Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Cross-Linguistic Influence between Aymara Evidentiality and Spanish Aspect in Bilingual Heritage Aymara-Spanish Children in the North of Chile

Cross-Linguistic Influence between Aymara Evidentiality and Spanish Aspect in Bilingual Heritage... 1IntroductionThe Aymara people in northern Chile are early bilinguals who acquire both the ancestral and socially dominant Spanish language at an early age. The sociopolitical status of the Aymara language (Southern Aymara, ISO 639–3: ayc, with approximately 20,000 speakers in northern Chile and classified as vulnerable by UNESCO, n.d) within a dominant Spanish country influences the language acquisition process resulting in Aymara in northern Chile being heritage speakers.Therefore, Aymara people in Chile have low proficiency levels when they speak Aymara (Espinosa, 2009; Fernandez, 2014; Espinoza, 2019), triggering the mainstream idea that Aymara people are Spanish monolingual with a strong indigenous identity (Ticona and Markovits, 2023). The previous generalization is reflected in the National Intercultural Language Program (MINEDUC, 2019) that aims to develop second-language Aymara courses, ignoring the initial exposure to the heritage languages (HL), which can be advantageous in revitalizing the language. However, Spanish literacy courses are designed for monolingual Spanish speakers, not considering the influence of the HL in the Spanish acquisition of the Aymara community members.The current study explores two side effects of Aymara evidentiality and Spanish Aspect and demonstrates that Aymara speakers in northern Chile are bilingual speakers who engage in cross-language effects. Thus, evidence of the cross-linguistic effect between Aymara and Spanish (and vice versa) will provide relevant information and an appropriate intercultural curriculum for Aymara children in northern Chile.2Theoretical Framework2.1Cross-Linguistic Influence (CLI) and the Functional Convergence HypothesisThe CLI concept emerged to explain different bilingual outcomes that the Autonomous Development Hypothesis, according to which bilingual children develop separate grammars at a very early stage, could not address (Meisel, 2001). Studies in this area refer to CLI as instances in which there is evidence of the effect of one language on the other generally at the morphosyntactic level (Serratrice, 2013). Theories like the Interface Hypothesis (Sorace and Filiaci, 2006; Sorace and Serratrice, 2009; Sorace, 2011) have opened a line of study to understand the influence of one language on another in a bilingual population. For instance, Sánchez et al. (2023) examine whether Spanish exposure (among other variables) affected pragmatic factors that guide The null subject in Spanish (a grammatical structure that lies on the syntactic-pragmatic interface) in bilingual heritage Spanish-English children. The authors evidenced a positive relationship between increased interaction at home and accurate comprehension of pragmatic conditions that license null subjects, and in turn, the results did not show the influence of Spanish on overt subject patterns in English, the dominant language.Building on this understanding of CLI, Sánchez et al. (2010) investigated cross-linguistic influence in a specific context: indigenous heritage languages in contact with Spanish. The author examined first-language morphology properties in the second-language grammars (residual transfer morphology) and cross-linguistic influence on the syntax–pragmatics interface in L1 speakers of Shipibo with different levels of formal instruction in Spanish as a second language. The authors investigated the impact on languague contact on subject-agreement morphology and the null subject properties in both languages. Shipibo has no verbal morphology on the verb and licenses null subject only in the third person, unlike Spanish, which is a null subject language with rich person morphology on the verb. The authors found that bilingual speakers produced a non-standard distribution of the null subject in Spanish, indicating the attainment of a complex syntax-discursive interface. However, the results did not reveal morphological residual transfer from Shipibo in Spanish subject-verb agreement. Previous studies partially supported the claims of the Interface Hypothesis since results showed that instability at the syntax–pragmatic interface persists only under certain conditions. Therefore, other conditions should be addressed in explaining the cross-linguistic effects of language contact (Odlin, 1989).Sánchez (2003) examined syntactic and morphosyntactic changes in the bilingual Quechua-Spanish discourse of children in Peru. She analyzed CLI in acquiring functional categories, such as the representation of the direct object and word order in both systems. Quechua has an obligatory accusative case morpheme ‑ta on the direct object, and the same morpheme can license null nouns in the direct object position. In Spanish, the definite determiners also license null nouns in certain semantic and pragmatic conditions. Sánchez revealed the absence of accusative marking on direct objects in Quechua and the presence of a definite determiner marking null nouns in a non-canonical position in Spanish. These partial similarities between the Spanish definite determiner and the Quechua accusative marker ‑ta indicate the grammatical changes in both languages, allowing the emergence of new structures, thus demonstrating that partial similarities between languages in contact can undergo CLI. These findings led the author to elaborate on the Functional Convergence Hypothesis (2004), which states that different functional features present in a language contact situation can be subject to CLI. The FCH initially stated that convergence, the structural similarities between two linguistic systems in a contact situation, occurs when functional features not present in one of the languages are mapped onto features of another language. Sánchez (2004) modified her proposal, arguing that convergence was favored when the matrix of features associated with a functional category was partially divergent. She tested the hypothesis examining CLI between Spanish Aspect and Quechua evidentiality since, in Quechua, past tense is closely tied to evidentiality, whereas in Spanish, it is linked to aspect. A group of Quechua-Spanish bilingual children performed a retelling story task in the target languages. The results showed a high frequency of the use of the Quechua reportative morpheme ‑sqa to convey an indirect source of information (in other words, that a third person reported the action) and a shallow frequency of the attested form ‑rqa that could convey Spanish aspect. The evidence supporting convergence in evidentiality features within the bilingual Spanish narratives was more compelling. Bilingual participants, on average, used imperfective morphology during a task involving the transmission of hearsay information. This suggests that bilinguals associate imperfective morphology with evidentiality features.This paper seeks to build on the foundational work of Sánchez (2004), which identified significant cross-linguistic influence between Quechua evidentiality and Spanish aspect in bilingual Quechua-Spanish children. By extending these findings, the present study examines a similar phenomenon in Aymara-Spanish bilingualism, focusing specifically on the interaction between Aymara evidentiality and the Spanish aspect. In doing so, this research broadens the scope of cross-linguistic influence to include another indigenous language. It contributes to a deeper understanding of how evidentiality systems in indigenous languages shape aspectual interpretations in bilingual contexts.2.2Evidentiality in Andean SpanishEvidentiality is a linguistic category in which the primary meaning of a statement is the source of information (Aikhenvald, 2018). Evidentiality specifies the origin of an utterance content. The origin refers to the degree of evidence between the speaker and the event. The degree of evidence ranges from testimonial evidence (e.g., vision, hearing, smell, etc.) to non-testimonial evidence, such as a second source of information, an inference based on visual traces, reasoning, and general knowledge (Aikhenvald, 2018: 1).Languages differ in the way they convey the source of information. Spanish does not have an obligatory grammatical evidentiality system. However, there are several lexical expressions to mark the source of information, such as vi lo que ella hizo, “I saw what she did” (direct evidence), or me dijeron lo que ella hizo, “they told me what she did”, (indirect evidence). Sometimes the evidential concepts are encoded in mental or perception verbs (e.g., saber, know; creer, guess; escuhar, hear).Furthermore, Sánchez (2004) analyzed different interpretations of Spanish imperfective morphology. She argues that the imperfect tense is used to report events occurring outside of consciousness, such as dreams. The author offered the following example:(1)En mi suenoyomeganabaLa loteriaIn my dreamICL1SGwinPAST IMPERFthe lotery‘In my dream, I won the lottery’.In the case of Andean Spanish, several authors have shown a systematic use of evidentiality features in this variety of Spanish. Fernandez (2014) examined dialectal convergence in the Andean Spanish divergence in contact with Chilean Spanish in the north of Chile. The author presented a description of the features of Andean Spanish that differed from Chilean Spanish, one of which was the use of pluperfect to express non-experienced past tense action. The author provided the following example:(2)Medijeron que la genteloshabía discrimadoaLos chilenos1. 