L 2 acquisition of Spanish dative clitics by English and Dutch learners

This article examines the second language acquisition (SLA) of Spanish dative clitics in clitic doubling (CLD) structures that are closely related to the double object construction (DOC) in English and Dutch. It also addresses the question of how adult English and Dutch speakers learning L2 Spanish in a formal setting develop knowledge and use of the animacy constraint in the target language, which is different from the first language (L1) counterparts. The role of transfer in acquiring new syntactic structures has been taken into account, where dative clitics appear and animate objects are marked by the dative preposition ‘to.’ New findings are obtained on CLD and the Spanish animacy constraint from a grammaticality judgement task (GJT), completed by English and Dutch learners at B1 and B2 CEFR levels. The difficulties learners experienced were not always due to negative L1 transfer, but also related to the complexity of the argument structure where the clitic is inserted. This has clear implications for the teaching of pronominal elements which are closely related to different syntactic configurations in Spanish.


Introduction
Dative alternation is a well-known phenomenon in second language acquisition (SLA).Recent studies, including Bresnan, Cueni, Nikitina, and Baayen (2007) for L2 English, and Jäschke and Plag (2016) for German-English interlanguage, have shown that the two alternative forms are not acquired at the same time.In particular, some factors, such as animacy and syntactic complexity, are key to successfully predicting which structure is chosen in a given context in English (En), either with a prepositional object, as in (1a), or without a preposition in the socalled double object construction (DOC), as depicted in (1b): (1) a. En.Mary gave the wonderful watch to her mother.
b. En.Mary gave her mother the wonderful watch.
Other languages like Dutch (and Spanish also allow dative alternation with a prepositional phrase structure and a DOC, similar to the English examples in (1).As illustrated by the examples in (2), Dutch (Du) allows dative alternation in similar semantic contexts forcing the prepositional dative to occur at the end of the sentence: 1 end of the sentence over double objects (Cuervo, 2007;Imaz Agirre, 2015;Perpiñan & Montrul, 2006,).Following Cuervo (2003), the Spanish DOC is an instance of dative clitic doubling (CLD).Note that the animate indirect object is next to the verb, and so is the dative clitic in the previous example in (3b).Yet a closer comparison reveals some mismatches between English and Dutch DOC, on the one hand, and Spanish CLD, on the other hand, as mentioned above.
Another irregularity deals with the so-called "animacy" effect, wherein the recipient (indirect object) must be animate in the DOC configuration in both English and Dutch, but not in Spanish CLD.Note that both recipients in the examples in (4) are next to the verb and are likewise introduced by the dative preposition 'to' regardless of whether they are animate or inanimate.Interestingly, the animacy constraint in Spanish needs to be formulated in different terms.In fact, only direct objects seem to be subject to animacy, which explains why animate direct objects must be preceded by the preposition a 'to', as illustrated by the minimal pair in (5): (4) a. Sp.Ana le dio a su estudiante el libro.
Ana CLDAT gave to her student the book 'Ana gave her student the book.' b.Sp.Juan le puso al coche gasolina.
Juan CLDAT put to the car the petrol 'Juan put petrol in the car.' (5) a. Sp.Ana visitó a sus padres.Ana visited to her parents 'Ana visited her parents.'b.Sp. *Ana visitó sus padres Ana visited her parents This article reports empirical data on the acquisition of Spanish dative clitics with CLD by speakers of English and Dutch, who share a similar dative alternation in their L1 but are faced with a different dative alternation in the foreign language.The study compares the L2 acquisition of Spanish CLD along with the L2 acquisition of objects marked by the dative preposition by two different groups of English and Dutch learners in foreign language settings, based on their performance in a comprehension task, for the first time to the best of our knowledge.The objective is to examine the development of dative clitics in their foreign interlanguage.
Analyzing the dative alternation in L2 Spanish is also key to L2 grammar teaching as it presents an interesting dilemma: deciding which structure needs to be taught first.And in this case, one assumes that SLA can be extended to explicit teaching and learning, which cannot change the natural order of acquisition.In order to determine the linguistic phenomena that can be critical, and hence relevant for the present study, this study focuses on the L2 acquisition of Spanish dative clitics with a special emphasis on CLD: When the recipient precedes the theme and as a consequence is doubled by a clitic, as illustrated by the previous examples in (3).
The paper is structured as follows.Section 2 mentions some previous studies on the acquisition of animate objects in Spanish as an L2; then section 3 presents an analysis of the syntax of dative clitics.Section 4 explains the method.Section 5 discusses the results.Finally, Section 6 deals with the interpretation of the findings and their implications for grammar teaching in the classroom of Spanish as a foreign language.

