ORIGINAL RESEARCH published: 20 January 2022 doi: 10.3389/fpsyg.2021.794587 The Distribution of Subjects in L2 Spanish by Greek Learners Panagiota Margaza* and Anna Gavarró Departament de Filologia Catalana, Universitat Autònoma de Barcelona, Barcelona, Spain Our study examines the expression and position of subjects in L2 acquisition, two phenomena that are studied within the framework of the Interface Hypothesis (IH). The first version of the IH predicts that interface properties involving syntax and another cognitive domain may not be fully acquirable in a second language (Sorace and Filiaci, 2006; also Sorace, 2011). The second version of the IH predicts that formal properties involving the syntax-semantics interface are unproblematic to acquire in L2 grammars compared to the vulnerable properties integrating syntax with the higher level of pragmatics (Tsimpli and Sorace, 2006). We test these IH versions in L2 Spanish as acquired by L1 Greek speakers, a language combination understudied in the literature. Both languages share the null subject parameter, but still the IHs predict incomplete command at the syntax-pragmatics interface. Two acceptability judgment tasks were designed for Spanish: the first task tested null/overt subjects in referential Edited by: contexts and the second task tested preverbal/postverbal subjects in informational José Antonio Hinojosa, contexts. Participants were L1 Greek intermediate and advanced learners of Spanish Complutense University of Madrid, and native speakers of Spanish (15 subjects in each group). In the first task, both Spain experimental groups showed target-like distribution of null/overt subjects in most non- Reviewed by: Cristina Maria Flores, contrastive and contrastive contexts, except for the advanced group in unambiguous University of Minho, Portugal referential contexts. In the second task, the respective groups accepted felicitous Verónica Moreno Campos, University of Castilla La Mancha, preverbal subjects with unergative verbs, but diverged from native-like distribution of Spain postverbal subjects with unaccusative verbs in neutral contexts. The L2 groups showed *Correspondence: a high preference for unfelicitous preverbal subjects with both intransitive verbs in Panagiota Margaza informational contexts, contrary to the subject inversion patterns of the control group. notabarcelona@yahoo.gr The results obtained were not consistent with the IH predictions, and other factors such Specialty section: as the type of subject, verb class and context played a role in L2 performance. This article was submitted to Language Sciences, Keywords: null subjects, preverbal subjects, postverbal subjects, syntax-pragmatics interface, syntax-semantics a section of the journal interface, Interface Hypothesis, L1 Greek – L2 Spanish Frontiers in Psychology Received: 13 October 2021 Accepted: 09 December 2021 INTRODUCTION Published: 20 January 2022 In this study, we explore the acceptance of null/overt subjects and preverbal/postverbal subjects in Citation: specific pragmatic contexts in L2 Spanish acquisition by Greek learners. Our aim is to examine Margaza P and Gavarró A (2022) The Distribution of Subjects in L2 both the expression and position of subjects within the framework of the Interface Hypothesis Spanish by Greek Learners. (IH), as most studies examine either the expression of null/overt subjects (see Clements and Front. Psychol. 12:794587. Domínguez, 2017; Lozano, 2018) or the position of preverbal/postverbal subjects (see Lozano, doi: 10.3389/fpsyg.2021.794587 2006a,b; Domínguez and Arche, 2014); scarce previous work on L2 Spanish addresses these two Frontiers in Psychology | www.frontiersin.org 1 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish properties of the null subject languages. After studying involving a higher level of language use, integrating properties the anaphora resolution of null/overt subjects, Sorace and of language and pragmatic processing. Their definition of the IH Filiaci (2006) introduced the first version of the IH (IH-1 is stated in (2). hereafter), which predicts that interface structures involving the mappings between syntax and other cognitive domains (2) IH-2 (after Tsimpli and Sorace, 2006: 656) such as pragmatics are more complex to acquire in L2. Formal properties involving the syntax-semantics interface After examining focalization and subject uses, Tsimpli and are unproblematic to acquire in L2 grammars compared to Sorace (2006) proposed a new version of the IH (IH- the vulnerable properties integrating syntax with the higher 2 hereafter), arguing that structures involving the syntax- level of pragmatics. semantics interface are easier to acquire than structures involving In their study, Tsimpli and Sorace (2006) examined word the syntax-pragmatics interface. In this study, we aim at order in relation to focalization and the expression/omission testing these versions of the IH in an understudied language of subjects in relation to person features (1st/2nd/3rd). The combination, L1 Greek – L2 Spanish, Greek and Spanish two phenomena were examined based on L1 Russian learners’ being two languages that share the null subject value and performance in L2 Greek. The results showed that the the unergative/unaccusative universal distinction. Still, the two L2 learners had acquired the felicitous word order with languages present some differences in the position of subjects focalization, while they overused 1st/2nd against 3rd person in informational focus contexts. In this case, our main aim is overt pronouns, showing that person had an effect on L2 to examine if acquiring interface phenomena remains difficult performance. The authors claimed that word order at the given the similar distribution of subjects in these two null syntax-semantics interface was easier to acquire than subject subject languages. use when the syntax-pragmatics interface was involved, in This paper is structured as follows: section “Background” support of the IH-2. However, in their own results L2 presents the IH and examines the distribution of subjects in learners had no problems with the distribution of null subjects, L2 Spanish. Section “The Study” presents our study: the IH so that the syntactic-pragmatic constraints were not always predictions and the methods of our research, including the compromised.1 experimental design, the procedure and the data analysis. The Sorace (2011: 15) rephrased the IH as follows: “L2 learners are main results of our analysis are detailed in section “Results.” less efficient than monolinguals at processing structures at the The discussion of the substantial findings and the overall syntax-pragmatics interface because their knowledge of or access conclusions of our study appear in section “Discussion and to computational constraints is less detailed or less automatic Conclusions.” than in monolinguals and they have fewer cognitive resources to deploy on the integration of different types of information in real- time language use.” The syntax-pragmatics interface is claimed BACKGROUND to be the main locus of processing difficulties and acquisition delays at the highest levels of L2 ultimate attainment. Interface The Interface Hypothesis: Versions and problems are attributed to the fact that L2 learners need to Objections acquire both the representational knowledge of the structure Sorace and Filiaci (2006) proposed the first version of the and the mapping conditions that operate within interface Interface Hypothesis, as defined in (1). components, and the processing principles that apply in real-time integration of different domains. Sorace (2012: 210) explicitly (1) IH-1 (Sorace