1S.DATtold that the people3.PL.AChave-IMP.3.S discriminatedDOMthe Chileans‘They told me that people had discriminated against Chileans.’Escobar (2011) reviewed the features of Andean Spanish that differentiated this dialect from other varieties of Spanish and described how some bilingual speakers in Bolivia and Peru, where numerous varieties of Quechua are commonplace, appeared to use the pluperfect for reported information and the present perfect to mark first-hand information. Bustamante (1991) found that the present perfect had a reportative function in Ecuadorian Spanish. These findings suggest that the evidential systems of indigenous languages in the Andean region of South America influence the restructuring of the perfective aspect in Spanish in multiple ways.To summarize, the expression of evidentiality in Spanish is mainly manifested through lexical expressions such as verbs, adverbs, modal verbs, and conjunctive locutions. Nevertheless, a vast amount of literature about Andean Spanish has found perfective/imperfective morphology to be a mark of evidentiality. Although there is no consensus about the contrast between the present perfect and the pluperfect to mark direct and indirect evidentiality, these studies suggest that functional features such as aspect and evidentiality are subject to permeability in a language contact situation. The current investigation aims to contribute to previous studies, analyzing the impact if Amymara on a less-studied variety, the case of Chilean Spanish spoken in the north of the country, examining the comprehension and production of the present perfect or the pluperfect to mark the source of information in the bilingual population.2.3Aymara EvidentialityThe Aymara language has a diverse evidential system, where different evidential categories are expressed through verbal inflections within the distribution of tenses.1 The Aymara understanding of time is distinct from European language-based models. The timeline in Western languages places the past behind the speech act (S) and the future in front of it. However, Aymara conceptualizes time so that the future is behind the speech act, and the past is in front (Miracle and Yapita Moya, 1975; Núñez and Sweetser, 2006). Then, Aymara conceptualizes time in terms of future and non-future actions. The non-future is divided into remote and present/recent past actions. The former refers to an action that occurred before the statement was completed (Ticona, 2007).In addition to tense, Aymara speakers must encode the source of information for events occurring in the past. Hardman et al. (2001) identify a distinction between personal knowledge tenses, associated with a non-remote past, and non-personal knowledge tenses, which include two remote past categories: remoto cercano (‘near remote past’) and remoto lejano (‘far remote past’). According to the authors, these tenses differ in their degree of remoteness, with remoto cercano referring to events that are relatively recent and remoto lejano indicating events that occurred further in the past. The non-remote past encodes a direct source of information, whereas the remote past tenses convey an indirect source. However, not all verbal inflections mark these distinctions consistently, suggesting a more nuanced evidential system.Figure 1Aymara tense paradigmFigure 1: Aymara tense paradigmSimilarly, Cerrón-Palomino (1994) describes two past tenses: pasado experimentado (‘experienced past’), overtly marked by the suffix ‑yä, and pasado no experimentado (‘non-experienced past’), indicated by the verbal morpheme ‑tay. According to him, this distinction is evidential: ‑tay encodes a non-experienced past, indicating an indirect source of information, while ‑yä marks a recent past (non-remote) associated with direct personal knowledge of the action. Ticona (2007) adopts the terms testimonial and non-testimonial to distinguish between recent personal knowledge and remote non-personal knowledge, respectively, using ‑yä and ‑tay to mark direct and indirect information sources.2 The following examples illustrate the verbal paradigm of evidentiality in the recent and remote past tenses:(3)Luysumanzanamanq’-ä-naLuisappleeatREC/DE3>3‘Luis ate an apple.’(4)Sapa uruthuqtayna (Ticona, 2007:42)Every daydanceREM/IE 3>33‘He danced every day!’ (the speaker is not witness of the action)2.4Spanish AspectAspect is a linguistic category that expresses different temporal properties of past events. Like evidentiality in Aymara, the Spanish aspect is overtly manifested in different past morphemes. Comrie (1976) described the grammatical aspect, or viewpoint aspect, as a distinction between perfectivity and imperfectivity. He described the perfective aspect as ‘looking at the event from the outside’ (Comrie, 1976: 17) and the imperfective aspect as ‘looking at the event from the inside’; (Comrie, 1976: 17).Spanish contains two past tense verbal paradigms, which align with the perfective/imperfective distinction. For example,(5)Jugu-éfútbolen la universidadPlayPRF.1.Ssoccerin the University‘I played soccer in college.’More recently, Bardovi-Harlig (1994; 1998; 2000) proposed the discourse hypothesis, which claims that aspects are determined by narrative structure. The author described different perspectives of the events within the narrative structure and distinguished between foregrounding and backgrounding in narratives by referring to the imperfective-perfective distinction. A background feature may have three interpretations: (1) the background verb aspect indicates that the verb occurred before the focal event; (2) the aspect of the verb indicates a prediction about the outcome of an event; or (3) the aspect could be interpreted as an evaluation of an action reported in the foreground. The foreground verb aspect, on the other hand, is in the event timeline. The properties of events in the past and their relation to aspect can be illustrated using Reichenbach’s (1947) speech parameter. The author described three different parameters of speech: a speech act (S), an event (E), and a reference point (R). In context, background and foreground clauses in the past take place before the speech act, but the event and reference points differ. Figure 1 explores this relationship.Figure 2Spanish aspect and the reference pointFigure 2: Spanish aspect and the reference pointBoth background and foreground events are located before the moment speech act (S). However, the chronological relationship between the point of the event (E) and the reference point (R) is different in the timeline. In the imperfective aspect, the reference point is before the event, unlike the perfective aspect, in which the event and the reference points are simultaneous.Here, we find a potential for convergence between the two systems, which could lead to CLI since Spanish and Aymara have two verbal morphology representations in the past tense. Such findings would be consistent with those of Sánchez (2004), who found that Quechua-Spanish bilingual children transferred evidential properties to the Spanish aspectual system and aspectual properties to the Quechua evidential system, as proposed in the present study. In addition, the distance between R and E in background events is greater than in foreground events. Similarly, the direct evidential Aymara morpheme ‑yä conveys a shorter distance between the reference point and the event, unlike the indirect morpheme ‑tay, which encodes a greater distance; therefore, the comprehension and production of both structures require the speaker to understand not only the morphosyntactic features of the structures but, also the pragmatic features linked to the reference point within the timeline.3The Present StudyThe present investigation examined the distribution of the two past verbal morphemes in Aymara, spoken in northern Chile, exploring whether the evidentiality system exhibits the value of Spanish aspects in the past narrative discourses of Aymara-Spanish bilingual children. The current research aims to demonstrate that bilingual speakers assign an evidential value when they produce and