course where, apart from other homework exercices, they had to present three 500 word compositions each, out of which all twenty-five items have been taken. They were the ones who came to class regularly and completed all the assignments.
The seven SFL items have been taken from a total of forty-nine compositions of an average of 650 words each belonging to seven postgraduate English speaking Spanish majors, enrolled in a SFL master’s program of the University of Southern Mississippi in their Madrid campus. The entrance requirements of such studies were the following: “Non-native applicants to the French and Spanish emphases should have an undergraduate degree in the language or present evidence of equivalent language experience”.
These conditions allow one to suppose safely that their proficiency level was, at least, advanced or high intermediate. In fact, they were able to follow the classes in Spanish, understand the content of the movies shown in the Spanish original version and express themselves fluently both orally and in writing. Five of them had already had some experience in teaching Spanish in the USA, three of which for several years. Although none of them was strictly bilingual, they had all spent more or less long stays in Spanish speaking countries. The compositions they had to hand in were part of the written activities about seven films, one of the requirements of a class called “Una Aproximación a la Problemática Social Espa?ola a través del Cine Espa?ol Actual” (‘An approximation to Spanish society through its present-day cinema’), of the 2004 summer intensive master.
The (b) items presented below correspond to five of the seven SFL students except for item (5b), which belongs to an SFL high intermediate student of the Written SFL Corpus project (Corpus Escrito del Espa?ol, CEDEL2: Cf. Lozano 2009), and has been chosen for lack of a corresponding SFL item in our own data to make our comparison between EFL and SFL items as complete as possible. The other fifteen sentences of the same type produced by our students can be consulted in appendix II. Besides, and in order to show these are not anecdotal data illustrate a very common error, as do the EFL ones, thirty sentences of the same type are also provided in appendices III and IV: on the one hand, fifteen produced by advanced SFL students of the Spanish Written Corpus or Corpus Escrito del Espa?ol (CEDEL2: Cf. Lozano 2009), cited on the reference bibliography at the end of this paper; and, on the other, another fifteen belonging to a public blog promoted by Dickinson College (in the USA) with undergraduate students’ comments and compositions in L2 Spanish along with materials from other subjects. In this latter case, although we do not have information about their proficiency levels, one can tell that the learners are at least of a high intermediate level, both because of the topics dealt with, as well as the length and structural complexity of the sentences they use.
There is one more unacceptable item for the SFL students, (7b), in order to show a sentence where preverbal se occurs with a finite clausal post-verbal subject, because the corresponding sentence for the EFL ones would be grammatical, as we show in the corresponding item (7a), a type they regularly produce correctly. For each EFL and SFL parallel example (in 1a-7a and 1b-7b, respectively), both the corresponding standard English and Spanish version are given below, as well as the English translation for the SFL items.
As can be seen, in each case a non-standard expletive, or semantically empty preverbal pronoun, has been inserted, it, for 1a-6a (7a would be acceptable), and se, for 1b-7b (even for 2b, as we will see). Standard English would either place the semantic subject pre-verbally, if the subject is not too long (as can be seen in all the glosses for examples (1a) to (6a)), or insert expletive there, which is theoretically possible in all items except 6a2 (see Quirk et al., 18.45, 49: 1404-1409). On the other hand, standard present day Spanish has no such overt expletives and it is really doubtful that our subjects may have been significantly exposed to those varieties which do have them. In the case of 2b, cuando se empieza la película, it is acceptable with a passive sense, meaning “when the movie is started” (in the sense of starting to shoot it or, probably more marginally, to screen it), clearly not intended by its author, who was just trying to situate her plot summary at the beginning of the film, not of its projection. The SFL items, as well as their standard Spanish glosses provided, have been checked and agreed upon with five native speakers of different varieties (one peninsular Spaniard, one Mexican, one Guatemalan, one Peruvian and a native from the Canary islands).
English expletive it can only anticipate post-verbal clausal subjects which have neither person nor number (functional) φ-features ―because it carries both person (3rd) and number (singular) to match the finite features of the corresponding Tense Phrase (TP). Pleonastic there, on the other hand, only carries (3rd) person features, leaving the number feature in T(ense) unvalued, rendering ungrammatical a sentence containing such expletive (e.g.: It/*there is impossible that she was at home). It can anticipate indefinite determiner phrases (DPs), as seen in examples (1a) to (5a), which have both person and number features to value the corresponding ones in T. There can be placed pre-verbally by merge3, the operation of assembling the sentence, in order to satisfy the EPP4 feature of T, which makes sure that all finite English sentences have overt pre-verbal subjects occupying the specifier position of TP (thus fully projecting T into TP).