The acquisition of animate objects in L2 Spanish
Previous studies have investigated the L2 acquisition of Spanish animate objects marked by the dative preposition a 'to,' including after instruction or feedback on this particular structure (cf.Martoccio, 2012 and all references cited therein).In particular, Martoccio (2012) found a wide array of difficulties regarding the use of the dative preposition a in front of affected objects, replicating previous and current research studies such as Montrul and Bowles (2009), Farley and McCollam (2004), Guijarro-Fuentes and Marinis (2007) and Guijarro-Fuentes (2012), among others. Martoccio (2012) argues that difficult structures need to be explicitly taught in order to increase learners' awareness of them.In Martoccio's study, instructed groups improved significantly more than uninstructed groups when tested on those structures not found in their L1.
Within the generative tradition, transfer effects have been found in the case of L2 acquisition of Spanish CLD.Bruhn de Garavito's (2000) study presented data confirming that L1 English learners could not distinguish grammatical double objects from ungrammatical ones when the dative was not animate.In Cuervo's (2007) study, English learners performed at chance level when the dative argument was inanimate, rejecting possessor datives in these contexts.Interestingly, Sikorska's (2009) experimental data on the acquisition of Spanish CLD by L1 Polish learners also showed chance performance when the dative argument was not animate.This finding supports the claim that these learners misanalyzed the CLD construction and mistakenly considered it to be constrained by the animacy of the dative clitic, which is a negative L1 transfer effect.Unlike English DOC or its Polish counterpart, the Spanish CLD is not constrained by an animacy constraint on the dative.Rather, the animacy constraint is present with some direct objects, as discussed below.

