L1 French learning of L2 Spanish past tenses: L1 transfer versus aspect and interface issues

This paper examines the process of acquiring L2s that are closely related to the L1 through data on how adult French speakers learning L2 Spanish in a formal setting develop knowledge and use of past tenses in this L2. We consider the role of transfer and simplification in acquiring mental representations of the L2 grammar, specifically in the area of tense and aspect, and how learners deal with integrating grammatically encoded, lexical and discursive information, including mismatching feature combinations leading to particular inferential effects on interpretation. Data is presented on the Spanish past tenses (simple and compound past, pluperfect, imperfect and progressive forms) from two tasks, an oral production filmretell and a multiple-choice interpretation task, completed by learners at A2, B1, B2 and C1 CEFR levels (N = 20-24 per level). L1 influence is progressively attenuated as proficiency increases. Difficulties were not always due to negative L1 transfer, but related also to grammar-discourse interface issues when integrating linguistic and pragmatic information in the interpretation process. This has clear implications for the teaching of closely related languages: instruction should not only José Amenós-Pons, Aoife Ahern, Pedro Guijarro-Fuentes 490 focus on crosslinguistic contrasts, but also prioritize uses requiring complex interface integration, which are harder to process.


Introduction: Past tenses in French and Spanish
In this paper we report empirical data on the acquisition, by instructed adult L1 French speakers learning Spanish (at A2, B1, B2 and C1 CEFR levels), of the ability to use L2 tense-aspect morphology: simple past (SP), compound past (CP), imperfect (IMP), pluperfect (PLP) and progressive (PROG) forms.The acquisition of the past tense morphology of L2 Spanish (as well as other Romance languages such as French, Italian and Portuguese) has received a great deal of attention, but most studies concentrate exclusively on the SP/IMP contrast, especially in groups of learners whose L1 does not have such aspectual contrast (mainly L1 English speakers).Little has been said, however, about closely related language pairings, and very few studies have included other tenses than SP and IMP.
Spanish and French share with the rest of the Romance languages the distinction between perfective and imperfective tenses, and the use of IMP is generally assumed to be similar in these two languages, although in fact some differences are found in modal contexts (Amenós-Pons, 2015;Escandell-Vidal & Leonetti, 2003), such as (1) and (2) below: (1) a. Sp.Si tuviera (IMP-SUBJ) dinero, daría (COND) la vuelta al mundo.
They went to bed early, because they had to leave at 8.00 in the morning.
On the one hand, as seen in ( 1), the imperfect subjunctive is generally the standard form in the antecedent of Spanish irrealis if-conditional clauses, while the indicative is compulsory in contemporary French. 1 On the other hand, as depicted in (2), the Spanish IMP is prone to prospective uses; they are not entirely ruled out in French, but the contexts allowing them are more restricted than in Spanish, and an auxiliary verb (in IMP) is often required in French (Azzopardi, 2011;Gosselin, 1999).These contrasts do not necessarily imply a crosslinguistic difference in the semantics of the IMP, but may rather be seen as a consequence of the overall diverging possibilities offered by the tense-aspect systems of each language (Amenós-Pons, 2015).
Besides, both languages have simple and compound past tenses, but, as shown in (2) above, they are used differently.In Spanish, the CP can be a perfect and a hodiernal past, but not a general narrative tense, since the SP is required for that purpose (Real Academia Española, 2009). 2 As for French, the SP is only found in written, mostly literary texts.In oral use, as well as in informal written language, the CP is the main narrative tense.The French CP can also be a perfect, both in oral and written usage (Corblin & De Swart, 2003).Crosslinguistic variation in the use of CP/SP in Romance is generally considered to reflect successive stages in the evolution of a common compound form, from its original resultative meaning towards a fully perfective tense-like function; the French CP is further along that journey than its Spanish counterpart (Lindstedt, 2000).
Finally, the PROG (not a tense, but a progressive operator) is found in Spanish and French, although its nature is different in each language: PROG is a morphosyntactic marker in Spanish, but a lexicosyntactic device in French.The mapping of PROG over the grammatical tense-aspect system also varies across the two languages, as in Spanish, PROG may be combined with perfective and imperfective tenses, while the French perfective tenses do not allow the use of PROG.In fact, even in imperfective contexts, the frequency of PROG is much lower in French than in Spanish.This may be related to the fact that the French construction is an innovation dating back only to the 18th century, and its use as a progressive device is not fully grammaticalized (Bertinetto, 2000): (3) a. Sp.Cuando Juan llegó (SP), Ana todavía estaba (IMP + PROG) trabajando.
b. Fr.Hier, Pierre a lu (CP) pendant deux heures.Yesterday, Peter spent two hours reading.
Within the syntactic framework of the minimalist program (Chomsky, 1995), tense and aspect are regarded as interpretable features, that is, elements that make a semantic contribution to the interpretation of utterances, apart from having a syntactic function.Similar features can be bundled in different ways, depending on the language.This is indeed the case of French and Spanish, as both have grammaticalized tense and aspect, and perfective and imperfective aspects may be combined, in turn, with all predicate types, telic and atelic.Still, as pointed out above, the uses of past tenses and aspectual operators (such as PROG) differ between French and Spanish.This implies that each of the two languages deploys these features in particular ways, and that the task of a French speaker learning Spanish will consist of appropriately re-assembling bundles of features, such as the tense-aspect ones, in the L2 lexicon, and determining the specific conditions under which their properties are morphophonologically expressed (Lardiere, 2008(Lardiere, , 2009)).The type of difficulties that this should cause and their permanent or transitory status are controversial issues within current generative approaches (see Ayoun & Rothman, 2013 and references therein), and upon which our study intends to shed new light.However, as is widely acknowledged, language use is not only a matter of syntax, but an interface phenomenon, where information from different systems (syntax, semantics and pragmatics) must be combined.When processing utterances, integrating linguistic and non-linguistic information has a cognitive cost, particularly in an L2, and this has often been reported as a source of variability (Sharwood Smith & Truscott, 2014;Sorace, 2011;Sorace & Filiaci, 2006).Thus, our study will also enquire into the specific complexities of the syntax/pragmatics interface for L2 acquisition of tense and aspect in closely related language pairings.The paper is organized as follows: Firstly, in Section 2, an overview of the empirical studies on the acquisition of L1 Spanish past tenses is offered; in Section 3, an introduction to the fundamental aspects of our own study is provided; in Section 4, we report on and discuss our results, finishing up with conclusions, limitations and suggestions for further research in Section 5, linking the experimental data with current theories on L2 acquisition by processing (Sharwood Smith & Truscott, 2014) and language/discourse-pragmatics interface issues.