Similar data to those in (1a)-(6a) have already been attested by researchers, such as Zobl (1989), Rutherford (1989) and Oshita (2000, 2004), who also provide written production from―among others―Spanish L1 intermediate students. Other authors have studied the English and Japanese L2 acquisition of unacussative predicates (see Hirakawa, 1995, 2001; Oshita, 2001; Sorace & Shomura, 2001; Oshita, 2004). Others have also dealt with them in studying the L2 acquisition of Romance languages (see Sorace, 1993, 2000; Toth, 2000; Montrul, 2004; Zyzik, 2006; Montrul, 2005; Lozano, 2006; Lozano & Mendikoetxea, 2008, 2010), in all cases, from a broader perspective than is taken here and not focussing on a particular type of error. Here, we do concentrate on a particular one because we think that not all Spanish L2 errors with pre-verbal se stem from the same source but we concur with other authors that their acquisition is not systematic (e.g. Escutia, 1998-99; Zyzik, 2006).
3. UNACCUSATIVITY IN ENGLISH AND SPANISH
According to modern linguistic theory, verbs such as, for instance, begin and its Spanish translation empezar, in examples (2a) and (2b), respectively, are intransitive ergative verbs, whose sole argument bears the (semantic) θ-role of theme (that is, the thing affected by or undergoing the action of the verb, with an understood and undefined external agent) and form part of a larger class of predicates that also includes passives (such as (5a) and (5b)) and raising predicates (like seem, happen, exist and be+(certain raising) adjectives (such as, likely or certain), as illustrated in (3a) - (3b), (4a) - (4b), (6a) - (6b) and (7a)-(7b), respectively). The property they all share has been termed unaccusativity. As pointed out by Levin & Rappaport-Horav (1995), it is a syntactically codified semantic property of certain predicates. From a syntactic viewpoint, unaccusative verbs do not form a homogenous class and are monadic predicates with a basic internal argument appearing, in English, in an external (subject, nominative) position because they cannot assign accusative case to its internal argument (Perlmutter, 1978; Burzio, 1986). With respect to their lexico-semantic composition, many unaccusatives express events where an object experiences an internal change related either to its state (e. g. become and disappear, Sp. hacerse and desaparecer, as in (1a) and (1b)) or its position (change of location, e.g. arrive, Sp. llegar). Other unaccusatives are inherently so, like the presentational/existential happen (Sp. ocurrir), exist (Sp. existir) and be (Sp. ser) in (3) to (6). All of the above are monadic predicates lacking an external argument and, consequently, no corresponding transitive counterpart.
Predicates with be are also found here alongside other predicates which signal change of state, because it is the linking verb in passive sentences, whose thematic structure is the same as that of unaccusatives in terms of theta role assignment and also makes it possible for the affected argument of predicative adjectives to raise to subject position. Furthermore, in Old English and even up until 19th century literary English- unaccusatives were conjugated with the verb to be as an auxiliary rather than with have, as in present day English5. This is also the auxiliary verb in modern German, Italian or French, although with certain differences between them (see Sorace, 1993, 2000).
Since unaccusatives have no underlying agent argument, they cannot assign accusative case to their sole argument; instead, they value it as nominative via its agreement relationship in φ-features with the finite T probe which structurally dominates it (c-commands, specifically). Subsequently, the EPP feature of T triggers its A(rgument)-movement in order to become the structural subject of T (see Radford, 2009). All this implies that―with respect to the example sentences―the following clauses―reflecting the type of errors that already both Zobl and Rutherford reported for their Spanish L1 subjects―are ungrammatical: *… and then (was) opened the door, *Happened many things, *Exist many people…, *When begins the second night..., *Was eaten a lot of food, *Is possible human cloning, since the theme argument has to raise to subject position in order to value and delete the EPP feature of T.