and Filiaci, 2006: 340) states that there is “a hierarchy of computational difficulty” Interface properties involving syntax and another cognitive with structures requiring proceduralized “internal” mappings domain may not be fully acquirable in a second language. being less taxing than structures requiring the integration of contextual information. In their work, Sorace and Filiaci (2006) examined anaphora White (2011b: 109) questioned Sorace’s claim that the IH resolution at the syntax-pragmatics interface, concerning the use does not hold at all L2 developmental stages, since interface of null/overt subjects in appropriate contexts. In particular, they problems, should they occur, would not emerge out of the focused on the mastery of anaphora resolution in L2 acquisition blue, but appear in the course of L2 language development, of Italian by L1 English. In their results, the L2 learners had not only at the near-native stages. White (2011a: 588) also problems with the interpretation of overt pronouns in relation argues that even if L2 non-native performance reveals processing to their antecedents, but showed target-like processing of null difficulties in acquiring interface phenomena, this does not imply pronouns. This asymmetry between null and overt subjects was not consistent with the IH-1, as both types of subjects were 1 Sorace and Serratrice (2009: 200) proposed an extension of the Interface predicted to present target-deviant distribution at the syntax- Hypothesis (IH-3) for bilingual acquisition, in the following terms: “Processing pragmatics interface. limitations in bilingual speakers may be responsible at least for some of the Tsimpli and Sorace (2006) also proposed a second version of difficulties attested at the interfaces, especially the ones requiring the coordination of syntactic and contextual information, while internal interfaces may be expected the IH (IH-2 hereafter), taking into consideration the distinction to be less sensitive to processing limitations because they involve mappings between the syntax-semantics interface, involving the formal between formal properties of the language system alone.” This is an extension of properties of grammar and the syntax-pragmatics interface, the IH to bilingual populations, which are not the focus of this manuscript. Frontiers in Psychology | www.frontiersin.org 2 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish permanent impairment at the interfaces. Slabakova (2009) and (4) a. El fin de semana proi salgoi con mis amigos. White (2011a,b) also cast doubt on what is considered “difficult” the end of week pro go.out.1SG.PRS with my friends and what “easy” under the IH, as the syntax-pragmatics interface “The weekend I go out with my friends.” is not necessarily found to be more problematic than other b. To savatokirjako proi vjenoi me tus filus linguistic domains, such as the syntax-semantics interface (see for example Serratrice et al., 2009; Sorace et al., 2009). The argument the weekend pro go.out.1SG.PRS with the friends is therefore that it is inappropriate to make broad generalizations mu. for interface domains. mine Taking into consideration that not all syntax-discourse “The weekend I go out with my friends.” interface properties are equally problematic, Lozano (2016) formulated a more specific proposal, the Pragmatic Principles (5) a. Juan habla con sus colegas. (unergative) Violation Hypothesis (PPVH), which makes predictions Juan speak.3SG.PRS with his colleagues mostly on the distribution of null/overt subjects. In his “Juan speaks with his colleagues.” study, Lozano (2016) examined anaphora resolution in L2 b. O Janis milai me tus sinaderfus tu. Spanish by L1 English speakers. His results showed that the.NOM Janis speak.3SG.PRS with the colleagues his the L2 learners presented native-like use of null subjects in topic-continuity contexts, while they avoided this type of “Janis speaks with his colleagues.” subject in topic-shift contexts. Regarding overt pronouns, the L2 learners preferred the expression of pronominal subjects (6) a. Ayer vino María. (unaccusative) in topic-shift contexts, but also used this type of subject yesterday come.3SG.PST María in topic-continuity contexts. Thus, it was easier to overuse “Yesterday María came.” the unfelicitous overt pronoun in topic-continuity contexts b. Xthes irthe i Maria. than to resort to the unfelicitous null pronoun in topic- yesterday come.3SG.PST the.NOM Maria shift contexts. These results were against the predictions of “Yesterday María came.” the IH of complete vulnerability at the syntax-pragmatics interface. Based on pragmatic principles related to redundancy (7) ¿Quién camina/va. . .? (“Who walks/goes. . .?”) and ambiguity, following the neo-Gricean principles of Informativeness and Manner, he proposed that overt- a. Camina Juan por el parque. when-null violation to mark topic-continuity leads only to walk.3SG.PRS Juan through the park redundancy and not to informative breakdown, while null- “Juan walks in the park.” when-overt violation to mark topic-shift leads to ambiguity b. Va Juan a la playa. that causes communicative failure in discourse. He stated his go.3SG.PRS Juan to the beach hypothesis as in (3). “Juan goes to the beach.” (3) PPVH (after Lozano, 2016: 243) (8) Pjos perpatai/pijeni. . .? (“Who walks/goes. . .?”) Advanced learners will violate pragmatic principles a. O Janis perpatai sto parko. banning redundancy more often than principles banning the.NOM Janis walk.3SG.PRS in.the park ambiguity, by being pragmatically more “redundant” “Janis walks in the park.” (producing redundant overt anaphors to mark topic- continuity) than “ambiguous” (producing ambiguous null b. O Janis pijeni stin paralia. anaphors to mark a shift in topic). the.NOM Janis go.3SG.PRS to.the beach “Janis goes to the beach.” This proposal follows the IH in placing the syntax-pragmatics interface as the locus of delay in L2 acquisition, but is While L2 Spanish acquisition with regard to subject expression more restrictive. and subject position is common in the literature, the combination of L1 Greek and L2 Spanish is understudied. Lozano (2018) explored the distribution of null/overt subjects in L2 Spanish Subject Distribution in L2 Spanish by L1 Greek learners. He focused on the development of Spanish and Greek share the null subject parameter (4) (see pronominal subjects at three proficiency levels (intermediate, Fernández-Soriano, 1989 for Spanish and Philippaki-Warburton, lower-advanced, and upper-advanced). In contrastive contexts, 1987, 1989 for Greek) and the unergative/unaccusative all L2 groups distinguished the felicitous overt pronoun from distinction affects word order (5) and (6) (see Eguren and the unfelicitous null pronoun. In this case, the upper-advanced Fernández-Soriano, 2004 for Spanish and Alexiadou and group showed convergence with native behavior, but presented Anagnostopoulou, 2004 for Greek). However, the two languages some persistent deficits in topic-continuity contexts, in which differ with respect to subject position in informational contexts, they accepted redundant overt pronouns. At lower levels, L2 VS in Spanish and SV in Greek (see Roussou and Tsimpli, 2006), learners alternated between null and overt subjects, confirming as illustrated in (7) and (8). a higher divergence from native-like patterns. Lozano (2009) Frontiers in Psychology | www.frontiersin.org 3 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish also claimed that deficits at the syntax-pragmatics interface were overt/null subjects (Clements and Domínguez, 2017; Lozano, selective, as the type of person played a role in the performance 2018) or