comprehend past narration in Spanish. Previous studies have demonstrated a non-standard distribution of present perfect and pluperfect in Andean Spanish. Therefore, the present work expects to find evidential marks in a high frequency of compound past forms in Aymara-Spanish bilingual speakers, which differ from the standard distribution of the simple and imperfect past in the Spanish Chilean variety (de Alba, 1998; Fernandez, 2014).I will test the Functional Convergence Hypothesis by Sánchez (2004) which states that convergence is favored when the matrix of features associated with a functional category is partially divergent. In the current study, the functional category is tense, and the divergent features are aspect (namely perfective and imperfective morphology) and verbal inflectional suffixes that mark evidentiality. Therefore, the current work proposes four research questions:Do heritage bilingual Aymara-Spanish children produce aspect morphemes with an evidential value in their past narrative discourse in Spanish?Do heritage bilingual Aymara – Spanish children comprehend aspect morphemes with evidential value in their past narrative discourse in Spanish?In response to these questions, it is hypothesized that evidentiality converges with aspect features in the Spanish discourse of Aymara-Heritage speakers in a contact situation.Do heritage bilingual Aymara-Spanish children produce past evidential morphemes with aspect value in their past narrative discourses in Aymara?Do heritage bilingual Aymara-Spanish children produce past evidential morphemes with aspect value in their past narrative discourses in Aymara?In response to these questions, it is hypothesized that aspect converges with evidentiality features in the Aymara discourse of child Aymara-heritage speakers in a contact situation.3.1MethodologyThe methodology included three groups of participation. The experimental group contained 21 heritage Aymara-Spanish bilingual children (HS) aged four to 12 years. These children live in the region of Tarapacá in the northern part of Chile, specifically in the community of Sibaya, one of the rural regions where there is a coexistence between the traditional Aymara culture and the Chilean culture (Goldstein, 2015). They used Aymara and Spanish simultaneously during their first two or three years. At the age of 4, they start pre-k schooling instruction in Spanish. From 6th to 8th grade, they have Aymara as a second language for three hours per week. After 8th grade, they must move to the biggest nearest town (Huara) to complete high school. When I conducted my research (2021), there were no Aymara L2 programs in Huara High School. The baseline of the heritage children was 15 heritage Aymara-Spanish bilingual adults (AD). The adults were the parents of the heritage children and lived in the same rural community. All the adult participants had migrated from Bolivia to Chile. In addition, 19 heritage Aymara-culture Spanish-dominant children (SP) of the same age as the experimental group participated in the study. The monolingual children lived in the capital of the Tarapacá region, Iquique, which is three hours away from Sibaya. They had been exposed to the traditional Aymara culture and knew some lexical items and phrases in Aymara. However, they did not understand or speak Aymara. Including the monolingual as a comparison group allowed us to understand the effects of the heritage language in the Spanish of the Aymara heritage speaker children, testing whether the two children groups behaved similarly (or not) when they needed to comprehend and produce Spanish sentences in the past. Table 1 includes relevant sociolinguistic features of the children’s participants for the current study.Table 1The demographic of the heritage (HS) and monolingual (SP) childrenGroupNMean ageFirst languageSecond languageDominant languageHS218.5AymaraSpanishSpanishSP197.9Spanishn/aSpanishThe following tests and tools were used:1. Modified sociolinguistic background questionnaire (Fernandez, 2014): The children’s parents completed a language questionnaire in Spanish, answering to questions about the children’s and family members’ activation, frequency, and exposure to Spanish and Aymara.2. A screening bilingual Spanish-Aymara test based on the textbook Para compartir voces: Texto de consulta para profesores de castellano como segunda lengua (Sánchez, 2014). The test was used to assess the Aymara and Spanish levels of the heritage children and adults. The maximum score was 43, and the minimum Aymara score to perform the Aymara version of the tasks was 15 points or intermediate I level.3. Experimental Items and Tasks: Spanish Production Tasks. To investigate Research Question 1 – whether Aymara heritage children mark Spanish past narratives with Aymara evidentiality – the following tasks were employed:3a) Elicited production task (EPT)This experiment attempted to elicit brief oral narratives in the Spanish past to examine whether Aymara-Spanish heritage bilingual heritage speakers produced the present perfect as a mark of direct evidentiality (Aymara experienced past tense). The previous assumption was based on Escobar (2011) and Rojas-Sosa (2008), whose findings showed evidence of a higher-than-standard distribution of the present perfect in narrative clauses in Andean Spanish speakers.To examine whether bilingual children used the present perfect instead of the simple past to narrate actions they had witnessed, the Principal investigator (PI) conducted ‘the see condition’ EPT. The interviewer told the children that she would show them pictures of Luis, a teenage boy. The PI then asked the children to tell her what Luis did yesterday. The children were shown eight pictures illustrating Luis’ actions from the previous day. The interviewer then asked the children to tell her what Luis did. For example:Picture 1: The child was shown an image of Luis waking up. The interviewer asked, ‘Can you tell me what Luis did yesterday?’Expected response: Luis ha despertado (‘Luis has woken up’).Unexpected response: Luis despertó (‘Luis woke up’).Picture 2: The child was shown an image of Luis buying tomatoes. The interviewer asked, ‘Can you tell me what Luis did yesterday?’Expected response: Luis ha comprado tomates (‘Luis has bought tomatoes’).Unexpected response: Luis compró tomates (‘Luis bought tomatoes’).If participants used the present perfect instead of the simple past tense preterit, I assumed they used it as a mark of direct evidentiality. On the other hand, if there were a preference for using the simple past tense, the Aymara evidential system would not have affected their production.3b) Retelling task.This experiment attempted to elicit past narratives in Spanish containing aspectual morphology to assess whether Aymara-Spanish bilingual children produced the pluperfect as a mark of indirect evidentiality (Aymara experienced past tense). The assumption was based on Fernandez (2014), who showed the use of the pluperfect to express non-experienced past tense actions in Andean Spanish. The PI presented a retelling story task to examine whether the bilingual children used the pluperfect, a periphrastic construction, instead of the simple past to describe an action in the past that they had not witnessed. The children listened to a Spanish recorded audio story, ‘La música en las montañas,’ after which the PI asked the children to tell her the story. For example:The child listened to (input): Todos los días Chuqu le pedía al yatiri que le mostrara el camino (‘every day Chuku asked the wise man to show the way’).Expected response (output): Todos los días Chuqu le había pedido al yatiri que le mostrara el camino (‘every day Chuku had asked the wise man to show the way’).4. Production tasks – Aymara versionThe current experiments were designed to answer research question 3 of whether Aymara heritage children produce Aymara past narratives marking the Spanish aspect. The tasks were the following:4a) EPT. This experiment attempted to elicit the production of Aymara evidential past morphemes to examine whether Aymara-Spanish bilingual children produced the non-experienced past morpheme (‑tay) as a marker of imperfective actions in the past. To examine whether the bilingual children used the non-experienced past morpheme with the imperfective value in Aymara, eight pictures of Luis as a baby appeared on the laptop computer, and the PI told the children that they knew Luis. They had seen what Luis did when he was young. Each child was administered eight trials. The interviewer then asked the children to describe what Luis did as a child. For example:Picture 3: The child was shown an image of Luis with a ball. The interviewer then told the child that he/she witnessed what Luis did as a child and asked, ‘Can you tell me what Luis did when he was a child?’ The expected answers were:– Expected response:Luis piluta- mpianata-tay-naLuis ball-INSTPLAYREM/IE3> 3‘Luis played with the ball – someone told me.’