Previous studies on L2 Spanish past tense acquisition
Researchers from different theoretical backgrounds tend to agree on certain core findings (Comajoan, 2014): (a) Before the emergence of tense-aspect morphology, learners make use of lexical and pragmatic devices to refer to past events (Dietrich, Klein, & Noyau, 1995); (b) in production studies, the SP tends to appear before the IMP (Bardovi-Harlig, 2000;Salaberry, 2011); (c) verbal forms are (at least partially) learned before their uses are mastered (Bardovi-Harlig, 2000;Dietrich, Klein, & Noyau, 1995;Montrul, 2004;Montrul & Slabakova, 2002); (d) the ability to use SP and IMP appropriately develops in stages, marked by prototypical associations of lexical aspect, discourse grounding and grammatical aspect (Andersen & Shirai, 1996;Bardovi-Harlig, 2000); this implies that (e) the use of SP in narrative discourses tends to be linked initially to telic predicates in the discourse foreground, while the IMP is related to atelic predicates in the background.Therefore, acquiring the use of past tenses may be described as learning to relate the tenses to the shifting needs of discourse, rather than prototypically relating verb forms to predicate types.In an L2, this can only be done at higher levels of proficiency, because it requires the ability to process and harmonize linguistic and extralinguistic cues (Comajoan, 2014 and references therein).
Among the studies mentioned in the previous paragraphs, however, only Dietrich, Klein, and Noyau's (1995) include the French-Spanish language pair, although they concentrate on naturalistic L2 learners and, compared to our own study, the language order is reversed (L1 Spanish/L2 French).Conversely, instructed French learners of Spanish were involved in a study carried out by Lorenzo (2002).Taking a discourse approach, the author studied the information structure of the narratives produced by advanced L1 French speakers of L2 Spanish; she claimed that the L1 habits strongly influence the way learners organize information, even at higher proficiency levels.In fact, as the author points out, due to the type of task (an orally told children's story), most of the participants chose to tell the story in the present tense (as often done in French), and not many past tense forms were produced.
Within the generative tradition, Diaz, Bel, and Bekiou (2007) investigated the acquisition of SP and IMP by different groups of L2 learners, including L1 Greek and also other Romance language speakers, whose languages encode the [± perfective] feature.It was found that the presence of this feature in the L1 helped those two groups of learners, compared with other L1 groups (from Chinese, Japanese and Slavic backgrounds), whose native languages do not specify that feature.After reviewing numerous generative studies on the acquisition of L2 English, Spanish, French and Portuguese by learners of diverse language backgrounds, Ayoun and Rothman (2013, p. 144) concluded that: (a) Instructed L2 learners tend to gradually improve, and eventually acquire, the ability to use the L2 tense-aspect-mood morphology, while at the same time, (b) interface-conditioned properties at the syntax-discourse interface are "more subject to vulnerability."The authors relate their claim to Sorace's interface hypothesis, whose "most current instantiation maintains that internal interfaces [i.e., syntax-semantics interfaces] should be less problematic than external ones [i.e., grammardiscourse pragmatics] for L2 convergence, due to differences in the attentional (processing) resource assignment needed (Sorace & Serratrice, 2009;Sorace, 2011)."Yet, only one of the tense-aspect studies described by Ayoun and Rothman (namely, Diaz, Bel, & Bekiou, 2007) included L1 Romance speakers.
Therefore, as anticipated in the Introduction, much remains to be done in the study of the acquisition of past tenses within close languages, and of the acquisition of tense-aspect devices other than the SP and IMP.These are the issues that will be the focus of the remainder of this paper.