The underlying post-verbal position of the theme argument can be seen in their co-occurrence in constructions with existential there, mainly affecting unacussatives of appearance and existence: There happened many things, There exist many people …, There was eaten a lot of food at the party. It is also evident in their direct object position in those cases with transitive counterparts (ergatives), like open (She opened the door) or begin (The teacher began the lesson) and in their corresponding active counterpart in the case of passives (The guests ate a lot of food at the party). Thus, two possibilities are available in standard English for examples (1a), (3a), (4a) and (5a): either the argument moves to the subject position (spec-TP) or the expletive pronoun there (a syntactic filler pronoun without a theta-role) is merged there in order to delete the EPP feature of T (or moves there, for those who maintain the base generation of non-thematic expletives in spec-VP). This is so because, in order to be interpretable and grammatical in sentences, DPs must have, respectively, both a theta-role (assigned by the corresponding predicate in base position) and abstract Case. For item (6a), an example of the predicate be+adj, the only viable possibility is the raising of the argument to subject position, as T c-commands a definite matching goal (human cloning=the cloning of humans).
As shown in the examples, the SFL learners tend to produce the unaccusative predicates with the preverbal morpheme se. This element in the L1 is a pre-verbal clitic both syntactically (a type of inflectional morpheme though written separately) and phonologically (belonging to the accentual scope of the following word), and appears enclitically in the infinitival form. Let us review briefly now the standard uses of this pronoun following Whitley’s (2002) contrastive classification but using our own comments and examples in order to see the potentially confusing data the SFL learners are exposed to.
This author distinguishes the following types: (1) reflexive or reciprocal se, with paradigmatic variants according to grammatical person, which is the realization of one of the arguments selected by the predicate and is semantically transparent (e. g. Pedro se lava (a sí mismo)) ‘Pedro washes himself’; Antonio y Raquel se abrazaron ‘Antonio and Raquel hugged each other’); (2) passive se (p. ej.: Se abren las puertas a las 7 ‘The doors are opened at 7’), which “absorbs” the agent argument of transitive verbs in the construction called “passive reflexive”, a kind of middle passive, where the verb agrees in number and person with the syntactic realization of the patient or theme argument; (3) impersonal se (e.g.: Se come/estudia/corre muy bien aquí ‘One eats/studies/runs very well here’), which, both with intransitive and pseudo-transitive verbs (no or implicit internal argument, respectively), absorbs the agent argument just like one in the corresponding English gloss- and which we add, and recommend to explain to L2 learners in order to pedagogically optimize the distinction with (2)- occurs with singular verbal forms, as with plural ones the corresponding construction could be equivalent to the preceding reflexive passive (e.g.: Aquí se escuchan canciones modernas ‘Here one listens to modern songs’/ ‘Modern songs are listened to here’); (4) inherent se (necessary morpheme agreeing with the 3rd person subject of the so-called Pronominal Verbs: Pedro se atreve a hacer cualquier cosa ‘Pedro dares to do anything’), without parallel in English; (5) aspectual se (necessary pro-clitic morpheme agreeing with the subject and signaling telicity or accomplishment: Se comieron todo lo que les pusimos ‘They ate (up) everything we served them’), which does not correspond to any argument and conveys the same meaning as that of an intensifying particle in English, as marked in the English gloss; (6) dynamic se, which changes the meaning of the accompanying verb adding different nuances, such as inchoative aspect or unexpectedness (e. g.: Entonces Pedro se fue a casa ‘Then Pedro left for home’, different from Pedro fue a casa ‘Pedro went home); and, finally, (7) intransitive or decausative se, which in absorbing the agent argument turns a predicate from transitive to intransitive, thus there being alternation between the transitive and intransitive variants of such a verb: La puerta se abrió de repente ‘The door opened suddenly’ (ergative construction) / El viento abrió la puerta de repente ‘The wind opened the door suddenly’(transitive). The last two also have paradigmatic variants according to grammatical person.
It is passive and decausative se that relate to the unaccusative characterization. Mendikoetxea (1999) explains that, in Spanish, besides distinguishing unaccusative verbs, we should also refer to “unaccusative constructions”, which seem to derive from the corresponding transitive ones, such as those with se and passive (or periphrastic, with be) constructions, as illustrated, respectively, in the examples: las puertas se cerraron ‘the doors closed’ (ergative, unacussative) and las puertas han sido cerradas ‘the doors have been closed’ (passive). The unaccusative construction could be defined as one where the notional object (theme or patient) is realized syntactically as a subject, an example of which is the passive, even when it is formed with verbs with no unaccusative use (e.g. construir: El puente ha sido construido ‘the bridge has been built’ vs *El puente se construyó (él solo) ‘*the bridge built’ (meaning “by itself”). She carefully distinguishes unaccusative sentences with se from se passives. They are formally identical: for example, La puertas se cerraron is ambiguous between ‘the doors closed’ (unaccusative) and ‘the doors were closed’ (by a non-mentioned external cause, passive).