– Unexpected response:Luis piluta- mpianata- tä-naLuis ball-INSTPLAYREC/DE3> 3‘Luis played with the ball – I was a witness.’Picture 4: The child was shown an image of Luis reading a book. The interviewer then told the child that he/she witnessed what Luis did as a child and asked, ‘Can you tell me what Luis did when he was a child?’ The expected answers were:– Expected response:Luis pankliyi-tay-naLuis bookreadREM/IE3>3‘Luis read the book – someone told me.’– Unexpected response:Luis pankliyï-naLuis bookread – REC/DE-3>3‘Luis read the book – I was a witness.’Suppose there was a preference for the use of -tay. In that case, it is plausible that the children produced the non-experienced past morpheme with the imperfective aspectual value since the children had witnessed the action that the character had performed habitually in the past. On the other hand, if there was a preference for the use of -yä, it was likely that the participants had not mapped an aspectual value onto any evidential marker in the past narration. In other words, the action in the past was modulated by the Aymara evidentiality grammar structure.4b) Retelling task. The children listened to a recorded version of the story ‘La música en las montañas’ in Aymara. After they had listened, the children were asked to tell the story to the PI. For example:– the child listened to (input):Chukumayantiqinaphusa-ñayatt ´ataynaChukuone more timequenaPLAYNMLZTRYREM/IE3>3‘One more time Chuku had tried to play the quena, someone told me.’– Expected response (output):Chukumayantiqinaphusa-ñayattänaChukuone more timequenaPLAYNMLZTRYREC/DE3>3‘One more time Chuku had tried to play the quena, I witnessed the action.’If there was a preference for ‑yä to describe a discrete event in the past, it was plausible that the children produced the experienced past morpheme with a perfective aspectual value since they did not witness the character’s action in the past. On the other hand, if there were a preference for the use of ‑tay in these contexts, the logical conclusion would be that Spanish perfectivity was not mapped onto the evidential morpheme in Aymara. In other words, the action in the past was modulated by the source of information.5. Comprehension tasks – Spanish versionThe current experiments were designed to answer research question 2 of whether Aymara heritage children comprehend Spanish past narratives marking Aymara evidentiality. The tasks were the following:5a) Forced-choice task (FCT) – see condition. The task aimed to examine whether the children assigned evidential value to aspectual morphemes in Spanish, that is, if they selected the present perfect for assigning the experienced past tense. The PI introduced two characters (the girl and the boy) and told the children that they were Luis’ friends and that they had seen what Luis had done the previous day. The PI told the child to listen to what the girl and the boy said and asked the participant to select which character sounded best. Each child was administered eight trials with present perfect and simple past audio recordings.For example:Audio 1: Luis tomó café (‘Luis had coffee’).Audio 2: Luis ha tomado café (‘Luis has had coffee’).If there was a preference for selecting the present perfect instead of the simple past, I assumed that the Aymara-Spanish bilingual participants were assigning the present perfect tense as a mark of evidentiality. On the other hand, if there was a preference for selecting the simple past, their selection of evidential morphology in Aymara was modulated by the Spanish aspect category.5b) FCT – hearsay condition. The task aimed to examine whether children assigned evidential value to aspect morphemes in Spanish, namely the pluperfect, to assign the non-experienced past tense. The PI told the children that the same characters in the previous task were told what Luis did when he was a child. The characters then described what they had heard about Luis, and the participants needed to decide which one sounded best. Each child was administered eight experimental trials with pluperfect and imperfect past audio recordings.For example:Audio 1: Luis jugaba a la pelota (‘Luis played the ball’).Audio 2: Luis había jugado a la pelota (‘Luis had played the ball’).If there was a preference for using the pluperfect instead of the imperfect past, it could be assumed that the participants were mapping the value of indirect evidentiality onto the pluperfect. On the other hand, if there was a preference for using the imperfect past, their production was only modulated by the Spanish aspect category.6. Comprehension tasks – Aymara versionThe current experiments were designed to answer research question 4 of whether Aymara heritage children comprehend Aymara past narratives marking Spanish Aspect. The tasks were the following:6a) FCT: See condition. This experiment examined whether Aymara-Spanish bilingual children assigned imperfect aspects to the non-experienced past morpheme ‑tay. The PI told the child that the same two characters were now seeing what Luis did as a child. Then, two audios popped up. After each character’s utterances, the PI asked who said it better. Each child was administered eight experimental trials with ‑tay and ‑yä verbal forms.For example:Audio 1: Luisa mä panq liyitayna (‘Luis read a book, someone told me’).Audio 2: Luis mä panq liyi:na (‘Luis read a book, I witnessed’).If there was a preference for the ‑tay option, I assumed the children were assigned an imperfect aspect value to the non-experienced past morpheme since the speaker witnessed the character’s action. On the other hand, if there was a preference for using ‑yä, I inferred that the children had not assigned any value to aspect features in the past narration. In other words, the action in the past was modulated by the source of information.6b) FCT: Hearsay condition. To examine whether the bilingual children assigned a perfective value to the experienced past morpheme – yä in Aymara, the PI told the children that the same characters had listened to what Luis did yesterday. The characters then described the picture, and the children needed to select who said it better. For example:Audio 1: Luis manq’äna (‘Luis ate, I witnessed’).Audio 2: Luis manq’tayna (‘Luis ate, somenone told me’).If there was a preference for selecting the past experienced morpheme ‑yä, I assumed that the children were assigned a perfect aspect value to the experienced past morpheme since the child did not witness the action in the narration in the past. On the other hand, if there was a preference for the use of ‑tay, I inferred that they had not assigned any value to the aspect feature in the past narration. In other words, the action in the past was modulated by evidentiality features.3.2ProcedureThe experimental phase was conducted in a single session, following this sequence: first, participants completed a 15-minute bilingual language screening test in Aymara and Spanish. This was followed by an Aymara Elicited Production Task (EPT) lasting approximately five minutes, and then a Spanish EPT, also taking around five minutes. Next, participants engaged in Spanish Forced-Choice Tasks (FCT s) for about five minutes, succeeded by Aymara FCT s of similar duration. The session continued with an Aymara retelling task (approximately five minutes) and concluded with a Spanish retelling task (also around five minutes).4ResultsThe analysis presents the results of the tasks performed by the 56. The participants who were divided into three groups, namely the heritage children group (HS), the Spanish-dominant-cultural heritage children group (SP), and the Aymara Adults heritage speakers (AD) group. Eight experiments were conducted: two Aymara forced-choice tasks (FCT), one Aymara elicited production task (EPT) and one Aymara retelling task, and two Spanish FCT s, one Spanish EPT and one Spanish retelling task. All groups performed the Spanish version of the exercises. However, only the Aymara-Spanish bilingual participants performed the Aymara version of the tasks since a minimum score of 15 out of 43 was required in the Aymara screening test to perform the Aymara exercises. The statistical analysis of the collected data was conducted using the R programming language.4.1Aymara Comprehension Tasks: Forced-Choice Tasks (FCT)The current analysis attempted to answer research question 4 of whether the HS group comprehended Aymara evidential morphemes with a Spanish aspect value. Each participant (21 HS; 15 AD) completed two FCT comprehension tests: the see condition and the hearsay condition. The participants completed eight experimental items in each test, distributed randomly by PPT. Task responses were coded as 1 for the expected response and 0 for the unexpected response. Figure 3 shows each group’s proportion of expected responses in each task by group. The