Research questions and hypotheses
Three research questions form the basis of our study, namely: (1) How do SP, CP, IMP, PLP and PROG develop?(2) What is the effect of transfer from L1?, and (3) How or when do L2 learners acquire and use, if at all, non-prototypical associations (e.g., cases of aspectual coercion, non-focalized PROG, or pragmatically based temporal reference)?
In relation to these questions, two alternative hypotheses were considered.Firstly, that L1 transfer could be the main force guiding acquisition of closely related languages, or, alternatively, that L1 transfer could be complemented and counterbalanced by the need to reduce the cognitive burden when processing and producing L2 utterances.If the first possibility is correct, it would be expected that, for L1 French speakers, variability would appear in the use of CP/SP and in the use of PROG, especially with perfective tenses; in contrast, the uses of IMP and PLP would not pose substantial challenges, except (in the case of IMP) in specific modal contexts.However, if the need to avoid cognitive overload significantly influences language processing across closely related languages, input and output simplification could come about as an attempt to reduce such overload.Sticking to prototypical aspectual associations could be an effect of that tendency; difficulty in integrating semantic, syntactic and pragmatic information (e.g., discourse relations) would also be expected, especially at lower levels of proficiency.

Task design
Originally, the study included three tasks: (1) an oral film retell task, in which the participants were asked to tell the story from the point of view of the main character, using the first person (as a personal experience); (2) a written narrative task ("What was your most memorable birthday?");(3) a written error correction task, based on a short narrative sequence.The tasks were performed by 33 L1 French speakers (university background, age 20-70), learning Spanish at a formal setting (Instituto Cervantes in Paris, A2, B1, B2 and C1; between 8-9 informants per level).3A control group of 9 L1 speakers of European Spanish also did the tasks.Due to space limitations, only the first task will be reported here; it will be henceforth referred to as Task 1. Task 1 (as well as the two non-reported tasks) was aimed at gathering information on the participants' general ability to use past tenses at different stages of the L2 learning process.
At a later stage, two successive follow-up tasks were added in order to gather information on specific tense uses that were infrequent or nonexistent in the production data.These tasks targeted only potentially difficult uses of the tenses, namely: (a) IMP with telic predicates and in non-prototypical uses; (b) the use of PROG with perfective tenses; (c) the choice between CP/SP with interval localizing expressions; and (d) the choice of tenses based on discourse relations.The first follow-up task was a multiple choice, paper-and-pencil task, with 30 multiple choice items (three options per item), integrated into narrative text.The second one was performed on line, and it consisted of 50 independent multiple choice items.The common goal of those two tasks was to enquire more deeply into L2 learners' language competence than was possible with the production task, and to find out whether L1 speakers and L2 learners (at different proficiency levels) have diverse tendencies when trying to make sense of complex language combinations, whose interpretation normally requires taking into account a substantial amount of contextual data.Additionally, the second followup task was meant to confirm the tendencies found in the previous one.Due to space limitations, only the second follow-up task (henceforth referred to as Task 2) will be considered here, along with the production task (Task 1).

Task 1 (film retell task)
Each participant individually watched a 7-minute film, with sound and music, but virtually no dialogues. 4The participants were previously informed that, just afterwards, they would be required to tell the story taking the role of the main character.Each participant told the story in front of the researcher, answering his initial question "What happened to you on the bus?"During the performances, researcher participation was restricted to back-channeling.
The performances were video recorded and transcribed; non-verbal behavior related to chronological relations was included in the transcripts.The total amount of tense forms was then computed, and the predicate types connected to each tense were classified.The data on predicate types are assembled here in two broad categories due to space limitations: atelics (including states and activities) and telics (including accomplishments and achievements).Additionally, the functions performed by each tense used in the stories were categorized using the following tags: "discourse foreground" (i.e., clauses describing temporally sequenced events, which constitute the backbone of the narrative discourse), "discourse background" (clauses providing information on causes, reasons, motivation or descriptive details that help the reader understand the story), "current relevance", "hodiernal past events" and "backshifting" (events in reverse order).Results per L2 level are summarized in Figures 1, 2, 3, 4 and 5.In each case, the following details are given: (1) number and percentage of tokens per tense; (2) discourse functions performed by each tense, at the corresponding L2 level; 5 and (3) predicate types (telics/atelics) used with each tense. 5Tenses are reported following its frequency in the data; the word "NO" is included in some functions, to account for the fact that, in some of the stories produced, such functions were not included.