the position of preverbal/postverbal subjects (Lozano, of learners who had problems with the anaphoric uses of 3rd 2006a,b; Domínguez and Arche, 2014). The combination of two person pronouns, while they presented better mastery of the null subject languages, L1 Greek and L2 Spanish, is also new, deictic uses of 1st/2nd person. Lozano (2006a) examined subject- as in most studies the combination involves a non-null subject verb alternations in the data of three (upper-intermediate, language and a null subject language, e.g., L1 English and L2 lower-advanced, and upper-advanced) experimental groups of Spanish (see Hertel, 2003; Montrul, 2005; Rothman, 2009). Our L1 Greek learners of L2 Spanish. All L2 groups showed native- main aim is to examine the extent to which Greek learners like mastery of the felicitous subject position with intransitive of Spanish show command of both null/overt subjects and verbs at the syntax-semantics interface. On the other hand, all preverbal/postverbal subjects in referential and informational L2 groups had problems with the distribution of subject-verb at contexts. These phenomena are tested under the light of the the syntax-pragmatics interface, except for the upper-advanced two versions of the IH as well as Lozano’s (2016) PPVH. The group that showed a clear preference for the felicitous VS with predictions of the above hypotheses are as follows. unaccusative verbs in informational contexts. Lozano’s (2006a) results supported again native-like command of syntactic- i According to the IH-1, L2 learners are expected to accept semantic properties, but did not confirm that all syntactic- the unfelicitous type of subjects, null or overt, preverbal pragmatic properties were inacquirable due to permanent or postverbal in non-contrastive, unambiguous and vulnerability. Hertel (2003) also found that learners of higher contrastive referent-shift contexts as well as informational levels performed native-like at the syntax-pragmatics interface, so contexts in which the syntax-pragmatics interface is that discourse-related word order was eventually acquired in L2. involved.2 Domínguez and colleagues also examined subject distribution ii According to the IH-2, L2 learners are expected to in L2 Spanish by L1 English speakers. Clements and Domínguez accept the felicitous subject position with intransitive (2017) found that, in contrastive contexts, L2 learners at an verbs in neutral contexts in which the syntax-semantics advanced level followed native-like intuitions with respect to interface is involved, while they will have problems the use of overt subjects, while they performed target-deviant with subject expression and position in referential and in the case of unfelicitous null subjects. In switch referent informational contexts in which the syntax-pragmatics contexts, though, advanced learners did not differ from native- interface is involved. like performance when the felicitous option was a null pronoun, iii If Lozano’s (2016) PPVH is accurate, L2 learners will showing command of the pragmatic constraint involving the overuse overt pronouns in referential contexts in which a possibility of omitting subjects in topic-shift contexts. In non- null pronoun is expected, while they will perform target- topic-shift contexts, L2 learners also approached the rates of like in pragmatic contexts in which an overt pronoun natives, showing no problems with the use of null pronouns. is expected. Pragmatic failure leading to redundancy is However, advanced learners did not reject unfelicitous overt predicted but not leading to ambiguity. pronouns to the same extent as the control group. The problematic nature of the syntax-pragmatics interface was not Methods always supported by the results. Domínguez and Arche (2014) In our study, we apply offline tasks that examine subject also examined the position of subjects at three proficiency processing under a particular time limit in the discourse. We levels (beginner, intermediate, and advanced). Results indicated have chosen a written task that allows to better control subject that beginner and intermediate groups preferred SV in all acceptability in relation to the type of context, referent antecedent contexts, while the advanced group accepted VS over SV with and verb class. Contextualized pragmatic felicitousness judgment unaccusative verbs in broad and narrow focus contexts, but tasks have been widely used in applied studies on L2 Spanish showed optionality between the two word orders with unergative acquisition (see Hertel, 2003; Lozano, 2006a,b, 2018; Clements verbs in focus contexts. Subject inversion acquisition was a slow and Domínguez, 2017), so this is a suitable method to examine process, as systematic preference for inversion was only observed the real preferences of L2 learners. at advanced levels. Persistent SV/VS problems were caused not exclusively by the syntax-pragmatics interface, but also by the Materials: Experimental Design syntax-semantics interface when verb class was involved (see also We designed two acceptability judgment tasks to examine Montrul, 2005). the type of subject, null or overt, preverbal or postverbal in contextualized sentences in Spanish. These experiments include a 5-point Likert-scale from −2 (fully rejected), −1 (rejected), THE STUDY 0 (neither rejected/neither accepted) to 1 (accepted) and 2 (fully accepted); this allows the rating of the exact degree of Predictions 2 In this study, we also examine the expression and position of Here we take into consideration White’s (2011a,b) claim that the IH should hold of developmental stages and we examine not only the very advanced levels (“the subjects in various pragmatic contexts in Spanish. The novelty near-natives” as originally proposed by the IH), but also the intermediate levels of our study is that we give an account of both phenomena, of knowledge, which are extensively studied in the literature for the interface as in the literature most studies examine either the use of framework (Lozano, 2006a,b, 2018; Domínguez and Arche, 2014). Frontiers in Psychology | www.frontiersin.org 4 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish acceptability of the type of subjects in various pragmatic contexts Subtests 2 and 3 involve the acceptability of 3rd person (see also Lozano, 2006a,b, 2018). null/overt subjects. Subtest 2 consists of five items requiring Experiment 1 consists of a total number of 21 stimuli: 16 null subjects in referent-shift contexts (with one unambiguous items with two sentences each (total: 32 sentences) testing antecedent). The variables tested are 3rd Person and Subject type the acceptance or rejection of null/overt subjects and five in a given context, giving rise to the following conditions: (i) 3rd distractors (total: 10 sentences). The variables checked are: (i) person, null subject, referent-shift context (with one antecedent), person (1st, 2nd, 3rd) and (ii) discourse context (referent- and (ii) 3rd person, overt subject, referent-shift context (with continuity, unambiguous referent-shift, and contrastive referent- one antecedent). Since the distribution of null pronouns is shift). Experiment 2 consists of a total number of 25 stimuli: constrained by the syntax-pragmatics interface, the IH-1 and 20 items with two sentences each (total: 40 sentences) testing IH-2 predict that L2ers will fail to accept the felicitous type the acceptance or rejection of preverbal/postverbal subjects and of pronoun in unambiguous referent-shift contexts; similarly, five distractors (total: 10 sentences) that do not involve the Lozano’s (2016) PPVH predicts that L2ers will be target-deviant, phenomena examined, so