As shown above, the preverbal morpheme se has several and diverse functions which, as pointed out elsewhere (cf. Escutia, 1992, 1998-99, 2005), further complicate its acquisition by SFL learners, since it is inconsistent with certain facilitating linguistic processing principles, which seem to be active in language acquisition. In particular, our morpheme is inconsistent with the One-to-One and the Semantic Transparency principles of linguistic items, the former about the bijective correspondence between their form and function, and the latter about their clear and unchanging meaning (cf. Slobin, 1985; Andersen, 1983, 1990). Apart from these two, our clitic is inconsistent with other processing principles which need not be mentioned here (see Escutia, 1998-1999).
With this in mind and looking at the above descriptions of its uses, it seems evident that the clearest function of this morpheme, when it does not correspond to an argument of the predicate (as with reflexives), is that of suppressing the agent argument. However, in the SFL items of our study, se is unnecessary since the accompanying predicates are unaccusative and do not need the clitic to “absorb” an agent argument, which they do not select. Thus, the se of these items does not correspond to any argument, as the semantic role theme, the only one assigned, belongs to the post-verbal subject in all cases. Therefore, this morpheme has no semantic role and must have some syntactic function unless it is just a lexical chunk. Advanced learners may know that se is present when suppressing the external argument and overgeneralize its use to all constructions with no external argument.
As Zyzik (2006) points out, it is unlikely for se in the case of advanced EFL learners to be part of an unanalyzed chunk of speech, but, rather, one would expect from them a productive use of the L2, that is, that they would have already extracted it from a lexical chunk in their IL and would not just overgeneralize it in production, particularly with transitive predicates6. In any case, if such “overgeneralization” could take place, unaccusative predicates might be likely candidates to be subject to it. They select only one argument generated post-verbally to which they assign the semantic role of theme, which could be anticipated syntactically by an expletive pronoun, as in their L1. This is what happens in the EFL of advanced speakers with very different L1s (Spanish and Japanese, for example, as mentioned above). Furthermore, it is hard to think of formulaic language applying in this case, since it is very unlikely that these learners may have encountered examples like se existe/ocurre/es posible, etc, because both the classroom input and the texts they have been exposed to tend to use standard Spanish, without localisms. Thus, the morpheme se of our learners’ examples must have some definite syntactic function. In this connection, a distinction will not be made between type and token in our discussion for either the EFL or SFL items because, in this case, we think all of them respond to the same underlying processes.
In Spanish, the internal argument theme of unaccusatives, whether ergative, passive or raising predicates, does not need to move pre-verbally to satisfy the EPP but can remain in their underlying post-verbal position because that is taken care of by the preverbal null subject pronoun pro present in null subject languages like Italian or Spanish (Rizzi, 1982, 1990) or a null affix, as other authors maintain (Kato, 1999). Thus, in the corresponding Spanish translations in 1a-6a to the non-native English items the DP theme remains in situ. However, in Spanish it is not possible to insert an overt expletive in preverbal position, which this language does not have7.
Whereas English finite morphology requires that the syntactic preverbal subject be realized overtly (even in cases like these, where it has no theta role), the differentiated person and number agreement inflections of Spanish licence null subjects. For some authors, agreement in Spanish is said to have the syntactic feature [+pronominal], which allows it to identify its null subjects, while English is characterised as [-pronominal], which does not (Alexiadou & Anagnostopoulou, 1998). Kato (1999) and Barbosa (1995, 1996, 2000) explain that “rich” agreement in nullsubject languages like Spanish is pronominal in the sense that, besides having person and number features, it has a nominal feature [N]. Because Agr in Spanish is [+N], it can erase the EPP feature in T by raising V+Agr to T. The agreement affix both erases the EPP and “identifies” an empty category in Spec-VP, which for Barbosa is a pro locally bound to the agreement affix and for Kato, the trace of the agreement affix itself, which initially merges as the external argument of the verb, like a clitic. Without further technicalities, once the agreement affix or pro is capable of erasing the EPP in T, movement of the subject to Spec-TP is superfluous and discarded by economy considerations. Therefore, lexical subject arguments in Spanish remain in situ, post-verbally, and “the cost” of raising it pre-verbally would only be justified by a corresponding change in the informational focus.