The analysis of dative clitics in CLD
This study addresses the question of how CLD is acquired in L2 Spanish, bearing in mind that two different classes of datives are examined.When dative determiner phrases (DPs) go with verbs such as give or introduce, two core arguments are presented, and the dative is usually interpreted as goal.In addition, some dative DPs that are interpreted as benefactives in Romance languages are therefore interpreted as non-core datives since they typically go with other types of verbs traditionally considered transitives like visitar 'visit' or conocer 'meet.'Hence, they are analyzed as affected objects introduced by the Spanish dative preposition a.Thus, the experimental question to be raised is how core and noncore dative clitics with these two types of verbs are acquired in L2 Spanish.
The question of what grammatical elements need to be learnt in order to target a different distribution of verbal arguments cannot be investigated without making some further assumptions about the argument structure of the verb including all its participants: agent, theme and goal.The first assumption is that internal arguments (the direct object and the indirect object) are internal to the event semantics and help construct the event.The external argument (the subject and the adjunct) are event modifiers.Following a generative grammar approach, arguments are DPs and the subject is related to the event by being the specifier of the verbal head.Applicative constructions are analyzed likewise, allowing both internal and external arguments in similar positions next to an applicative head.
Following the argument structure above, the analysis of the English dative alternation as previously illustrated in (1) posits the idea that the DOC alternative in (1b) exhibits a functional projection responsible for the case assignment of the non-core internal argument her mother, namely an applicative head, as in Pylkkänen's (2008) analysis, as depicted in (6b) below.A similar analysis including an intermediate applicative head can also be extended to the L1 Dutch DOC since indirect objects (without any dative preposition) also precede direct objects, as illustrated by its Dutch counterpart in (2) above.
As for Spanish, according to Cuervo's (2003) analysis of benefactive constructions, we can assume that the dative clitic in examples such as those depicted in (4a) is inserted in the applicative head between the functional lowercase v, where the external subject is projected, and the V-domain, that is, the thematic environment for internal arguments, as depicted in (7). 2 The clitic will further incorporate into the verb as an instance of head movement from its canonical position in APPL.
In the literature on Romance languages, other analyses can be found which have accounted for the presence of clitics in ditransitive configurations with non-core arguments, such as in Torrego (2010) on the Spanish phenomenon of leismo.3Interestingly, Laughren and Eisenchlas's (2006) analysis claims that the same applicative head is responsible for affected objects, which are crucially definite and animated, and therefore preceded by the dative preposition a, as illustrated by the minimal pair in ( 5) above.
This account predicts that the affected or animated object preceded by the dative preposition a is in complementary distribution with a second dative phrase doubled by a clitic in ditransitive configurations with verbs like presentar 'introduce.'This prediction is well supported by the examples in (8) below.Note that the affected or animated object su jefe 'his boss' in the sentence depicted in (8a) is, however, in complementary distribution with the indirect object a su mujer 'to his wife' doubled by the dative clitic le in the sentence depicted in (8b) since it is now preceded by the dative preposition a.
Juan le presentó a su mujer su jefe.
Juan CLDAT introduced to his wife his boss 'Juan introduced his wife his boss.' b.
*Juan le presentó a su mujer a su jefe.
Juan introduced to his wife to his boss Extending Laughren and Eisenchlas's (2006) analysis to the grammatical counterpart in (8a), as in ( 9), the same applicative head is responsible for the presence of the dative preposition in front of animate objects as well as for the dative clitic heading the applicative head, as indicated by the tree in (9).Further evidence for this unitary analysis comes from the fact that the cluster of argumental clitics that includes an accusative clitic and a dative clitic (CLDAT + CLACC) bearing different indexes (i, j) are in complementary distribution, when referring to two animated objects, as illustrated by the minimal pair depicted in (10).Sei loj presentó a su mujeri.CLREFL CLACC introduced to his wife 'He introduced him to his wife.'A dative clitic cannot double an indirect object in the presence of an accusative clitic since it also doubles an affected object and competes for the same applicative head, which explains the ungrammaticality of the example in (10a).In contrast, the combination of the reflexive clitic se together with the accusative clitic lo is possible, as illustrated by the grammatical sentence in (10b).This possibility has been argued to constitute evidence for the presence of an additional applicative head for reflexive datives (Escobar & Teomiro, 2016;Teomiro, 2013).An analysis that includes a different applicative head for the clitic se in French can also be found in Boneh and Nash (2011).

Method
The main objective of this paper is to present new experimental data concerning the acquisition of Spanish dative clitics by English and Dutch adults learning Spanish as a foreign language in a formal context.The fact that these learners need to discriminate between two configurations where dative clitics appear, namely as clitic pronouns with ditranstive verbs and as doubling clitics in CLD will be discussed.

Grammaticality judgment tests
It is now generally accepted that SLA requires implicit and explicit learning.Following Ellis (2004Ellis ( , 2009)), one way to test metalinguistic knowledge is by using some reliable mechanisms like grammaticality judgment tests (GJT).In this way we could be able to obtain metalinguistic judgments referring to the two main questions of our study, namely: 1. Do participants in our study discriminate between the dative clitics that appear in each of the two syntactic configurations as discussed above? 2. Are they aware of the animacy constraint associated with animate objects marked by the dative case marker a in L2 Spanish?
In relation to these questions, two alternative hypotheses were considered.The first one is that L1 transfer could have a negative effect on the acquisition of apparently closely related syntactic structures.Or, on the other hand, implicit knowledge could complement and counterbalance that negative effect by acquiring the animacy constraint associated with the target syntactic structures.If the first hypothesis is correct, it would be expected that the Spanish CLD would always be a challenge for L1 English or L1 Dutch learners, provided that they start with a DOC configuration constrained by a different animacy effect on the dative next to the verb, as discussed above.On the other hand, if learners had already acquired both syntactic configurations, they would have implicit knowledge and would naturally select a number of features in the L2 in order to construct each syntactic structure (Chomsky, 1995;Lardiere, 2008;Van Patten, 2011), namely the features that conform to particular elements functioning as syntactic heads like APPL, to form the new syntactic configurations, as mentioned above.