y-axis represents the total number of expected or ‘1’ responses.Figure 3Proportion of expected responses across bilingual participants in the Aymara comprehension taskFigure 3: Proportion of expected responses across bilingual participants in the Aymara comprehension taskThe results for the heritage children showed that in the see condition (right block), the mean was above chance (X= 0.53, SD = 0.50), demonstrating that more than 50 % of the participants selected the indirect morpheme ‑tay to describe a remote action that they were witnessing in the past, thus suggesting that the participants assigned an imperfect aspect value to the indirect evidential morpheme. In the hearsay condition task, less than 50 % of the heritage children selected the morpheme ‑yä to describe a recent past action secondhand information (X= 0.45, SD = 0.50), showing that more than 50 % of the participants selected the unexpected response (the selection of the indirect morpheme ‑tay), suggesting that children did not assign an aspect value to the direct morpheme ‑yä. Regarding adult heritage speakers, the frequency of ‘1’ responses was higher than in see (X= 0.88, SD = 0.32) and hear (X= 0.51, SD = 0.50) conditions compared to the HS group. An explanation of mean differences by group could be that Spanish influence has a greater impact on how heritage adults comprehend the source of information than children (see discussion).4.2Spanish Comprehension Tasks: Forced-Choice Tasks (FCT)The current analysis attempted to answer research question 2 of whether the HS group comprehended Spanish aspect morphemes with an Aymara evidential value. Specifically, the current task aimed to identify whether the participants selected the present perfect verbal form to describe an action they had witnessed in the past (see condition), as well as to identify whether the participants selected the pluperfect form to describe pictures of secondhand source of information (hearsay condition). Each participant completed two comprehension tasks, namely the see and hearsay conditions. The participants completed eight experimental items in each test, distributed randomly by PPT. Task responses were coded as 1 for the expected response and 0 for the unexpected response. Figure 4 shows each group’s proportion of expected responses in each task by group (21 HS, 15 AD, and 19 Spanish dominant children with a strong Aymara identity or SP). The y-axis represents the total number of expected or ‘1’ responses.Figure 4Proportion of expected Spanish comprehension task responses groupsFigure 4: Proportion of expected Spanish comprehension task responses groupsThe HS group performed differently across conditions. Heritage children’s score was above chance in the see condition (X=0,59; SD= 0,49), showing that more than 50 % of the heritage children selected the expected response. In other words, they preferred the present perfect rather than the simple past to describe an action they witnessed. On the contrary, they performed on chance in hear conditions (X= 0.5; SD= 0.50), showing no tendency for any choices (imperfect or pluperfect). The results could be due to a task effect (see discussion). Adults had fewer expected responses in the see condition (X= 0.4, SD= 0.49) than children. Adults tended to select the use of simple past tense to express a witnessed action in a recent past. In addition, AD expected response in the hear condition was higher (X= 0.57; SD= 0.49) than in the HS group, showing a tendency to select the pluperfect to describe a reported action in the past. Differences in conditions across heritage groups are discussed in the next section (see discussion). Finally, the SP group’s performance was below chance in see (X= 0.32; SD: 0.46) and hearsay (X= 0.30; SD: 0.46) tasks, demonstrating that Spanish-dominant children tended to select the simple past instead of the present perfect and the imperfect past instead of the pluperfect in their responses to describe actions in the past, as it was expected.4.3CLI in Production Tasks4.3.1Aymara Elicited Production Task (EPT)The current analysis attempted to answer whether the heritage children (HS) produced Aymara evidential morphemes with a Spanish aspect value. Specifically, the current task attempted to identify whether the participants elicited the non-direct morpheme ‑tay with an imperfect aspect value. Each participant completed an EPT task in which they witnessed the actions that happened long ago (see condition). The participants completed eight experimental items that were randomly distributed by PPT. Only one of the 21 HS participants and five of the 19 AD participants performed the task. The low performance was because that participant said they could not speak Aymara. They said they did not know how to describe the pictures task orally. Due to the low number of participants, the current sections provide the frequency of evidential forms elicited from the participants in each group. The number of verbal tokens was counted by group. Then, the number of evidential morphemes was counted to examine the frequency. Participants did not produce any ‑yä morpheme. Participants produced verbal forms with ‑tay morpheme or verbal forms with ‑ki morpheme. ‑ki morpheme is an aspectual- long-lasting morpheme formed by two suffixes: the suffix of incompletive aspect ‑ka and the suffix ‑i, which indicates the third person of the simple tense (Cerron-Palomino, 1994).Table 2 shows the frequency of the ‑tay and ‑ki morphemes by group in the see-condition EPT task.The frequency table shows that the most produced morpheme was ‑tay, followed by ‑ki. It is helpful to remember that the participants were told they had witnessed a character’s actions long ago. Table 3 provides examples of the expected responses for each experimental item and the participants’ most frequent responses.Table 3 shows the data that most participants produced verbal forms with indirect evidential morphemes. This behavior differed in item 3 and item 6, in which the most frequent verbal type did not attach the non-experienced past morpheme ‑tay. Conversely, the participants in item 3 and item 6 tended to elicit the aspect-durative morpheme ‑ki. The reason could be a task effect. In other words, images 3 and 6 could depict a progressive action, unlike the rest of the pictures. In addition, using the imperfective morpheme -ki could be a case of attrition of the syntactic evidentiality representation. However, further investigation needs to be done to support the previous claim.Table 2Frequency of evidential morpheme across groupsn ofn tokens (numbern ‑kin ‑tayparticipantsof verbal forms)morphememorphemeHS1826AD532626Table 3Expected responses versus frequent responsesExperimental itemExpected responseMost frequent response1Luis piluta anatatayna‘Luis played with the ball, someone told me.’piluta anatatayna‘with the ball he played, someone told me.’2Luysu wasa isi uchasitayna‘Luis put on cloth from someone else, someone told me.’Disfrasasitayna‘(he) put on a costume, someone told me.’3Luysu mapanka liyitayna‘Luis read a book, someone told me.’Libri liyiki‘Luis is reading.’4Luysu uta luratayna‘Luis built a house, someone told me.’Luis anatayna‘Luis played, someone told me.’5Luysu jarisitayna‘Luis washed himself, someone told me.’Luis wañasitayna‘Luis took a bath, someone told me.’6Luysu mansana maq’atayna‘Luis ate an apple, someone told me.’Manzana katuski‘(he) grabbing the apple’7Luysu mamapampi saratayna‘Luis walked with his mother, someone told me.’Mamallpampi sarkatayna‘He walked with his mother, someone told me.’8Luisu karta qilqatayna‘Luis wrote letters, someone told me.’Karta qilqatayna‘(he) wrote letters, someone told me.’4.3.2Spanish Elicited Production Task (EPT)The current analysis tried to answer whether the HS group produced Spanish aspect morphemes with an Aymara evidential value. Specifically, the current task attempted to identify whether the participants used the present perfect instead of the simple past to describe actions that they had witnessed. Each participant completed an EPT task in which they witnessed the actions they needed to produce (see condition). The participants completed eight experimental items that were randomly distributed randomly by PPT. The expected response was the production of the present perfect. However, participants produced six verbal forms, as shown in Table 4.Figure 5Production of verbal forms across the groupsFigure 5: Production of verbal forms across the groupsTable 4Verbal forms produced by participantsVerbal formExampleImperfect (imperf)Luis leíaimperfect progressive (imPR)Luis estaba leyendoPast periphrastic form other than imperfect progressive (periph)Luis se fue a leerPluperfect (pluper)Luis había leídoPresent perfect (PP)Luis ha leídoPresent progressive (presPR)Luis está leyendoSimple past (simplep)Luis leyóDue to the variability in participant responses, the expected response was any