4. TRANSFER AND LINGUISTIC ARTICULATION OF L2 SE/IT UNACCUSATIVITY ERRORS
Our L2 students, both of EFL and SFL, as well as those of the studies mentioned above, (Zobl, 1989; Rutherford, 1989; Oshita, 2000; Hirakawa, 1995, 2001; Oshita, 2001; Sorace & Shomura, 2001; Oshita, 2004; Lozano & Mendikoetxea, 2008), seem to react to the common lexical property of unaccusativity; that is, L2 learners seem to pick up unconsciously on the fact that unaccusative predicates have undergoer subjects. For example, many of the errors Oshita (2000, 2004) found in the corpus he studied of L2 English unaccusatives production by different L1 speakers (Spanish-speaking ones included) corresponded to passive unaccusatives (e.g. *“My mother was died when I was just a baby”). This somehow points to the learners’ implicit knowledge of the patient role of the subject in these constructions. As this property is the result of the interaction of argument structure, Case assignment and satisfying the EPP, it corresponds to constants in the grammatical design of natural languages, or, in chomskyan terms, to UG aspects. This has also been supported by L1A findings of V(erb)-S(ubject) order for unaccusative verbs, even in English speaking children (cf. Deprez and Pierce, 19938).
Oshita (2000, 2004) found that L2 learners treat unaccusatives and unergatives differently in terms of always observing the S-V order with the latter while often allowing the V-S order with the former, up to high intermediate proficiency levels. This is taken by some linguists as an example of the poverty of the stimulus problem (Hirakawa, 2001; Montrul, 2005), that is, a sign that UG must be at work, since there is nothing in the input that might lead them to such different treatment. Consequently, unaccusative verbs are a good test case for implicit learning for the following reasons: they are found in a great variety of languages and behave in the same way, although they vary as to their syntactic reflexes; they are never taught explicitly and native speakers are unaware of this phenomenon in their L1. Nevertheless, in spite of all this, L2 learners acquire the distinction between these two types of intransitive verbs (Montrul, 2005). Both Sorace (1993) and Montrul (2005) even show that L2 students are sensitive to the semantic distinctions between its different verb subtypes. Mendikoetxea and Lozano (2008) confirm this with respect to word order in their corpus study of VS order in L2 English directed towards unveiling the contexts in which the learners produce it: while Spanish and Italian advanced learners still make V-S errors with unaccusatives, they hardly do so with other intransitive verbs.
Thus, our students represent the argument structure of unaccusative verbs as having an internal theme argument, which they map appropriately to a position within the VP. This representation could also derive from L1 transfer, rather than UG, in the case of the Spanish EFL learners, but not so in the case of the English SFL ones. The former’s problems in production relate to moving this argument to subject position in the syntax and inserting a preverbal non-standard expletive instead; the latter’s in frequently adding the preverbal morpheme se.
Our Spanish EFL students somehow know that the (preverbal) subject position has to be overtly filled (that is, the EPP has to be satisfied as in English) and so they provide a syntactic filler. The only divergence from L1 standard English speakers is that they use a different expletive, it instead of there, which, on the other hand, is also present in analogous cases in other natural English varieties, like Black Vernacular English (in examples such as it ain’t no heaven for you to go to: see Labov, 1969)9. It would seem, then, that in the EFL learners’ IL, expletive it carries a 3rd person and an unvalued number feature as, both in the examples under study as well as in those from Appendix I, it occurs with either singular or plural post-verbal subjects. Native there, on the other hand, only has a (3rd) person feature and no number feature.
On the other hand, the English speaking SFL learners fill the preverbal subject position overtly (satisfying thus the EPP as in their L1), probably seeing in the clitic se the best candidate to do so, using it as some kind of pleonastic pronoun. This is hardly surprising, as our clitic appears in the grammatically related passive and impersonal constructions, where it signals the suppression of the external argument. As Toth (2000) points out, SFL learners cling to the semantic similarity of se in reflexive passive (e.g. se ocultaron pruebas ‘evidence was concealed’), impersonal (e.g. se ocultó al testigo ‘theyimp hid the witness’) and middle constructions (se ocultan facilmente las mentiras ‘lies are easily concealed’), and we add- they might use it to anticipate the non-agentive post-verbal subject. They may just be doing the same thing as our advanced EFL learners or those of other studies (Zobl, 1989; Rutherford, 1989; Oshita, 2000), which provide examples, mainly for Spanish and Italian L1 speakers, whose most frequent errors belong to the type *it-V-DP. Compare our (3a) and (3b) items: in both cases, it and se seem to function as an expletive placed there to erase the EPP feature.