that they are not further analyzed. as in the previous subset of items. The variables controlled for are: (i) verb class (unergative and In example (10) the inflection of the verb decir (“say”) allows unaccusative) and (ii) context (neutral or informational). the identification of the 3rd person of the antecedent referent el Experiment 1 includes three subtest conditions. Subtest 1 profesor (“the teacher”) so that the production of a null subject involves the acceptability of 1st and 2nd person null/overt is acceptable in Spanish (10a). However, an emphatic/contrastive subjects in non-contrastive referential contexts. This subtest overt pronominal subject él (“he”) is possible in referent-shift contains six items: three items with 1st person subjects and three contexts (10b). (In Greek, a null subject is also the first choice in items with 2nd person subjects. Both types of person demand the equivalent contexts, but the expression of an overt pronoun is the production of null subjects in non-contrastive referential not disallowed). contexts. The variables tested are Person and Subject type in a given context, giving rise to the following conditions: (i) 1st (10) Cuando el profesor imparte clases person, null subject, non-contrastive referential context, (ii) 1st when the teacher give.3SG.PRS classes person, overt subject, non-contrastive referential context, (iii) de matemáticas, ___________ 2nd person, null subject, non-contrastive referential context, of mathematics, ___________ and (iv) 2nd person, overt subject, non-contrastive referential (a) sus alumnos no entienden ni context. Under the IH-1 and IH-2, L2ers are expected to his students not understand.3PL.PRS not.even accept unfelicitous overt pronouns of both persons due to the involvement of pragmatics; likewise, under Lozano’s (2016) la mitad de las PPVH, L2ers will overaccept overt pronouns when null the half of the pronouns are expected. cosas que dice To illustrate, in example (9) the null subject is felicitous in things that say.3SG.PRS referent-continuity contexts in which the inflection of the verb −2 −1 0 1 2 volver (“return”) shows the 1st person singular in (9a). In this (b) sus alumnos no entienden ni la case, the expression of the overt pronoun yo (“I”) (9b) would his students not understand.3PL.PRS not.even the be redundant. Still, an overt pronoun would be acceptable with mitad de las cosas que emphatic/contrastive interpretation, but this is not the first choice half of the things that in the sentence examined in Spanish. (In Greek, subject omission él dice. is also the preferred option in the equivalent contexts). he.NOM say.3SG.PRS −2 −1 0 1 2 (9) Ayer, cuando salí del trabajo, “When the teacher gives mathematics class, his students yesterday, when leave.1SG.PST from.the work, do not understand even half of the things that he says.” __________. __________ Subtest 3 involves five items that require the expression of 3rd (a) volví a casa para cenar con mis padres. person subjects. The variables tested are 3rd Person and Subject return.1SG.PST to house to dine.INF with my parents type in a given context, giving rise to the following conditions: (i) −2 −1 0 1 2 3rd person, overt subject, contrastive referent-shift context (with two antecedents), and (ii) 3rd person, null subject, contrastive (b) yo volví a casa para cenar con mis referent-shift context (with two antecedents). Under the IH-1 I.NOM return.1SG.PST to house to dine.INF with my and IH-2, L2ers will fail with the felicitous overt pronoun, due padres. to difficulties in acquiring referential properties. On the other parents hand, Lozano’s (2016) PPVH predicts that L2ers will perform −2 −1 0 1 2 target-like. “Yesterday, when I left work, I went back home to In example (11), the inflection of the verb hablar have dinner with my parents.” (“speak”) shows the 3rd person singular in Spanish, but Frontiers in Psychology | www.frontiersin.org 5 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish it does not distinguish the antecedent referents Manolo (12) Estás en casa y oyes las voces de or Sofía, so the expression of the pronoun él (“he”) is be.2SG.PRS in house and listen.2SG.PRS the voices of obligatory in (11a) to refer to the antecedent (Manolo). la gente que A null subject would generally refer to the closest singular the people that antecedent (Sofía) in the discourse (11b). (In Greek, está en la calle. Luego, tu madre the felicitous option is also an overt pronoun in the be.3SG.PRS in the road. then, your mother equivalent contexts). vuelve de su return.3SG.PRS from her (11) Manolo y Sofía trabajan en una empresa trabajo y le preguntas: ¿Qué sucede? Manolo and Sofía work.3PL.PRS in a company work and her.ACC ask.2SG.PRS: what happen.3SG.PRS? multinacional. Y ella responde: multinational. and she.NOM answer.3SG.PRS: (a) Sus colegas dicen que él no habla (a) Mucha gente camina por la calle. their colleagues say.3PL.PRS that he.NOM not speak.3SG.PRS many people walk.3SG.PRS around the road muy bien inglés. −2 −1 0 1 2 very well English (b) Camina mucha gente por la calle. −2 −1 0 1 2 walk.3SG.PRS many people around the road (b) Sus colegas dicen que no habla −2 −1 0 1 2 their colleagues say.3PL.PRS that not speak.3SG.PRS “You are at home and listen to the voices of the people muy bien inglés. being in the road. Then, your mother comes back from very well English her work and you ask her: What is happening? And she −2 −1 0 1 2 answers to you: A lot of people are walking in the road.” “Manolo and Sofía are working in a multinational company. Their colleagues say that he does not speak English very well.” (13) Ayer, mientras estabas en el banco, yesterday, while be.2SG.PST.CONT in the bank, viste a un ladrón. In Experiment 2, we examine the position of subjects see.2SG.PST to a thief. with intransitive verbs in various contexts. Two subtest Hoy tu amiga Juana te pregunta: conditions are included. Each subtest contains five items with unergatives and five with unaccusatives. Subtest 1 today your friend Juana you.ACC ask.3SG.PRS: involves the unergative/unaccusative distinction that allows ¿Qué sucedió en el the anteposition of unergative subjects and the postposition what happen.3SG.PST in the of unaccusative subjects (see also Lozano, 2006a,b). This banco? Y tú contestas: distinction is examined in direct question-answer pairs in bank? and you.NOM answer.2SG.PRS: which the informational focus is neutral, so the syntactic- (a) Entró un ladrón. lexical-semantic properties of verbs constrain the position enter.3SG.PST a thief of their subjects. The variables tested are Word Order and −2 −1 0 1 2 Verb Class in a given context, giving rise to the following conditions: (i) SV, unergative verb, neutral context, (ii) (b) Un ladrón entró. VS, unergative verb, neutral context, (iii) SV, unaccusative a thief enter.3SG.PST verb, neutral context, and (iv) VS, unaccusative verb, −2 −1 0 1 2 neutral context. Under the IH-2, L2 learners will accept “Yesterday, while you were at the bank, you saw a thief. the felicitous subject position with both verb classes at the Today your friend Juana asks you: What happened in syntax-semantics interface. the bank? And you answer: A thief walked in.” In the contextualized examples (12) and (13), the broad focus questions ¿Qué sucede? (“What happens?”) and ¿Qué sucedió Subtest 2 examines the distribution of subjects in en el banco? (“What happened in the bank?”) trigger as new informational focus contexts. In this case, the syntactic- information the entire answer, allowing SV with the unergative pragmatic properties of focus determine word order, so the caminar (“walk”) in (12a) and VS with the unaccusative entrar