verbal form other than simple past. This allows us to examine a non-standard distribution of verbal production to describe discrete points in the past. Figure 5 shows the frequency of each verbal form produced by the group.The data shows that the imperfect progressive (imPR) had a high frequency within the HS (N=68) and the AD (n=41) groups. Regarding HS, the most frequent form after imPR was the simple past (26), followed by the present progressive (N=9). The expected present perfect form (PP) had a low frequency within the SP corpus (N=6), similar to the AD group (N=2). As was expected, the SP group showed a higher frequency of simple past. The most frequent form produced by the Spanish-dominant children after the simple past was the present progressive (n=18), followed by the past imperfect (n=3).4.4Aymara Retelling Task (RT)The current analysis aimed to answer whether the HS group produced Aymara evidential morphemes with a Spanish aspect value. Specifically, the current task attempted to identify whether the participants elicited the direct morpheme ‑yä with a perfective value in a secondhand narration to describe discrete points in the past. Each participant listened to a recorded Aymara narration and was then asked to retell the story to someone else. Similar to the previous task, only one of the 21 HS participants and only one of the 19 AD participants performed the task. The reason for not performing the task was the same as on EPT. Most of the participants argued that they were not able to retell the story in Aymara. The current sections analyze the most frequent type of evidential morphemes used to describe the story. The input story contained a total of 17 verbs, of which one token was a verbal form attached with a future morpheme (marked in bold), and 16 were verbal forms attached with the non-evidential morpheme ‑tay (underlined) (see supplemental material). No attested morphemes appeared in the story due to the secondhand source of information nature of the instrument.The number of tokens (verbal forms) in both participants was counted. The HS participant produced zero verbal tokens. The HS participant retold the story based on nouns, as the following example illustrates:(10) PI: ¿Le puedes contar la historia a tu hermana?‘Can you tell the story to your sister?’HCH: mmmm, no sé‘Mmmm, I don’t know.’PI: ¿y de que te acuerdas?‘What do you remember about it?’HCH: Jamp’atu, uma, yuqalla, imilla‘Frog, water, boy, girl.’Unlike the HS participant, the AD participant produced nine verbal forms, all of were evidential tokens (the ‑tay morpheme attached to the verbal form). The following example illustrates the corpus of the AH participant.(11)YatiriawisataynaukathakhisasaukaawisataynasasaukatkullakapampisarataynaWise manTo tellREM/ID3>3thatwaytellThatTo tellREM/ID3>3tellthatsister3.posCOMgoREM/ID3>3‘The wise man told about that way, that way he told. Then he went with her sister, he told.’The oral production transcription showed a high frequency of-tay morphemes in the past narration, demonstrating the use of the indirect morpheme to assign the origin of the utterance in hearsay and remote narration in the heritage language. However, the direct morpheme-yä was not used to describe a discrete point during the narration.4.5Spanish Retelling Task (RT)The current analysis attempted to answer whether the HS group produced Spanish aspect morphemes with an Aymara evidential value. Specifically, the current task attempted to identify whether the participants produced the pluperfect form with an indirect evidential value in a secondhand narration. Each participant listened to a recorded Spanish narration. The retelling nature of the task was an ideal context for bilingual children to use reportative past forms. Fifty-five participants performed the task in the same three groups: 21 participants in the HS group, 15 in the AD group, and 19 in the SP group.The input story contained a total of 26 verbs, of which 13 tokens were past imperfect forms (underlined), and 13 were past perfective forms (marked below in bold) (see supplemental material); there were no tokens of present perfect forms. A descriptive analysis of the types of verbal forms produced by the participants is presented in the current section. The forms were as follows:the imperfect past.the simple past.the past imperfect progressive.pluperfect.others. This classification included the simple present (indicative and subjunctive) and the pluperfect subjunctive. No other verbal forms were produced.The number of tokens (verbal forms) distributed in the three groups is shown in Table 5.Table 5Number of verbal forms by groupGroupImperfectSimple pastImperfect past progressivePluperfectOthersHS334941103AD426911210SP5381647The maximum number of verb forms for the three groups was similar. The most frequent verb form was the simple past, demonstrating the consistent use of the simple past to express discrete points in the three groups.The analysis also showed a high frequency of the past imperfective progressive in the HS group (n=41) compared to the AH (n=11) and HCU (n=6) groups. There was a frequent mismatch between input and output among heritage children. The mismatch was using the past progressive form in the output when the input had a simple past form. The following examples show the mismatch between simple past (input) and past imperfect progressive(output) in HS participants.4.6Input Simple Past–Output Imperfect Progressive(14) input: sopló su quena‘(he) blew his flute.’output: estaba soplando la quena‘(he) was blowing his flute.’The previous example shows that bilinguals have imperfective morphology more frequently than monolinguals in the foreground context, thus suggesting that the Aymara-heritage children preferred periphrastic forms to express simple forms. Finally, the frequency of use of the pluperfect form was lower than was expected in the HS (n=10) and AD (n=21) groups; however, the frequency of the pluperfect form in the heritage groups was higher than in the SP group (n=4).5Discussion5.1Cross-Linguistic Influence (CLI) in Comprehension TasksTwo Aymara comprehension FCT s and two Spanish FCT s were conducted to identify whether there was CLI between the Aymara evidentiality and the Spanish aspect (and vice-versa) in the past bilingual discourse of the heritage children. The assumption was based on the Functional Convergence Hypothesis (Sánchez, 2004), which states that cross-linguistic influence occurs when functional features not active in one of the languages become activated in the other when the matrix in both languages is partially divergent. The results for the FCT s partially confirmed this hypothesis.The Aymara FCT see condition task (that depicted a scenario in which two characters witnessed a remote past action) showed that more than 50 % of the participants preferred the ‑tay constructions, thus providing explorative evidence of discourse-oriented aspectual features in the comprehension of the indirect evidence morpheme ‑tay. By contrast, the Aymara FCT hearsay condition (that depicted a scenario in which two characters heard what someone else did in a recent past action) showed that less than 50 % of the participants selected the direct evidential morpheme ‑yä, suggesting that the direct evidential morpheme does not encode a foreground aspect feature. Two explanations can be drawn.On the one hand, in the heritage speakers’ morphology, perceptually salient elements are better maintained and are more readily noticed (Polinsky, 2018). The morpheme ‑yä is phonetically weaker than the indirect morpheme ‑tay, thus resulting in a possibility of attrition; however, it is impossible to confirm a case of attrition based on the current results. Another explanation could be that the verbal morpheme ‑yä tends to overlap with the third-person nominal morpheme ‑na. Therefore, children could comprehend the morpheme ‑yä as a mark of third person singular instead of as a mark of a direct source of information. Following Lightfoot (2010), the input from the baseline generation contributes to the changes in the grammar of heritage speakers. Adult heritage speakers performed similarly to children in the same condition. Children received non-standard input from their parents, which resulted in ambiguity in their mental representations of the target structure. The ambiguity forces heritage speakers to (economically) reanalyze the constructions (Scontras et al., 2015), thus selecting the less ambiguous option or the indirect morpheme ‑tay.An interesting result displayed by the FCT Aymara tasks was that the proportion of expected responses by adult heritage speakers was higher than that of children in both conditions. In other words, higher scores in the tasks mean more influence from the Spanish aspect in the Aymara evidentiality representation. The former interpretation led us to think that children have a more stable Aymara representation