What we are proposing could correspond to a (simplified) representation of examples (3a), (3b), (5a) and (5b) as given below (where ? stands for a null complementizer carrying a declarative force feature, AFF for the verbal tense affix, and QP for Quantifier Phrase, a kind of DP whose head is a quantifier).
Both the EFL and SFL advanced learners seem to be fully aware of, respectively, the non-null and null subject nature of English and Spanish. The former provide overt expletives, which are non-existent in their L1, and these elements seem to be the most difficult type of subjects to be incorporated by L1 null-subject students. Unlike Oshita (2004), who treated ?/it-V-NP errors as alternative forms, these EFL learners’ production might suggest that it-V-NP replaces ?-V-NP at more advanced proficiency levels, though the data is probably not enough. At the same time, this is compatible with not having learned to raise the theme argument of unaccusative verbs to subject position in all cases, even when it is obligatory because the theme argument is definite and expletive there could not be inserted in English, as seen in examples (2a) and (6a).
On the other hand, the SFL advanced students, who do not tend to produce unnecessary preverbal pronominal subjects in their L2 compositions, appear to need the pronoun se with the post-verbal subjects of unaccusative predicates. It does not seem to correspond to pleonastic there but rather a generic expletive because it serves to anticipate both clauses, like it, and DPs, like there: see (1b) and (7b), for example (and all the other (b) items, of course).
The parallelism of the SFL and EFL items is clearly evident. In both cases, the non-native expletives used, se and it, have the 3rd person feature plus an unvalued one for number, as they occur both with singular and plural post-verbal subjects, while native it has 3rd person and singular features agreeing with the corresponding finite features of ST and anticipates a clausal post-verbal notional subject. As explained before, native expletive there has only the 3rd person feature and the number feature of the head T is left unvalued, thus making the corresponding clause ungrammatical: e.g. it/*there is impossible that she was at home. It does anticipate indeterminate DPs, having both person and number features which can value the corresponding ones in T: e.g. there is-are/came (a) car(s) in the yard. This makes it also different from the SFL learners’ non-native se, which can anticipate determinate DPs, as seen in (2b) and (4b). Evidently, it goes without saying that the EFL leaners have no problem using well anticipatory it with clauses, as shown in (7a), provided to show this.
All these learners seem to have grasped the presentational capacity of unaccusative predicates in the distribution of information10, with the post-verbal subject which tends to be heavy, as pointed out by Mendikoetxea and Lozano (2010), as in some of our examples- signaling new information, consistently with processing principles of the syntax/discourse interface. As these authors also conclude with respect to Spanish EFL students in general, both our EFL and SFL learners correctly produce post-verbal subjects with unaccusative predicates, but show consistent non-native errors in their syntactic coding. In their framework, the it/se-V-S word order shows that our learners acquire the corresponding syntactic knowledge but may signal processing difficulties in the integration of the syntactic and discursive information due mainly to L1 transfer.
These learners show difficulties with non-interpretable syntactic features, specifically, the SFL students with the [+pron/+N] of Agreement and the [+null] of Spanish expletives, and the EFL ones with the functional specifications of native it and the [-pron/-N] nature of Agreement. This results in choosing a non-native expletive instead of a null one, in the first case, and, in the second, in not raising the post-verbal DP subjects to pre-verbal position (not even short ones), but rather anticipating them non-natively with it. However, in either case, the non-native production is not a mere reflex of L1 transfer because both groups go beyond what we might expect if only transfer were at work: the EFL learners show their knowledge that the subject pre-verbal position must be filled in English (to satisfy the EPP) by providing an overt expletive where it is not necessary in their L1, Spanish; the SFL ones show their knowledge about keeping the subject post-verbally with unaccusative predicates in the L2 but, and here it is where L1 transfer may act, filling as well the subject pre-verbal position by turning the Spanish morpheme se into an expletive of sorts. The latter is not equivalent to their corresponding one there for this kind of contexts since their se has different syntactic features not present in either of the expletives of their L1. That is, neither group simply relexifies the L1’s surface structure in L2 words but rather create surface structures not present in their L1s.