felicitous word order is VS with both unergative/unaccusative (“enter”) in (13a). On the other hand, the second word verbs. The examined contexts contain direct question-answer order option is not acceptable in Spanish neutral contexts in pairs, as in Subtest 1, but, in this case, the question is with (12b) and (13b). (In Greek, unergative verbs also accept the ¿Quién. . .? (“Who.?”), triggering a focalized subject that SV order, while unaccusative verbs allow the VS order in introduces new information into the discourse. The variables neutral contexts). tested are also Word Order and Verb Class in a given context, Frontiers in Psychology | www.frontiersin.org 6 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish giving rise to the following conditions: (i) SV, unergative verb, regresa María, te pregunta: ¿Quién informational context, (ii) VS, unergative verb, informational return.3SG.PRS María, you.ACC ask.3SG.PRS: who context, (iii) SV, unaccusative verb, informational context, and ha (iv) VS, unaccusative verb, informational context. Under both have.AUX the IH-1 and IH-2, L2ers will accept the unfelicitous option of preverbal subjects as the syntax-pragmatics interface is involved venido? Tú respondes: in the contexts examined. Lozano’s (2016) PPVH does not make come.3SG.PRS.PRF? you.NOM answer.2SG.PRS: predictions for this subset of cases. (a) Ha venido una chica. In the contextualized examples (14) and (15), the narrow have.AUX come.3SG.PRS.PRF a girl focus question with ¿Quién.? (“Who. . .?”) receives as answer the −2 −1 0 1 2 VS order with the unergative reírse (“laugh”) in (14a) and the (b) Una chica ha venido. unaccusative venir (“come”) in (15a). The anteposition of subjects a girl have.AUX come.3SG.PRS.PRF is unacceptable in (14b) and (15b) contexts in Spanish. (In Greek, −2 −1 0 1 2 though, the felicitous option is the SV order with both intransitive “María leaves the library for a moment and in a minute a verbs in informational focus contexts). girl appears that you do not know. When María comes back, she asks you: Who has come? You answer: A girl has (14) Estás en una clase de física. Todo el mundo come.” be.2SG.PRS in a class of physics. all the world está callado All test items were fully randomized.3 be.3SG.PRS silent Participants mientras el profesor explica Participants of both experiments were two groups of L1 while the teacher explain.3SG.PRS Greek learners of L2 Spanish and a group of native Spanish la lección, pero un speakers. The non-native groups consisted of intermediate and the lesson, but a advanced students who were learning Spanish as an L2 at the chico se ríe. El Instituto Cervantes de Atenas. At the time of the experiments, boy REFL laugh.3SG.PRS. the intermediate and advanced learners were attending the respective profesor no ve quién third and fifth Spanish courses for 4 hours a week. Both groups teacher not see who had passed the Examination for the Diploma of Spanish as a Foreign Language (DELE). Intermediate learners had obtained se ríe, así que an average rate of 86% in the B1 Exam, while advanced learners REFL laugh.3SG.PRS, so that had attained a mean of 89% in the C1 Exam, according te pregunta: ¿Quién se to the European Framework for Foreign Languages. Spanish you.ACC ask.3SG.PRS: who REFL native speakers were living in Madrid and were students at the ríe? Tú respondes: Universidad Autónoma de Madrid. This third group served as laugh.3SG.PRS? you.NOM answer.2SG.PRS: a control group and established the rate of acceptability of the (a) Se ríe un chico. various types of subjects in Spanish. Table 1 shows the essential REFL laugh.3 SG . PRS a boy information for the three groups. −2 −1 0 1 2 (b) Un chico se ríe. TABLE 1 | Participants. a boy REFL laugh.3SG.PRS −2 −1 0 1 2 Groups Intermediate Advanced Control “You are in the physics class. All people are silent while First language Greek Greek Spanish the teacher explains the lesson, but a boy is laughing. Number 15 (3 males 15 (2 males 15 (5 males The teacher does not see who is laughing, so that he asks and 12 females) and 13 females) and 10 females) you: Who is laughing? You answer: A boy is laughing.” Age range (SD) 30-60 (9.27) 34-62 (8.24) 30-50 (7.07) Studies in L2 3rd L2 course 5th L2 course —— Spanish (15) María sale de la librería un momento y Duration 3 years 5 years —— María leave.3SG.PRS from the library a moment and Proficiency level B1 C1 Native al minuto Average score in 86% 89% —— in.a minute DELE exams aparece una chica a la que no conoces. appear.3SG.PRS a girl to the whom not know.2SG.PRS. 3 A reviewer suggests that a between-subjects design would have been more Cuando appropriate, with several lists and the distribution of the experimental items per list following the Latin Square procedure, and would have allowed for more distractors, when something to take into account for future research. Frontiers in Psychology | www.frontiersin.org 7 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish Procedure referential, and contrastive) and Subject (null/overt) are Both Experiments 1 and 2 were administered in the Instituto presented in Table 2. In the three subtest conditions, the Cervantes de Atenas, where the classes of L2 Spanish were L2 intermediate and advanced groups showed a higher rate taught and in the Universidad Autónoma de Madrid, where of felicitous than unfelicitous type of subjects, following the the native speakers were studying. All participants provided patterns of the control group, except for the advanced group in written informed consent to participate in the tasks of our study. the overt subject condition in 1st person non-contrastive and Participants also answered a language questionnaire about their unambiguous referential contexts. L1 (Greek being the only L1 examined in this study), and their In the GLM, no significant differences between groups knowledge of L2 Spanish, to distinguish between two competence for both null/overt subject conditions in non-contrastive levels for L2 learners, which determined the factor of Group and contrastive contexts were found. On the other hand, in the statistical analysis. Then all groups were instructed as to in unambiguous referential contexts there were significant how to complete the acceptability judgment tasks and how to differences in the overt subject condition (F = 5.23, rate the two-sentence items. The five points of the scale were p-value = 0.0093). The statistical differences were detected explained, as follows: −2 (fully rejected), −1 (rejected), 0 (neither in the comparison between advanced and control groups rejected nor accepted), 1 (accepted), and 2 (fully accepted). The (t = 3.23, adj p-value = 0.0066, according to Tukey correction). participants were also given a distractor example that indicated To examine the interaction of Person (1st and 2nd) and how to rate the felicitous and unfelicitous options. All questions Group (intermediate, advanced, and control) in non-contrastive and doubts were answered to avoid misunderstandings. The contexts, a GLMM was applied. In the null subject condition, duration of each task was 45 min, but participants were given there were no significant differences detected. In the overt subject extra time if necessary. condition there were significant differences with respect to Person (F = 9.54, p-value = 0.0036), but no interaction of Person and Coding of Data and Statistical Analysis Group. In the post hoc test, there were significant differences In Experiments 1 and 2, the ratings of subject types (null/overt between 1st and 2nd person (t = −3.09, adj p-value = 0.0036 or preverbal/postverbal) on the five-point scale were classified according to