Participants
Two groups of English and Dutch learners participated in the study.Participants were university students taking Spanish as a foreign language in the second or third year at two different universities: the University of Bristol (UK) and the University of Ghent (Belgium).All of them had studied Spanish in secondary school before attending college.As for their educational background at university, the learners had attended formal Spanish courses using a communicative approach for two years prior.
The first group consisted of a total of 69 English learners, with an age range of 22, enrolled in a Spanish course (B1+ level) that was divided into three different groups at the University of Bristol.The second group consisted of a total of 73 Dutch learners, with an age range of 24, enrolled in three different Spanish courses (B1 = 28, B1+ = 24 and B2 = 21) at the University of Ghent.All courses followed a communicative teaching approach and the grammar of Spanish dative clitics had not been explicitly taught to any of the groups.4

Procedure
Learners in each group were asked to freely participate in the study.They were given a comprehension test preceded by some instructions on how to complete it.The task was presented as an optional exercise to revise their grammar.Students had as much time as they needed.In order to motivate them, the students were offered a certificate of participation.The materials consisted of a grammaticality judgment test that contained five items as a placement test, followed by another 35 experimental grammar items.All items were distributed on the same answer sheet.
As for procedure, each participant had to complete the test on their own.In the case of grammatical sentences, the correct answer was "yes."And in the case of ungrammatical sentences, the correct answer was "no."All of these questions (test items and placement test items) had to be answered next to each sentence on the questionnaire, which was distributed amongst the two groups of participants (English and Dutch groups) in a formal setting in the classroom of Spanish as foreign language, with a time limit of 15 minutes.Both studies were approved by each corresponding institution and were performed in accordance with ethical standards.

Materials
In this section we describe the instrument designed to measure the sample items and the response options.The study implemented a comprehension task based on a GJT, where inanimate datives and animate/direct objects appeared together with a large number of other experimental items, which were employed as distractors.The test included 41 test sentences, eight of which were test items for the present study.The distribution of the test items consisted of four grammatical Spanish applicative dative constructions and four ungrammatical ones.The test items were presented with very different verbs and in different configurations listed in a counterbalanced order to impel learners to use their unconscious knowledge of the target language.Hence, the test contained grammatical sentences with a dative clitic, and ungrammatical sentences without either an obligatory dative clitic or the dative case marker a, as depicted in (11).Firstly, grammatical experimental sentences included CLD with a dative clitic after the verb in the imperative form, as in (11a); CLD with an animate dative phrase, as in (11b); CLD with an inanimate phrase doubled by a dative clitic, as in (11c); and, CLD in a ditransitive configuration with a reflexive dative, as in (11d).( 11 As for the ungrammatical sentences, different linguistic contexts for dative alternation were included.These contexts consisted of a number of CLD configurations without a dative clitic, as illustrated in the examples in ( 12): a fronted dative not doubled by a dative clitic, as in (12a); a ditransitive configuration with two animate objects without a dative clitic or a dative case marker, as in (12b); and a ditransitive configuration without a dative clitic or a dative case marker, as in (12c).( 12 In this sense, a grammatical sentence with the relevant morphology of the applicative structure was tested together with an ungrammatical sentence lacking the morphosyntactic dative marker (i.e., the dative clitic, the dative preposition or the reflexive dative clitic), among other distractor sentences.The task for all test items above, which appears in the appendix at the end of this article, consisted of indicating whether the sentences sounded natural or not in the learner's L2 Spanish.We believed that this was more informative in terms of their unconscious competence than having to say whether the sentences were grammatical or ungrammatical.