The fact that both groups keep the notional subject post-verbally could be in the case of the Spanish EFL learners due to L1 transfer, but being, as it is, concomitant with an anticipatory non-null morpheme erasing the EPP feature, looks more like UG-derived knowledge as well. In the face of it, it is very interesting to see that both groups of students are really doing the same thing in producing it/se-V-S sequences since they all keep the same word order, the expletives can anticipate both finite clauses (native-like in the case of it) and DPs (non-native-like in either case) and the surface structure of these strings is not present in their respective L1s.
As pointed out in Escutia (2008), it appears that non-null subject and subject raising need not be related in L2A, as it used to be thought in the first L2A accounts of the Null Subject Parameter (White, 1985; Phinney, 1987, etc.), where verb-subject inversion used to be considered one of the derived properties of its positive setting (with test sentences normally involving unaccusative verbs). Here, satisfying the EPP and subject raising seem to be independent. Montrul (2004) points out, citing López-Ornat (1994), that the morphosyntactic aspects of the Null Subject Parameter and the unaccusative/unergative distinction are acquired in very early childhood (before four years of age) in monolingual Spanish children. In L1A, their syntactic consequences might well be linked and fall into place seamlessly. Probably, the situation is the same for English L1, butL2A seems to follow different paths, as has been assumed here. In fact all our L2 students seem to have the same type of difficulties in the acquisition of the functional features of expletives (including their (non) overt nature) while at the same type keeping subjects with unaccusative predicates post-verbally.
In sum, our learners seem to build their mental grammar in a UG-determined way since they show evidence of the involvement and interaction of the theta Criterion (each argument must be assigned a semantic role), the EPP, the thematic hierarchy and Case assignment. It is possible that in constructing their grammatical specifications for unaccusative structures they may start with their L1 knowledge, which would lead them, first, in the case of the EFL learners, to simply leave in place the internal argument of the unaccusative predicate in surface syntax, without either raising it to subject position or providing a subject expletive (as Zobl, 1989 and Rutherford, 1989 report for their less proficient learnes), because their L1 Spanish satisfies the EPP with a null expletive or affix; in the case of the SFL ones, L1 transfer might lead them to raise the notional subject pre-verbally to satisfy the EPP feature. Nevertheless, our learners have gone beyond the confines of their L1s: the EFL ones in also providing an expletive (although a non-standard one11) to satisfy the EPP as in the L2; and the SFL ones in keeping the subject post-verbally, as in the L2, anticipating it with a non-null expletive non-existent in the L2 but congruent with their L112.
Therefore, although there may be at work at the beginning either UG, in the sensitivity to unaccusatives’ underlying structure for both types of learners, or transfer, in knowing about it through the L1, the transitional grammar at this stage seems to have been restructured towards the full realization of the non-null subject nature of English and the lack of subject raising with unaccusatives in Spanish.
CONCLUSION
The data we have reviewed and compared brings home once more the realization that learning a foreign language goes far beyond simply relexifying or clothing the old L1 structures in L2 words. It reveals itself rather as a distinct process in which there seem to work together UG (in the learners’ sensitiveness to the unaccusative thematic structure) and both the L1 grammar (in not raising the subject pre-verbally for the EFL learners and providing a non-existent overt expletive se for the SFL ones) and the L2 grammar (in providing an overt expletive, non-native in its distribution, for the EFL learners; and keeping the subject post-verbally, for the SFL ones). The reference framework used here has allowed us to articulate this process linguistically and provide for it a common source resorting to the same underlying principles.
This process would be compatible with Liceras’ (1996) L2A model, according to which, L2learners restructure certain parts of their linguistic representations following the model of the L2’s surface structure because their learning procedures, as opposed to those of the L1A, have no access to the functional feature values of the L2: in this case to [± pronominal/N] of Agreement, which sanctions both subject raising and the presence of a (non) null expletive and to the functional features of the expletive itself. As stated above, we are also conscious that the data examined could also be explained with the model of Lozano & Mendikoetxea (2010), which ascribes this kind of deficit to the processing difficulties inherent to L2 production, causing competition from the L1 at the syntax-discourse interface but our main purpose was to show how the learners’ IL grammar for unaccusative predicates, as seen in their production, is different from their L1’s and not merely ascribable to transfer but to construction and restructuration of functional elements of the latter.