Tukey correction). See Figure 1 for the 1st vs. 2nd as follows: the accepted (1, 2) and rejected (−1, −2) values person comparison. were grouped together, while the neither accepted/rejected (0) Regarding Experiment 2, the number and percentage of value was also noted as third category. Subjects were coded acceptance values (1, 2) by Context (neutral and informational), in accordance with the context of each given condition. For Verb Class (unergative and unaccusative), and Subject each condition, a Generalized Linear Model (GLM) was used (preverbal/postverbal) appear in Table 3. In the first subtest to compare percentages of accepted items (1, 2) across different condition, the L2 intermediate and advanced groups showed levels, using the binomial distribution (see Dobson and Barnett, a native-like rate of the felicitous preverbal subjects with 2018). Also we examined the interaction of Person (1st and 2nd) unergatives in neutral contexts, while both groups showed and Group (intermediate, advanced, and control) in Experiment variability between SV and VS with unaccusatives, diverging 1 and Verb Class (unergative and unaccusative) and Group from target patterns. In the second subtest condition, both L2 in Experiment 2 applying a Generalized Linear Mixed Model groups showed a higher preference for the unfelicitous subject (GLMM) of accepted items (1, 2) with the binomial distribution position in informational contexts, and the intermediate group (see Moscatelli et al., 2012). The GLMM has a high statistical presented full variability in the case of unaccusatives, against power as it estimates the variability of fixed and random native intuitions for postverbal subjects. effects. P-values were adjusted according to Tukey correction In the GLM there were no significant differences between for multiple comparisons. The statistical analysis was performed groups in the case of unergatives in the first subtest condition, using the software SAS v9.4, SAS Institute Inc., Cary, NC, while the statistical differences were significant in both preverbal United States. The statistical decisions were made taking as (F = 4.37, p-value = 0.0189) and postverbal subject conditions significance level the value 0.05. (F = 4.44, p-value = 0.0179) with unaccusatives. In the preverbal subject condition, the significant differences were detected in the comparison between advanced and control groups (t = 2.58, RESULTS adj p-value = 0.0353, according to Tukey correction). In the postverbal subject condition, the differences were found in Experiment 1 yielded a total of 1,440 responses (480 from each both intermediate-control (t = 2.48, adj p-value = 0.0449, group), while Experiment 2 yielded a total of 1,800 responses according to Tukey correction) and advanced-control group (600 from each group). The responses to the distractors were comparisons (t = −2.98, adj p-value = 0.0131, according to not included in the analysis because they did not involve the Tukey correction). phenomena examined. Prior to analysis, the responses were In order to examine the interaction of Verb Class (unergative categorized following the grouping of accepted and rejected and unaccusative) and Group (intermediate, advanced, and values, while the zero value was not selected by any of control) in neutral contexts, a GLMM was applied. In both SV the participants. (F = 40.87, p-value < 0.0001) and VS conditions (F = 75.7, Regarding Experiment 1, the number and percentage of p-value < 0.0001), there were significant differences regarding accepted values (1, 2) by Context (non-contrastive, unambiguous Verb Class, but no interaction of Verb Class and Group. In Frontiers in Psychology | www.frontiersin.org 8 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish TABLE 2 | Overall means and Standard Deviation in the contexts of Experiment 1. Acceptance percentage INTERM ADVAN CONTR N Mean Std N Mean Std N Mean Std Non-contrastive context Null 15 97% 9% 15 99% 4% 15 100% #Overt 15 28% 26% 15 42% 33% 15 36% 34% 1st person Null 15 98% 9% 15 100% 15 100% #Overt 15 18% 25% 15 42% 41% 15 22% 37% 2nd person Null 15 96% 12% 15 98% 9% 15 100% #Overt 15 38% 35% 15 42% 37% 15 49% 40% Unambiguous referential context Null 15 92% 17% 15 88% 13% 15 97% 7% #Overt 15 48% 42% 15 60% 28% 15 33% 40% Contrastive context #Null 15 11% 15% 15 15% 26% 15 5% 9% Overt 15 89% 21% 15 91% 21% 15 97% 7% Overt. 100% 80% Estimated Accepted percentage 60% 40% 20% 0% 1st person 2nd person 1st person 2nd person FIGURE 1 | Estimated acceptance percentage according to Person in the overt subject condition. the post hoc test, there were significant differences between postverbal: t = 4.16, adj p-value = 0.0004, according to Tukey unergatives and unaccusatives in both SV (intermediate: correction) and advanced-control group comparisons (preverbal: t = −3.37, adj p-value = 0.019; control: t = −4.64, adj t = 5.02, adj p-value < 0.0001; postverbal: t = −4.77, adj p-value = 0.0005, according to Tukey correction) and VS p-value < 0.0001, according to Tukey correction). Regarding conditions (intermediate: t = 5.69, adj p-value < 0.0001; unaccusatives, statistical differences between groups were also advanced: t = 5.58, adj p-value < 0.0001; control: t = 4.85, found in both preverbal (F = 26.27, p-value < 0.0001) and adj p-value = 0.0002, according to Tukey correction). See postverbal subject conditions (F = 8.18, p-value = 0.0010). Figures 2, 3 for unergative and unaccusative comparison in SV The significant differences were detected in both intermediate- and VS conditions. control (preverbal: t = −5.76, adj p-value < 0.0001; postverbal: In the second subtest condition, the GLM showed that t = 2.98, adj p-value = 0.0131, according to Tukey correction) there were significant differences between groups with preverbal and advanced-control comparisons (preverbal: t = 6.25, adj (F = 17.80, p-value < 0.0001) and postverbal subject options p-value < 0.0001; postverbal: t = −3.92, adj p-value = 0.0009, (F = 11.45, p-value = 0.0001) with unergatives in informational according to Tukey correction). contexts. The significant differences were detected in both To examine the interaction of Verb Class (unergative and intermediate-control (preverbal: t: −4.6, adj p-value = 0.0001; unaccusative) and Group (intermediate, advanced, and control) Frontiers in Psychology | www.frontiersin.org 9 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish TABLE 3 | Overall means and Standard Deviation in the contexts of Experiment 2. Acceptance percentage INTERM ADVAN CONTR N Mean Std N Mean Std N Mean Std Neutral context Unergative SV 15 97% 7% 15 95% 21% 15 97% 7% #VS 15 37% 28% 15 33% 31% 15 51% 20% Unaccusative #SV 15 76% 31% 15 77% 26% 15 57% 33% VS 15 84% 20% 15 79% 22% 15 97% 10% Informational context Unergative #SV 15 85% 18% 15 89% 17% 15 48% 36% VS 15 63% 34% 15 55% 33% 15 95% 12% Unaccusative #SV 15 81% 32% 15 87% 25% 15 32% 38% VS 15 79% 34% 15 65% 37% 15 97% 10% Neutral Context. SV. 100% 80% Estimated Accepted percentage 60% 40% 20% 0% INTERM B1 ADVAN C1 CONTR Unergative Unaccusative FIGURE 2 | Estimated acceptance percentage for the SV condition in neutral contexts according to Verb Class. Neutral Context. VS. 100% 80% Estimated Accepted percentage 60% 40% 20% 0% INTERM B1 ADVAN C1 CONTR Unergative Unaccusative FIGURE 3 | Estimated acceptance percentage for the VS condition in neutral contexts according to Verb Class. Frontiers in Psychology | www.frontiersin.org 10 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish in informational contexts, a GLMM was applied. In the Non-contrastive Unambiguous Contrastive SV condition, there were no significant differences regarding 120% Verb