A2 results
At A2, all learners use the present indicative (PRES) as the main verbal device to tell the story (71% of the verb forms used are PRES), for every discourse function.In most productions, however, PRES occasionally alternates with CP (17%) to represent foregrounded events; CP is also used to express temporal backward shifting.The CP is correctly used both with telics and with atelics, although telics predominate significantly (p < .005 in the one-way ANOVAs).SP use is rare (only 6% of verb forms) and it never becomes the main narrative tense.Its appearances, however, always correspond to foregrounded events, and its use does not entail any chronological contrast with events in the PRES and in the CP.As in the case of the CP, the SP is correctly used with telic and (occasionally) with atelics.The occurrence of IMP at A2 is almost nonexistent (3% of the total verb forms), and its correct use is only found with atelic predicates in the discourse background. 6ikewise, the PLP is never found in the A2 productions: depending on discourse needs, the participants make gestures, or occasionally resort to the CP (contrasted with the PRES), when they intend to include events in reversed chronological order.7However, this is hardly a surprise, since the PLP has not been explicitly taught at this level, even though the learners have frequently encountered it in written texts.

B1 results
At B1, learners produce longer stories, with a wider array of tenses.The PRES (36% of the total amount of verb forms) is still the most frequent tense in the foreground, but its relative weight diminishes, and the SP (24%) is already second, ahead of the CP (14%), whose occurrence also decreases.Still, most B1 learners do not consistently differentiate the uses of these three tenses: only one participant told the story in the SP; another one chose to PRES to do so; in the rest of the productions, the alternation of PRES, CP and SP in the foreground did not parallel any meaningful contextual effect.
Besides, the B1 productions showed a dramatic increase in the use of the IMP (23% of the total verb forms), always linked to atelic predicates (both states and activities);8 the IMP is, from that level on, the main tense for backgrounded events.Activities in IMP (33% of the total amount) are often inflected in the PROG (at B1, PROG is used with 59% of the activities in the IMP).The PROG form never extends to other predicate types.
The PLP is still infrequent (3%) at B1, but it is nevertheless the main form used for backwards shifting purposes.Among the six occurrences of the PLP found in the B1 narratives, five come with telic predicates, and only one is with an atelic predicate; this is not surprising, since the PLP is a perfective tense.As for SP and CP, telic predicates also predominate, but atelics are found as well.

B2 results
Compared to the previous levels, participants at B2 show two main differences: on the one hand, stories are longer, and the accounts tend to be more detailed; on the other, the use of the SP to depict mainly telic, foregrounded events, keeps increasing (28%), while the occurrence of the CP radically decreases (4%).However, the PRES is still the most frequent tense (47%), and it always occurs in the foreground.The frequent amalgamation of PRES and SP (and occasional CP forms) in many productions does not seem a deliberate choice, as no clear connection to particular meaning effects is shown.
To denote backgrounded events, the IMP (18% of the verb forms) appears systematically, and participants make use of PROG + IMP, combined with activities (26,66% of the IMP forms with activities are inflected in the PROG) or, less frequently, with accomplishments (16.66%).The PROG form, however, never occurs with perfective tenses.Finally, a slight increase is shown in the use of PLP (3% of the verb forms), always in relation with events in reverse order.

C1 results
At C1, the use of the tenses in the narratives change substantially.The percentages are in fact very close to those found in the control group, with no significant (p > .05)differences between them.The SP is the most common tense (42%), and it has become the main form in the foreground.The PRES is now only the third form (18%), below the SP and the IMP (31%).However, in all the productions but two, foregrounded events in SP occasionally alternate with isolated PRES and CP forms, with no apparent expressive intention.
C1 learners often provide more background details than in previous levels.As a result, a considerable increase in the use of the IMP is found (31% of the total verb forms); 48.85% of the activities and 33.33% of the accomplishments in the IMP are inflected in the PROG.In addition, events in reverse order are also more frequent, with a relative increase in the presence of the PLP (6%).In the C1 data, all tenses appear both with telic and atelic predicates, although the percentage of telics, as expected, is substantially higher for the perfective tenses, and atelics are more frequent with the IMP.

Control group results
In the control group, the relative percentage of each tense is similar to the percentages found at C1: the SP is the most abundant tense (40%), followed by the IMP (31%) and the PRS (22%).The PLP is scarce (5%), and so is the CP (2%).
All native speakers consistently employed the SP as the main foreground tense.The PRES was also used to refer to past events in two native speaker narratives, but not in the same way as the L2 speakers, even at C1.In the narratives of the control group, the PRES forms did not appear in isolation.On the contrary, they were concentrated at certain parts of the narratives to create a dramatic effect of immediacy, which was never found in the non-native speaker narratives.Also, three (out of nine) native speakers made occasional use of SP forms inflected in PROG, a combination that was never detected in the productions of the L2 groups.The IMP was always used in the control group (mainly with atelic predicates) to signal events in the background; 24.39% of the activities and 15.38% of the accomplishments in the IMP were inflected in the PROG.As in the C1 data, in the native speaker narratives all tenses appear both with telic and atelic predicates, although the percentage of telics is substantially higher for the perfective tenses, while atelics in the IMP are more widespread.