of the source of information than their parents. It is meaningful to remember that all children’s parents are Bolivian migrants, and the children were born in Chile. Schools in Chile include Aymara L2 courses at elementary and intermediate levels (see participants). Three significant educational reforms took place in Bolivia during the twentieth century. The 1905 reform established a national education system, the 1955 reform broadened educational coverage and promoted a homogenous national culture, and the 1994 reform, which is currently in effect, restructured the educational system, encouraged popular participation, and promoted intercultural education (Taylor, 2003). The mean age of the parents was 35; therefore, it is highly likely that the adult heritage-speaking participants had yet to have formal Aymara education, unlike their children in Chile.The Spanish FCT see condition revealed that the bilingual heritage children tended to select the present perfect in the see condition instead of the simple past. Regarding the use of present perfect, there is abundant evidence of the use of compound forms with evidentiality value in Andean Spanish in Peru, Bolivia, and Ecuador (Sánchez, 2014; Ruiz, 2017; Laprade, 1981; Babel, 2009; Dankel, 2017). Therefore, the current study provides more evidence regarding using Spanish compound forms to mark evidentiality, adding comprehension data from the Aymara language in northern Chile.The Spanish FCT also revealed no tendency to use imperfect or pluperfect in the hearsay condition. In other words, participants did not perceive a difference between the two stimuli. These results could be a task effect. Fernandez (2014) examined dialectal convergence in the Andean Spanish in northern Chile. He described the features of Andean Spanish that differed from Chilean Spanish, namely the use of a lexical phrase formed with pluperfect with evidential value at the end of the utterance (me había dicho que / ‘they had told me that.’). The design of the current task did not include a lexical phrase to mark the indirect source of information. Instead, the task included the selection of an imperfect versus a pluperfect sentence (Luis se bañaba v.s. Luis se había bañado/ ‘Luis used to have a bath.’ v.s. ‘Luis had taken a bath.’) Based on that, children may not notice a relevant difference between the two experimental options, suggesting that both options satisfy the input perception.The results of the heritage adults differed from those of the experimental group. In the see condition, adults tended to select the simple past over the present perfect. One possible explanation is that responses could be based on parental perceptions regarding the use of the present perfect linked to a Bolivian variant. Fernandez (2014) assessed how Spanish-speaking migrants from the Bolivian Andes to San Pedro de Atacama (Chile) adapted their speech to Chilean Spanish. The author found a new preference for the simple past over the past perfect to express aspect, which was increasingly marked depending on the length of stay in Chile. Therefore, there is the possibility of a perception of stigmatization among Bolivian Aymara immigrants in Chile for using the present perfect as a strong marker of the Bolivian Spanish variety.In summary, the bilingual comprehension data in the current analysis partially support the hypothesis that the aspect and evidentiality features undergo CLI due to the convergence and divergence of the functional matrix tense in both target languages. On the one hand, it was possible to interpret that the participants had a mental representation of the aspectual background features encoded in the Aymara reported form -tay. However, confirming that Aymara-heritage children were assigned an aspect value to the direct evidentiality morpheme ‑yä was impossible. Furthermore, the present study clearly showed the influence of Aymara in the Spanish discourse of the Aymara-heritage children when they comprehend the direct source of information; however, the results were not compelling for the indirect source of information. Therefore, further reformulation of the methodology design is necessary to explore the target structure.5.2CLI in Production Tasks: EPT sThis analysis attempted to determine whether Aymara-heritage speakers undergo CLI between the Spanish aspect and Aymara evidentiality (and vice versa) in their bilingual oral past narrations. Two EPT s were conducted: an Aymara version performed by the Aymara-Spanish bilingual heritage participants (children and adults) and a Spanish version performed by the three groups (heritage-bilingual population and Spanish-dominant children).The Aymara EPT results showed that most participants produced verbal forms by attaching the indirect evidential morpheme ‑tay. As expected, none of the participants produced the direct evidential morpheme ‑yä. Although there was a low participation rate (only one of the 21 bilingual and five of the 19 bilingual adult participants), the experiments revealed CLI between the two target structures. In other words, Aymara-Spanish bilinguals demonstrated background aspects associated with indirect evidential morphology. The previous results supported previous descriptive studies on the relationship between evidentiality and aspect across languages. Forker and Aikhenvald (2018) described several languages with aspect-evidentiality morphology overlapping. For example, in Wanano, aspect and evidentiality are fused into verbal suffixes that convey visual evidence +perfective semantic value and morphemes that express visual evidence + imperfective semantic value. In Kashaya Pomo, the perfective and imperfective aspect feature is overtly manifested in suffixes that combine visual evidentiality features. If these categories can interact within languages, there is a possibility that the morphosyntactic exponents of Aymara evidentiality can interact with the Spanish aspect at the semantic level due to the long-lasting contact situation between the two languages.The Spanish EPT showed that the present perfect had a shallow frequency across the heritage groups; however, there was a high frequency of the past imperfect progressive verbal form in the Aymara-heritage children compared to the Spanish-dominant group. An explanation for the high frequency of the past progressive form in the heritage population may be understood as a cognitive requirement for fulfilling a morphemic gap due to the complex Aymara morphemic agglutinative structure. The use of compound forms in Andean Spanish to mark evidentiality (see the authors mentioned above) is straightforward. Therefore, it can be inferred that bilinguals who speak Spanish and Quechua or Aymara need to fill the morphological gap with a functional item to surface the source of information in their mental representations. Following Torres (2013) and her analysis of mirative constructions in Spanish and Albanian, I suggest a syntactic agreement relationship between aspect and evidentiality encoded in the aspect phrase and interpreted in the TP domain. This agreement relationship must meet + imperfective and + indirect evidentiality at the left peripheral of the Aspect phrase domain to trigger the imperfective progressive auxiliary in tense. Since no other authors have suggested using different compound verbal forms to express evidentiality, the results of the Spanish production task cannot be considered conclusive regarding the use of past progressive forms to mark evidentiality.In summary, the results of the two production EPT tasks presented partial evidence of functional convergence between two partially divergent categories: Aymara evidentiality and Spanish aspect. The Aymara task results indicated the emergence of a background value in indirect evidential morphology. On the other hand, the Spanish task results showed that the direct evidentiality features might be associated with Spanish morphology but not with the present perfect morphology. This paper proposes the use of the imperfect progressive form as a restructured marker of evidentiality due to the frequency of the form in the participants’ input. However, further studies are required to support this proposition.5.3CLI in the Production Tasks: Retelling Task (RT)The data for the Aymara RT s did not reveal marks of aspect in the evidentiality morphology. In other words, the participant did not produce the attested evidential morpheme ‑yä to describe discrete points in the past. The task results revealed that the attested form was not used in secondhand narration, thus rejecting the hypothesis concerning the use of direct evidentiality morphology with perfective aspectual value in the population. However, the task demonstrated a strong use of the morpheme ‑tay to express the second source of information in the adult population. On the other hand, the data for the Spanish RT task did not show the use of pluperfect