Class and there was no interaction of Verb Class and 100% Group in the acceptance values. In the VS condition, the 80% Means differences were significant with respect to Verb Class (F = 5.45, 60% p-value = 0.0244), and there was no interaction of Verb Class and 40% 20% Group. In the post hoc test, there were no significant differences 0% between unergatives and unaccusatives in neither condition for Null #Overt Null #Overt Overt #Null the three groups. Intermediate Advanced Control FIGURE 4 | Subject uses in referential contexts. DISCUSSION AND CONCLUSIONS In this study, we have examined the distribution of subjects in the judgments of L1 Greek intermediate and advanced learners of L2 Spanish in two contextualized acceptability tasks. Neutral Informational In Experiment 1, both intermediate and advanced learners Unergative Unaccusative Unergative Unaccusative showed native-like acceptance of the felicitous null subjects 120% 100% against the unfelicitous overt subjects in non-contrastive 80% Means referential contexts (Subtest 1). Their performance was 60% 40% independent of the type of person, 1st or 2nd in the null 20% subject condition. However, in the overt subject condition, 0% person had an effect on subject expression, but this was not related to the factor of group, as both L2 groups did not significantly diverge from native patterns. In unambiguous Intermediate Advanced Control referent-shift contexts (Subtest 2), the L2 groups also showed a higher acceptance of 3rd person null subjects than overt FIGURE 5 | Subject positions in pragmatic contexts. subjects, though they did not reach the ceiling rates of 1st/2nd person in the previous contexts. The intermediate group showed a tendency toward target-like patterns with both null and overt subjects, while the advanced group presented significant divergence from native rates in the case of the unfelicitous overt we observe that the intermediate and advanced learners of subjects. In contrastive referent-shift contexts (Subtest 3), the Spanish had no persistent problems with the distribution of L2 groups followed native-like judgments with both felicitous null/overt subjects in non-contrastive and contrastive referent- overt subjects and unfelicitous null subjects of 3rd person. In shift contexts in Experiment 1, so that their performance runs this case, all groups clearly rejected null pronouns in favor of against the predictions of the IH-1 (Sorace and Filiaci, 2006) expressing overt pronouns with contrastive interpretation. The and the IH-2 (Tsimpli and Sorace, 2006). Lozano’s (2016) PPVH tendencies of the intermediate and advanced groups with respect can also be rejected in non-contrastive contexts, as the L2 to the control group are shown in Figure 4 for the three contexts groups did not show significant differences from natives with examined in Experiment 1. overt pronouns. However, the advanced group diverged from In Experiment 2, the intermediate and advanced groups native rejection of overt subjects in unambiguous referent- showed target-like acceptance of felicitous preverbal subjects shift contexts, showing optionality in their judgments along the with unergative verbs in neutral contexts, but diverged from predictions of the IH-1/IH-2 and the PPVH. native-like subject inversion with unaccusative verbs in neutral Regarding Experiment 2, the results run against the IH-2 contexts (Subtest 1). Thus, verb class played a significant (Tsimpli and Sorace, 2006), as both L2 groups had difficulties role in the judgments of L2 learners with respect to subject with the distribution of postverbal subjects-unaccusative verbs position. In informational contexts (Subtest 2), both L2 groups at the syntax-semantics interface. The syntactic-lexical/semantic showed divergence from native-like distribution of subjects properties were not always acquired, and verb class played a role with unergative and unaccusative verbs. Both groups accepted in neutral contexts, as the position of subjects with unergative unfelicitous preverbal subjects, compared to the control group verbs was acquired earlier by L2 learners. In informational that showed a clear preference for the discursive VS order. In this contexts, the non-target performance of both L2 groups showed case, context type had a higher effect on learners’ performance that they had not yet acquired the syntactic-pragmatic properties than verb type. See Figure 5 for the word order patterns of of subject distribution. Here the influence of the L1 Greek the L2 groups with respect to the control group in neutral and that allows SV in informational contexts against VS in Spanish informational contexts in Experiment 2. (Roussou and Tsimpli, 2006) might be the source of non-target If we examine the overall results of the two experiments performance, so that the L2 learners overgeneralized the L1 against the predictions formulated in section “Predictions,” felicitous option. Frontiers in Psychology | www.frontiersin.org 11 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish The results showed that both L2 groups performed better DATA AVAILABILITY STATEMENT than expected under the IH-1 and IH-2 in non-contrastive and contrastive contexts in Experiment 1. The involvement The raw data supporting the conclusions of this article will be of pragmatics did not necessarily lead to unacceptable use of made available by the authors, without undue reservation. subjects in referential contexts. The predictions of Lozano’s (2016) PPVH were not correct in contexts where a null subject was the felicitous option, and the distribution of ETHICS STATEMENT redundant overt pronouns was variable because they were accepted in unambiguous referent-shift contexts, but rejected in The studies involving human participants were reviewed and non-contrastive contexts, while ambiguous null subjects were approved by the Comissió d’Ètica en l’Experimentació Animal correctly highly avoided in contrastive contexts. Regarding i Humana (CEEAH), Universitat Autònoma de Barcelona. Experiment 2, the predictions of the IH-2 were not fulfilled in The participants provided their written informed consent to the performance of L2 groups. The syntax-semantics interface participate in this study. was not necessarily acquired earlier than the syntax-pragmatics interface. The L2 groups showed a better performance in the case of referential null/overt subjects than in the case of informational AUTHOR CONTRIBUTIONS preverbal/postverbal subjects, to the effect that not all syntactic- pragmatic properties were equally acquirable or inacquirable in PM and AG designed this study. PM conducted the experiments L2, against both versions of the IH. and statistical analysis. Both authors are responsible for Overall, the IH-1 and IH-2 failed to account for the results of interpreting the results, and writing the manuscript. the two experimental tasks. The PPVH was not fulfilled either in Experiment 1 (the only experiment here for which it made any predictions). The IH-1 did not capture the performance FUNDING of the intermediate and advanced learners of Spanish in the case of null/overt subjects, but only for the advanced group in This manuscript has received the support of projects 2017 SGR unambiguous referent-shift contexts, and for both L2 groups in 634 from the Generalitat de Catalunya and FFI2017-87699-P informational subject-focused contexts. The IH-2 fared well for from the Ministerio de Economía y Competitividad. both L2 groups only in the case of the unergative word order (not for unaccusatives) –not in the case of referential uses of null/overt subjects. The PPVH fared better than the IH in the case of ACKNOWLEDGMENTS