Placement test items
Results for the placement test appear in Table 1.In this case, the following items were tested: (a) the correct use of the Spanish verbs ser/estar 'be' with individual predicates (cf.Item 2), (b) the correct use of the auxiliary verb haber 'have' in a present perfect verb form (cf. Item 3) and (c) the correct use of morphological agreement for 1st person singular (cf.Item 7).Table 1 indicates the correct response means for the three relevant placement items represented by ungrammatical sentences in Spanish, assuming that rejection of ungrammatical items can be a measure of explicit knowledge (Gutiérrez, 2013).In particular, the placement items range from incorrect use of the Spanish verb ser (Item 2) to the Spanish auxiliary verb haber (Item 5) to verbal inflectional agreement of first person singular (Item 7).An independent-samples t test showed that, as far as language proficiency is concerned, there was a significant effect for L1, t(df) = 2.64, p = .009,with the Dutch speakers significantly outperforming the English speakers.

Experimental test items
Table 2 indicates the correct response means of the experimental conditions ranging from the grammatical (G) sentences including a dative clitic (cf.Items 3, 6, 14 and 20) to the ungrammatical (UG) sentences without an obligatory clitic (cf.Items 16,22 and 38).Likewise, in order to determine whether the difference between the experimental test conditions was significant between groups, a t-test analysis was conducted, contrasting the means of both of them.Table 3 below contains the resulting values per each experimental item.Considering these statistical values, the Dutch learners of Spanish performed better than English learners in both grammatical and ungrammatical sentences, especially regarding the CLD condition (Items 3 and 14), and the animate object condition (Items 22 and 38).Both groups, however, had similar difficulties with non-animate dative clitics (Item 6) and reflexive dative clitics (Items 16 and 20).Note that significant differences between both groups were obtained except for the latter conditions (Items 6,16 and 20).The fact that the Dutch students obtained better results than the English students on the placement test items above indicates that the former had a higher level of Spanish.Yet, there were no significant differences in terms of the experimental items regarding a complex argument structure, such as Item 6 (CLD with an inanimate dative), Item 16 (CLD with a reflexive dative clitic) or Item 20 (a ditransitive structure with a reflexive dative clitic).This result clearly suggests that both English and Dutch learners had difficulties with the CLD condition and with reflexive dative clitics in complex argument structures.Our results also pinpoint a similar tendency to reject the dative clitic in two-participant events deriving the DOC alternative in L2 Spanish, regardless of the learner's proficiency level.This indicates that there is a particular difficulty in acquiring clitics in complex argument structures, as has also been observed in the acquisition of other clitics with applicatives (Escobar & Teomiro, 2016).