Task 1: Conclusions
According to our data, the A2 and B1 learner participants generally use the PRES as the main narrative tense, while they make an undifferentiated use of occasional CP/SP forms to convey the idea that the events are located in the past.The use of SP as the main foreground tense starts at B2 and consolidates at C1. 9  The IMP appears later than the perfective tenses, although at B1 it has become the main device to signal backgrounded events.As for the PLP, even though its morphology and use can be directly transferred from French, it is not frequent before B1.In fact, the PLP is a relatively marginal form in the productions of all groups, including the control group, since natives and non-natives alike show a clear preference for events presented in chronological order.
The overuse of the PRES at A2, B1 and (to a lesser degree) B2 may be seen as a single effect caused by several factors.The PRES is often described in the literature as a base form, used in L2 narratives, especially at lower levels, with no specific temporal meaning (Bardovi-Harlig, 2000).At intermediate and higher levels, however, it can also be an avoidance strategy (learners use the PRES to avoid conjugating past tense forms that may be difficult for them). 10Additionally, L1 discourse habits (Lorenzo, 2002) may influence the preferences of the French speakers when choosing a main tense to tell stories. 11 At all levels, perfective tenses attract telic predicates, and the IMP attracts atelics.Aspectual restrictions seem to be stronger in the IMP: (Correctly used) 9 However, these are general trends that are not evenly distributed in all the productions within a single level; individual variability exists at all levels. 10In the written production task ("What was your most memorable birthday?"),at B1, the SP was already the main narrative tense.This strongly suggests that some degree of avoidance of the SP in the oral narratives is a fact. 11According to Lorenzo (2002), French speakers often use the PRES to tell children's stories; still, personal reports do not necessarily show the same tendency (Amenós-Pons, 2010).telic predicates in IMP are not found until B2.However, this almost exclusive association of atelics and the IMP, in this task, is not rare in the productions of the native speakers; thus, there seems to be an effect of the task in the type of predicates used in the narratives. 12verall, cases of incorrect use of the SP instead of the IMP, or the reverse, are only occasional in the data, and practically non-existent at B2 and C1.Therefore, at lower levels, avoidance of past morphology is more frequent than incorrect use.A potential, as well as extreme, case of avoidance is found in the complete absence of the PROG form associated to perfective tenses; it never appears in the L2 productions (unlike in the native speaker narratives).Indeed, the impossibility of such combination in French may be a crucial fact.
Generally speaking, the evidence does not show L1 French speakers directly transferring the use of the past tenses from their native language.However, there is indeed an L1 effect, in that the tense uses that do not coincide in L1 and L2 are a source of variability.This is in line with the prediction of feature accounts on L2 acquisition (especially, Lardiere's [2009] feature reassembly hypothesis).Output simplification also plays a role, as some non-prototypical combinations of tense and aspect (particularly PROG + SP/CP) tend to be avoided, principally at lower levels.
Nevertheless, the fact that perfective tenses are employed, already at A2, with all types of predicates, is not compatible with the predictions of those hypotheses that highlight the primacy of aspect in the development of L2 past tense use (Andersen & Shirai, 1996).Conversely, our findings are compatible with those hypotheses that emphasize the role of discourse functions (Bardovi-Harlig, 1994;Salaberry, 1999): Verbs representing foregrounded events are the first to be inflected in perfective forms, especially (but not exclusively) if they are telic; the IMP appears later, related to atelic predicates in the background.In the case of L1 French speakers, the inflected form in the foreground is not always the SP, but the CP and the SP in undifferentiated alternation.
On the whole, Task 1 shows that the L2 learners improve gradually, and eventually acquire the tense/aspect system of Spanish (as postulated by Ayoun & Rothman, 2013), even though some degree of variability persists at C1 level.However, since Task 1 relies only on production data, a fundamental question remains: The omission of some forms (e.g., PROG + SP) and the variability found in others (e.g., SP/CP), are just performance effects due to communicative pressure or, on the contrary, do those trends stem from a deep representation of meaning and use that is different from that of native speakers?To shed light on those issues, we will now present the results of Task 2.