contrastive contexts, but not for non-contrastive contexts. Thus, none of the hypotheses considered captured the performance The authors wish to thank all the students of the Instituto of L2 learners in all cases examined. Our interpretation would Cervantes de Atenas and the Universidad Autónoma de Madrid be that the performance of L2ers was affected by grammatical that participated in the experiments in this study. The authors factors, such as the type of subject, verb class and context, and also wish to thank Ester Boixadera from the Servei d’Estadística best accounted for in terms of transfer effects. The involvement Aplicada of the UAB for helping with the GLM and GLMM of interfaces was orthogonal to performance in L2 Spanish statistical analyses and the two reviewers of Frontiers for their by L1 Greek speakers, a result that questions the empirical comments and suggestions. Any remaining errors are the authors’ adequacy of the IH. responsibility. REFERENCES Hertel, T. J. (2003). Lexical and discourse factors in the second language acquisition of Spanish word order. Second Lang. Res. 19, 273–304. doi: 10.1191/ Alexiadou, A., and Anagnostopoulou, E. (2004). “Voice morphology in the 0267658303sr224oa causative inchoative alternation: Evidence for a non-unified structural analysis Lozano, C. (2006a). “The development of the syntax-discourse interface: Greek of unaccusatives,” in The Unaccusativity Puzzle: Explorations of the Syntax- learners of Spanish,” in The Acquisition of Syntax in Romance Languages, eds Lexicon Interface, eds A. Alexiadou, E. Anagnostopoulou, and M. Everaert V. Torrens and L. Escobar (Amsterdam: John Benjamins), 371–399. (Oxford: Oxford University Press), 114–136. Lozano, C. (2006b). Focus and split-intransitivity: the acquisition of word order Clements, M., and Domínguez, L. (2017). Reexamining the acquisition of null alternations in nonnative Spanish. Second Lang. Res. 22, 145–187. doi: 10.1191/ subject pronouns in a second language: focus on referential and pragmatic 0267658306sr264oa constraints. Linguist. Approaches Biling. 7, 33–62. doi: 10.1075/lab.14012.cle Lozano, C. (2009). “Selective deficits at the syntax-discourse interface: evidence Dobson, A., and Barnett, A. (2018). An Introduction to Generalized Linear Models, from the CEDEL2 corpus,” in Representational Deficits in SLA, eds N. Snape, 4th Edn. Boca Raton: CRC Press, doi: 10.1201/9781315182780 Y. Ingrid Leung, and M. Sharwood Smith (Amsterdam: John Benjamins), Domínguez, L., and Arche, M. J. (2014). Subject inversion in non-native Spanish. 127–166. doi: 10.1075/lald.47.09loz Lingua 145, 243–265. doi: 10.1016/j.lingua.2014.04.004 Lozano, C. (2016). “Pragmatic principles in anaphora resolution at the syntax- Eguren, L., and Fernández-Soriano, O. (2004). Introducción a una sintaxis discourse interface: advanced English learners of Spanish in the CEDEL2 minimista. Madrid: Gredos. corpus,” in Spanish Learner Corpus Research: Current Trends and Future Fernández-Soriano, O. (1989). Strong pronouns in null subject languages and the Perspectives, ed. M. Alonso Ramos (Amsterdam: John Benjamins), 235–265. avoid pronoun principle. MIT Work. Pap. Linguist. 11, 228–240. doi: 10.1075/scl.78.09loz Frontiers in Psychology | www.frontiersin.org 12 January 2022 | Volume 12 | Article 794587 Margaza and Gavarró Subjects in L2 Spanish Lozano, C. (2018). The development of anaphora resolution at the syntax-discourse Sorace, A., and Serratrice, L. (2009). Internal and external interfaces in bilingual interface: pronominal subjects in Greek learners of Spanish. J. Psycholinguist. language development: beyond structural overlap. Int. J. Biling. 13, 195–210. Res. 47, 411–430. doi: 10.1007/s10936-017-9541-8 doi: 10.1177/1367006909339810 Montrul, S. (2005). On knowledge and development of unaccusativity in Sorace, A., Serratrice, L., Filiaci, F., and Baldo, M. (2009). Discourse Spanish L2 acquisition. Linguistics 43, 1153–1190. doi: 10.1515/ling.2005.43.6. conditions on subject pronoun realization: testing the linguistic intuitions of 1153 older bilingual children. Lingua 119, 460–477. doi: 10.1016/j.lingua.2008.09. Moscatelli, A., Mezzetti, M., and Lacquaniti, F. (2012). Modeling psychophysical 008 data at the population-level: the generalized linear mixed model. J. Vis. 12:26. Tsimpli, I., and Sorace, A. (2006). “Differentiating interfaces: L2 performance doi: 10.1167/12.11.26 in syntax-semantics and syntax-discourse phenomena,” in Proceedings of the Philippaki-Warburton, I. (1987). The theory of empty categories and the pro- 30th Annual Boston University Conference on Language Development, eds drop parameter in Modern Greek. J. Linguist. 23, 289–318. doi: 10.1017/ D. Bamman, T. Magnitskaia, and C. Zaller (Somerville, MA: Cascadilla s0022226700011282 Proceedings Project), 653–664. Philippaki-Warburton, I. (1989). “Subject in English and Greek,” in Proceedings of White, L. (2011a). Second language acquisition at the interfaces. Lingua 121, the 3rd Symposium on the Description and/or Comparison of English and Greek 577–590. doi: 10.1016/j.lingua.2010.05.005 3, (Thessaloniki: Aristotle University), 11–32. doi: 10.26262/istal.v3i0.7161 White, L. (2011b). The interface hypothesis: how far does it extend? Linguist. Rothman, J. (2009). Pragmatic deficits with syntactic consequences? L2 pronominal Approaches Biling. 1, 108–110. doi: 10.1075/lab.1.1.19whi subjects and the syntax-pragmatics interface. J. Pragmat. 41, 951–973. doi: 10.1016/j.pragma.2008.07.007 Conflict of Interest: The authors declare that the research was conducted in the Roussou, A., and Tsimpli, I. M. (2006). On Greek VSO again! J. Linguist. 42, absence of any commercial or financial relationships that could be construed as a 317–354. doi: 10.1017/S0022226706003914 potential conflict of interest. Serratrice, L., Sorace, A., Filiaci, F., and Baldo, M. (2009). Bilingual children’s sensitivity to specificity and genericity: evidence from metalinguistic awareness. Publisher’s Note: All claims expressed in this article are solely those of the authors Bilingualism 12, 239–257. and do not necessarily represent those of their affiliated organizations, or those of Slabakova, R. (2009). “What is easy and what is hard to acquire in a Second the publisher, the editors and the reviewers. Any product that may be evaluated in Language,” in Proceedings of the 10th Generative Approaches to Second Language this article, or claim that may be made by its manufacturer, is not guaranteed or Acquisition Conference (GASLA 2009), eds M. Bowles, T. Ionin, S. Montrul, and endorsed by the publisher. A. Tremblay (Somerville, MA: Cascadilla Proceedings Project), 280–294. Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguist. Copyright © 2022 Margaza and Gavarró. This is an open-access article distributed Approaches Biling. 1, 1–33. doi: 10.1075/lab.1.1.01sor under the terms of the Creative Commons Attribution License (CC BY). The Sorace, A. (2012). Pinning down the concept of interface in bilingual development. use, distribution or reproduction in other forums is permitted, provided the A reply to peer commentaries. Linguist. Approaches Biling. 2, 209–217. doi: original author(s) and the copyright owner(s) are credited and that the original 10.1075/lab.2.2.04sor publication in this journal is cited, in accordance with accepted academic practice. Sorace, A., and Filiaci, F. (2006). Anaphora resolution in near-native speakers of No use, distribution or reproduction is permitted which does not comply with Italian. Second Lang. Res. 22, 339–368. doi: 10.1191/0267658306sr271oa these terms. Frontiers in Psychology | www.frontiersin.org 13 January 2022 | Volume 12 | Article 794587