Discussion
In this paper, we have examined the results obtained by two different groups of adult students learning Spanish as a foreign language, presumably at the same B2 level.Both groups shared a similar dative alternation configuration in each of their L1 languages: English and Dutch.Likewise, they had to learn a different syntactic structure in Spanish, their target language, in which a dative clitic doubles a dative phrase.In addition they also had to acquire an animacy constraint, nonexistent in their L1 languages, through the marking of animate or affected objects by the dative preposition a. Relevant data on the question of how dative clitics develop in foreign language settings were gathered for this study by means of the comprehension task.
First of all, it was found that, at lower levels of acquisition, the English learners incorrectly accepted the omission of the dative clitic.In contrast, the presence of dative clitics with CLD was generally accepted by the Dutch learners, who demonstrated a better level of Spanish according to our placement test results.However, this was not always the case since the Dutch learners also failed to reject the omission of dative clitics with CLD when the dative phrase was fronted (cf.Item 16).As for the second question, no systematic transfer from L1 was found.There was, however, an L1 effect: Both groups at all levels incorrectly rejected CLD with an inanimate dative phrase doubled by a dative clitic (cf.Item 6).Finally, the Dutch learners that exhibited a better level of Spanish on our placement test also obtained better results in terms of the Spanish animacy constraint.However, they also appeared to have some difficulty with the experimental conditions when some complex structures were at stake, namely Item 22, in which both animate direct object and indirect object compete for the dative preposition, and Item 20, which contains a reflexive dative in an apparent ditransitive configuration.
These results indicate that they are compatible with hypotheses that emphasize the natural development of syntactic structures in second or foreign languages.This then suggests that the acquisition of new complex configurations leads to different performances at all levels, although at higher levels it is expected to be completed.We would like to argue in favor of at least two developmental stages for the learning of dative clitics based on the syntax of the particular configurations that need to be acquired over time, as depicted in (13).Thus, salient patterns when teaching dative clitics according to the argument structure where they appear should be taken into account.First, transitive configurations should be presented where animate objects are preceded by dative preposition a; then dative clitics could be introduced as an image pattern of the English ditransitive construction with a prepositional dative pronoun at the end of the sentence.Finally, the intricate phenomena of dative clitic doubling regarding a complex argument structure should be explained.
Regarding the type of verbs to be taught, the easy transitive configuration should be introduced with verbs that take non-affected objects such as: pagar 'pay,' comprar 'buy,' vender 'sell' or alquilar 'rent,' among others, as in ( 14), since the accusative clitic (masculine lo or feminine la) can act as the verbal object in all these cases.The second pattern to be taught would deal with transitive verbs taking animate objects introduced by the dative preposition a such as: conocer 'meet,' visitar 'visit' and presentar 'introduce,' as in (15) As part of the second pattern, the Spanish verb escuchar 'listen' can also be taught as a transitive verb which goes with an animate object understood as affected in the examples provided in ( 16), where the dative clitic is also required.Significantly, it is important to note that the English verb also takes the dative preposition to, as illustrated in the translations of these examples.
¿Has escuchado al médico?'Have you listened to the doctor?' b.
Sí, ya le/(*lo) he escuchado.'Yes, I have already listened to him.' Once transitive configurations with the affected/non-affected object dichotomy are presented, animate objects co-occurring with datives in ditransitive configurations should be explained.This pattern turned out to be particularly challenging for both groups of learners in our study.Finally, dative reflexive clitics should be introduced as a complex argument structure.In sum, our analysis is based on two main tenets: rethinking the language knowledge base, and modifying language instruction in accordance with our understanding of the nature of the linguistic phenomena to be acquired.
Following Hawkins (2001), the main evidence for a nativist approach to SLA that supports language development, and not just transfer from learners' L1 grammatical options, comes from studies that identify UG-specific constraints on L2 grammatical knowledge that do appear to be derived from L2 input.The fact that this study has observed development in the acquisition of the L2 Spanish CLD phenomena, especially by the group of Dutch learners with a more advanced level, seems to support such an approach.In other words, an approach to SLA based on explicit and implicit knowledge predicts that L2 learners are bound to overcome negative transfer effects at a later stage.The acquisition data observed during this study refer to a developing language system, the participant' interlanguage, where characteristics of both the L1 system and the target system are found, as is also extensively discussed by the literature of other studies (Benati & Angelovska, 2016).

Conclusion
To conclude, we have examined the L2 acquisition of Spanish dative clitics by English and Dutch learners.The linguistic phenomena concerning clitic doubling (CLD), where the dative clitic doubles a full dative phrase introduced by a dative preposition, has been a challenging configuration for both groups of learners.We have shown that CLD involves quite an intricate syntactic structure of several functional heads.To palliate the effect of such a complexity, we have tentatively proposed a learning method based on a gradual acquisition of salient patterns including transitive and more complex argument structures where dative clitics are inserted.We have put forward a view of language knowledge as the initial stage followed by explicit grammar instruction based on practices that can help capture the nature of the linguistic phenomena to be acquired.A proposal for further research is to study the acquisition of other types of clitics, like ethical dative clitics.Not only would this provide us with an analysis of other developmental stages, but the comparison of different methodological approaches would deepen our understanding of the whole process of L2 grammar acquisition.

( 13 )
Two developmental stages I. Dative clitic pronouns with simplex ditransitive configurations Juan le dio el libro.Juan DCLI gave the book (to Maria) 'John gave the book to her.' II.Dative-cltic doubling with complex ditransitive configurations Juan le dio a María el libro.Juan DTCL gave to Mary the book 'John gave Mary the book.'

Table 1
Placement test results: Correct response

Table 2
Experimental test results: Grammatical and ungrammatical sentences