Task 2 (online multiple choice task)
As mentioned above, the purpose of Task 2, an online multiple-choice interpretation task, consisted in exploring the process of tense and aspect acquisition from the perspective of the learners' ability to interpret discursive and linguistic elements that determine the appropriacy of certain verb tenses in potentially complex, non-prototypical environments.The task design eliminated the element of communicative pressure that was present for Task 1, thereby providing a possibly more accurate insight into the learners' mental representations of Spanish tense and aspect grammar and usage.Furthermore, this task developed an inquiry which honed in on forms that the results of Task 1 suggested were more challenging for the L2 learners.
The 70 native speakers of French who participated were learning Spanish in the formal setting of the Instituto Cervantes in Paris or Toulouse, and had been placed by a standardized placement test at levels A2 (N = 15), B1 (N = 20), B2 (N = 16) or C1 (N = 19); and additionally, a control group of 25 Peninsular Spanish L1 speakers also did the task.None of these participants had taken part in the previously mentioned stages of the study. 13However, the tendencies shown in Task 1 were assumed to be generalizable to the L1 French learners of Spanish as a whole, and thus formed the basis for the design and choice of items of Task 2.
The task format consisted of an online questionnaire which began with a brief text providing instructions, followed by 50 randomized gap-fill items.8 of these were distractors offering choices of various verb tenses, and the 42 remaining items were the target ones in which, for a given sentence in Spanish, participants chose between 3 tenses provided in the same order for each verb, namely, CP, SP or IMP.Response time was measured in seconds for all 50 items.Finally, an ethnolinguistic questionnaire was presented, requesting information on the participants' age range, first and other languages spoken, and Spanish languagelearning experience, in terms of whether this included any experience of study, or any periods visiting or living in a Spanish-speaking country and for how long. 14 Among the tense uses seldom produced, or totally absent, in the narratives of Task 1, those involving certain kinds of interpretations of SP and IMP with different predicate types, and combinations of these tenses with certain temporal adverbials, were focused on in this task, as well as PROG with SP.Thus, the item conditions were as shown in the following chart, with 6 items for each condition set.Three sets of conditions fall under the umbrella of interpretations connected to the perfect (Dahl & Hedin, 2000), in Items 1 to 18; the following sets covered the hodiernal CP, progressive readings of IMP, echoic interpretations of IMP, and finally, PROG with the auxiliary (estar) in CP.In general, the response options include one which is identified as more expectable given the sentential and/or discourse context provided; however, for the CP items, as is reflected by the control group results, a degree of flexibility is present regarding what an individual participant may find to be the best option, depending for instance on factors like diatopic variation or the way the situation is conceptualized by each individual (e.g., more or less currently relevant, or whether the event has finished or not).The general overview of the results of this task can be summed up as follows.
To begin with, unsurprisingly, the learner groups were significantly (p < .05)less accurate than the control group in the great majority of their responses; however, for certain sets of items, specifically for readings of CP that are obtained in both French and Spanish under similar discourse conditions-resultative CP and experiential CP (1-6 and 7-12, respectively)-the difference in accuracy between learners and controls was not significant.
As to the other item conditions, upon contrasting and comparing the results obtained, some general patterns are observable.Firstly, in relation to choosing perfective tenses, as required in Items 1 to 24, probably as a result from L1 transfer, learners at A2 often preferred the CP, which was an advantage for them since the CP was in fact the expected answer.At B1, however, the effect of positive transfer was attenuated, and more participants (incorrectly) chose the SP.Overall, little, if any, progress takes place across the general proficiency levels.Likewise, the ability to adequately identify and respond to discourse contexts where PROG with SP, a combination that the learners lack in their L1, did not increase significantly across proficiency levels either.And, finally, an area in which the learners did show progression was identified when they were required to choose between perfective and imperfective tenses.These three patterns are explained in further detail below.
Regarding the choice of perfective tenses, as just mentioned, the results show that the learners make no significant progress towards convergence with the native speakers' responses, despite managing to respond with fairly high accuracy to these items from the elementary level of A2.In fact, in one group of items (Items 19 to 24), whose most appropriate response was the CP with a hodiernal interpretation, learners tended to respond differently to the native speaker group more often as general proficiency increased (see Figure 7).For the PROG with SP items (37-42), the difference in accuracy between the learners and the control group remained more constant than in cases like those just shown, where an initial advantage in choosing SP for the hodiernal interpretation was lost as learners gained in general proficiency.Thus, in items 37-42, a stable and significant (p < .05)difference was detected between the learner groups across the four levels of general proficiency, on the one hand, and the control group, on the other (Figure 8).A slight increase in accuracy of the learner groups across proficiency levels is shown, but does not reach significance.

Figure 8
Learner and control group results for Items 37 to 42, PROG with SP As to the learners' performance in the use of the IMP, the results suggest that as their proficiency increases, learners do manage to improve their ability, even in the complex conditions tested, for choosing between IMP and perfective tenses, in contrast with what was observed for the item conditions seen above.A significant increase in accurate tense choice is observed across the proficiency nor found within the control group, but native speakers (of European Spanish) clearly favored interval location (with the CP).Items 37 to 42 levels for items 25 to 30, where the IMP (combined with telic predicates) has a progressive interpretation, although at C1 there is still a significant difference between the learners' accuracy and that of the control group, which chose the expected tense more frequently.But the tendency to improve in tense use as general proficiency is gained does not apply to all of the IMP uses tested.The items in which this tense was elicited in a context that provokes a quotative/echoic interpretation 16 proved particularly challenging for the participants, regardless of their general proficiency level; and the learner groups' accuracy was markedly, and of course statistically significantly, lower than that of the control group.In contrast, the control group was practically unanimous in its-predicted-preference for the IMP in these items (see Figure 9).The difficulty for those items may be due to the combination of two facts: (a) the environments in which such readings are possible do not coincide in French and Spanish; and (b) the recognition of those readings largely depends on the hearer's ability to recognized complex sets of contextual cues, both linguistic and pragmatic (Amenós-Pons, 2015;Escandell-Vidal & Leonetti, 2003).

Figure 9
Learner and control group results for Items 31 to 36, IMP with quotative readings

Task 2 response time data
The data on the participants' response time in Task 2, briefly described in this section, may provide some interesting insights into the levels of challenge it presented to each group.One of the main facts observed in this respect was that the control group consistently responded with significantly (p < .05)greater 16 In quotative/echoic interpretations of the IMP, the tense is understood as reporting words or thoughts from someone who is not the speaker (or, alternatively, from the speaker at a time different from that of the main utterance).speed than the learner groups, regardless of whether the learners are considered as separate groups by proficiency level, or collectively as a whole. 17Figure 10 (below) illustrates this difference in relation to an item group in which it was more marked.As to the learner groups, separated from one another according to their general proficiency, there were no statistically significant differences among them in response time.Additionally, the results show that response time was not significantly related to accuracy for any of the participant groups.
Figure 10 Response time mean (in seconds) by participant group for Items 31-36

Task 2: Conclusions
The intended focus of Task 2, consisting of offering more direct evidence of the acquisition of certain tense, morphological and lexical aspect combinations that were suggested to be more problematic for the learners by the results of Task 1, led to the prediction that their interpretation would prove challenging in general for these participants.What the data seems to indicate, besides corroborating the generalized difficulty for learners in aligning with the control group in their choice of the most appropriate form for each sentential context, is: (a) that the challenge is more prevalent for certain item conditions, and (b) that these learners' initial advantage in certain tense-aspect uses is lost as progress in general proficiency is gained.
In connection with the variation according to the different conditions tested in each set of items, that is, depending on the tenses and interpretations that were elicited, it is worth highlighting that mixed results were obtained.So, as 17 This was so for every set of conditions except that of Items 19-24, in which the controls were still faster but the difference with respect to the learner groups' times was not statistically significant.shown in the previous section, progression in accuracy was observed as statistically significant only for the choice between perfective and imperfective tenses (Items 25 to 30).On the other hand, certain items were less challenging for all the proficiency levels, and the learner groups equaled the accuracy-in statistical terms-of the control group, specifically, those with current relevance readings of the CP.This difference in the relative ease or difficulty of the acquisition process according to the item conditions is related to the learners' L1 grammar, as also suggested in connection with the Task 1 results (Section 4.1.6).Thus, it can be assumed that the learners' tense and aspectual parameters have been set for French; this language is characterized by its many features in common with those of Spanish.Developing familiarity with the specific conditions under which each of these properties are expressed in the L2 becomes, thereby, the main obstacle in the acquisition process, as in Lardiere's (2009) feature reassembly hypothesis.
Clearly, the need to integrate grammar and discourse-pragmatic cues is a central element of this challenge at upper intermediate and advanced levels.
For instance, the items with IMP where a quotative interpretation was obtained show a case in which the learners could take advantage of similarities between French and Spanish, since the same kind of interpretation is obtainable in French when IMP appears under certain discourse conditions.However, that advantage is not found in the results, since the learner groups still showed a significantly lower accuracy rate even at C1 level, while at lower levels, the kind of transfer effects found, for instance, in the CP uses of Items 1 to 17 were not shown.These differences should be seen in light of the fact that French quotative interpretations of IMP are much less frequently found and take place under more highly restricted conditions (Amenós-Pons, 2015).
The data obtained from Task 2 provide clear evidence for the positive effect of learners' transferring knowledge of CP into the L2.However, taking into account the lack of improvement in the ability to use and interpret some of the tense/aspect combinations tested in this task among learners at increasing levels of general proficiency, the conclusion in this respect highlights the limitations of transfer.It may be that increasing knowledge of the range of tense morphology and usage weakens the possible effects of direct transfer.Considered from an acquisition-by-processing perspective of SLA, such as those put forth in work by Sharwood Smith and Truscott (2014, and references therein), this may be due to learners' increasing metalinguistic knowledge of the L2 tense and aspect system.That is, learners take into account a wider range of tense and aspect options as they acquire more of the L2, and therefore also develop awareness of the lexical and discourse conditions that affect appropriateness or grammaticality of a tense/aspect choice.In some cases, this may result in obstructing the efficiency of the learners in those conditions common to both French and Spanish.

General conclusions and implications for L2 teaching
In this paper, two different tasks have been discussed: an open production task (Task 1) and a controlled interpretation task focusing on complex tense and aspect combinations (Task 2).Three research questions were asked, namely: (1) How do SP, CP, IMP, PLP and PROG develop?(2) What is the effect of transfer from L1?, and (3) How or when do L2 learners acquire and use, if at all, nonprototypical associations?
Relevant data on the first question was gathered by means of Task 1.It was found that, at A2, the learners made little use of past morphology.Only occasional, functionally undifferentiated CP and SP forms were found to signal foregrounded events, alternating with pervasive PRES morphology.Perfective past morphology increased its presence at B1, but only at B2 did the learners make consistent, clearly differentiated use of the two forms.At all of the general proficiency levels, perfective tense was mainly, but not exclusively, used with telic predicates.The IMP was seldom used at A2, but it became the main form in the discourse background from B1 onwards.Telic events in IMP were also infrequent, but this was mainly found to be an effect of the task.PLP forms were not detected at A2, but it was coherently used by the learners from B1 onwards to report events on reversed chronological order.However, the production of PLP forms did not increase with general proficiency, since native and non-native speakers alike showed a clear tendency to report the events respecting their chronology.Finally, PROG was first seen with activities in IMP at B1, and later on, from B2 onwards, its use extended to accomplishments in IMP.Combinations of PROG with perfective tenses, which are possible in Spanish but not in French, were never detected in the L2 productions.
Overall, the results from T1 are compatible with hypotheses that emphasize the role of discourse functions in the development of tense-aspect morphology (Bardovi-Harlig, 1994;Salaberry, 1999).In Task 1, no systematic transfer from L1 was found at any proficiency level.There was, however, an L1 effect, which may have both positive and negative consequences.On the one hand, L1 French speakers at A2 were already able to differentiate perfectivity and imperfectivity; they were also capable of combining tense morphology with different predicate types, telic and atelic.On the other hand, the tense uses that do not coincide in French and Spanish remained a source of difficulty, suggesting, therefore, that feature reassembly leads to variability at all levels, although at higher proficiency levels it decreased substantially.Further effects of L1 transfer were found in Task 2: It was seen that, in complex combinations of tense, aspect and discourse environments, the A2 learners were strongly reliant on L1 knowledge when deciding the appropriateness of a tense.Yet as participants' proficiency increased, L1 transfer was less prevalent, possibly counterbalanced by increasing reliance on metalinguistic knowledge.Paradoxically, this did not systematically result in increasing accuracy.
In relation to the third research question, our research offers mixed results.In Task 2, progress was found for some conditions, but not others.In particular, no significant progression was seen in those items requiring complex integration of syntactic, lexical and discourse-pragmatic cues.This can be seen as indirectly related to the claims of Sorace's interface hypothesis (Sorace, 2011;Sorace & Filiaci, 2006)-although this hypothesis is primarily related to ultimate attainment, not to developing L2 grammars-that variability stems from the complexity of processing at the syntax-pragmatics interface.
Overall, Task 1 and Task 2 offer complementary results.This clearly implies that, in order to investigate the relationship between active use and mental representations, the combination of open production and controlled interpretation tasks is required.The way learners use tenses as a discursive device is clearly seen only in production tasks, but in these tasks, learners use forms that are contextually accessible for them.Still, the absence of a form does not necessarily mean lack of knowledge.Therefore, only controlled interpretation tasks can effectively test specific hypotheses on learners' representations of tense-aspect relations.
As a final point, some implications of our research for L2 teaching could be emphasized.The idea that our data show that feature reassembly is a source of variability highlights the need for attention to those uses that differ in L1 and L2.However, interface integration is also a source of variability, so working on L2 vocabulary building and discourse structuring are paramount to ensure L2 grammar proficiency.In teaching L2 at higher levels, tense uses requiring complex interface integration should be given priority.Finally, it is clear that discourse processing is a demanding task at all levels of L2 proficiency.A fundamental symptom of the evolution of L2 competence is the increasing ability to take into account global discourse relations when interpreting and producing utterances (Comajoan, 2014).Thus, in teaching closely related L2s, grammar instruction should not merely rely on lists of convergences and divergences in isolated sentences; it should also include extensive practice, both receptive and productive, based on increasingly complex discourse sequences.

Figure 7
Figure 7 Learner and control group results for Items 19 to 24, Hodiernal CP (interval adverbials + telic predicates)

Task 2 results: Accuracy and progress across general proficiency levels
se ha examinado/se examinó/se examinaba + How's Pedro?-Very busy I think.He said_______________ for his licence next week.(Sit the exam) se ha examinado/se examinó/se examinaba