as a marker of indirect evidentiality; however, there was a high frequency of past imperfective forms to retell discrete points in the past. Based on the EPT Spanish results (low frequency of present perfect form and high frequency past imperfective forms) and the Spanish FCT results (more than 50 % of the participants selected the present perfect option), there was an asymmetry between the comprehension and production of some grammatical features in the heritage population. The previous assumption was based on the Differential Access Hypothesis (Perez-Cortes et al., 2019), which claimed that, while some features of heritage grammars were difficult to lose in the individual mental representation, these features became more challenging to access for production due to the lack of activation and the constant competition with other grammatical structures. This account could explain the absence of the present perfect as a marker of evidentiality and the high frequency of the progressive form in the participants’ Spanish production.Regarding evidentiality, abundant data show a restructuring of the grammatical representation of the attested morpheme through the present perfect form. The present perfect form is more difficult to access due to its low frequency of use in Chilean Spanish. Therefore, the speakers used a more dominant grammar source with a similar morphological structure (imperfect progressive), which allowed them to restructure their bilingual representation of the direct Aymara evidentiality category. In other words, a bilingual representation of the Spanish morphology encodes Aymara’s evidentiality. However, the target speakers chose a more dominant linguistic option that was present in the input, which was the past progressive form.In sum, the Aymara RT tasks could not find evidence of using the Aymara direct morpheme to describe perfective actions in Aymara. On the other hand, the Spanish version of the task did not display a high frequency of pluperfect form to express an indirect source of information in Spanish past narrations. However, the tasks allow us to identify a strong mental representation of the indirect source of information within the Aymara past discourse. Furthermore, the Spanish RT task evidenced a high frequency of past imperfective forms describing discrete points in the past, suggesting that the verbal form could manifest direct evidentiality, it is necessary to extend the methodology to other participants with similar linguistic features (for example, Quechua speakers in the north of Chile).6ConclusionsThe present work has demonstrated that partially divergent structures are susceptible to CLI, extending Sánchez’s findings (2004). The analysis of the interaction between Aymara evidentiality and the Spanish aspect not only confirmed the presence of cross-linguistic influence similar to that observed in Quechua-Spanish bilinguals but also revealed new insights specific to the discussion on functional features that may undergo cross-linguistic effects. The current investigation demonstrated more evidence in favor of convergence in evidentiality features in the bilingual Spanish narratives in production tasks and emerging evidence of using Spanish compound forms to mark evidentiality. In addition, the current study employed both comprehension and production tasks, providing a more comprehensive approach to assessing bilingual performance. Differences in results across the comprehension v.s. production tasks contribute to understanding universal principles that may govern the heritage language (HL) acquisition by adding data from Aymara, which had not been previously analyzed regarding CLI. The differences between the children and the adults in the CLI tasks also point to interesting conclusions, particularly for the sociolinguistic field. The analysis suggested that preferences in adult comprehension responses were conditioned by their migrant status. Moreover, the analysis allowed for the inference that Aymara lessons in Chile facilitate the maintenance of the HL by validating the ancestral language through formal education.Thus far, there has yet to be an investigation of indigenous heritage languages in Chile, and sociolinguistic research in Chile has a secondary place in the studies concerning the Aymara people, who have received much less attention than other communities. Consequently, the current work focused on Aymara people in northern Chile as early bilinguals who acquired the ancestral and socially dominant Spanish languages early on. Therefore, the relevance of this thesis extends beyond academia, as it also has educational implications. The current project is the first step toward understanding how children acquire Aymara as an HL and Spanish as a second language, thus paving the way for developing an appropriate curriculum for the target population.Supplemental Materialhttps://github.com/jmarkovitsr/CLIAymaraSpanish to review all materials and statistics scripts used in the current research. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Language Contact: Evolution of Languages, Contact and Discourse Brill

Cross-Linguistic Influence between Aymara Evidentiality and Spanish Aspect in Bilingual Heritage Aymara-Spanish Children in the North of Chile

Download PDF
37 pages

Loading next page...
 
/lp/brill/cross-linguistic-influence-between-aymara-evidentiality-and-spanish-qLvamIAYWm

References (32)

Publisher
Brill
Copyright
Copyright © Koninklijke Brill NV, Leiden, The Netherlands
ISSN
1877-4091
eISSN
1955-2629
DOI
10.1163/19552629-bja10092
Publisher site
See Article on Publisher Site

Abstract

1IntroductionThe Aymara people in northern Chile are early bilinguals who acquire both the ancestral and socially dominant Spanish language at an early age. The sociopolitical status of the Aymara language (Southern Aymara, ISO 639–3: ayc, with approximately 20,000 speakers in northern Chile and classified as vulnerable by UNESCO, n.d) within a dominant Spanish country influences the language acquisition process resulting in Aymara in northern Chile being heritage speakers.Therefore, Aymara people in Chile have low proficiency levels when they speak Aymara (Espinosa, 2009; Fernandez, 2014; Espinoza, 2019), triggering the mainstream idea that Aymara people are Spanish monolingual with a strong indigenous identity (Ticona and Markovits, 2023). The previous generalization is reflected in the National Intercultural Language Program (MINEDUC, 2019) that aims to develop second-language Aymara courses, ignoring the initial exposure to the heritage languages (HL), which can be advantageous in revitalizing the language. However, Spanish literacy courses are designed for monolingual Spanish speakers, not considering the influence of the HL in the Spanish acquisition of the Aymara community members.The current study explores two side effects of Aymara evidentiality and Spanish Aspect and demonstrates that Aymara speakers in northern Chile are bilingual speakers who engage in cross-language effects. Thus, evidence of the cross-linguistic effect between Aymara and Spanish (and vice versa) will provide relevant information and an appropriate intercultural curriculum for Aymara children in northern Chile.2Theoretical Framework2.1Cross-Linguistic Influence (CLI) and the Functional Convergence HypothesisThe CLI concept emerged to explain different bilingual outcomes that the Autonomous Development Hypothesis, according to which bilingual children develop separate grammars at a very early stage, could not address (Meisel, 2001). Studies in this area refer to CLI as instances in which there is evidence of the effect of one language on the other generally at the morphosyntactic level (Serratrice, 2013). Theories like the Interface Hypothesis (Sorace and Filiaci, 2006; Sorace and Serratrice, 2009; Sorace, 2011) have opened a line of study to understand the influence of one language on another in a bilingual population. For instance, Sánchez et al. (2023) examine whether Spanish exposure (among other variables) affected pragmatic factors that guide The null subject in Spanish (a grammatical structure that lies on the syntactic-pragmatic interface) in bilingual heritage Spanish-English children. The authors evidenced a positive relationship between increased interaction at home and accurate comprehension of pragmatic conditions that license null subjects, and in turn, the results did not show the influence of Spanish on overt subject patterns in English, the dominant language.Building on this understanding of CLI, Sánchez et al. (2010) investigated cross-linguistic influence in a specific context: indigenous heritage languages in contact with Spanish. The author examined first-language morphology properties in the second-language grammars (residual transfer morphology) and cross-linguistic influence on the syntax–pragmatics interface in L1 speakers of Shipibo with different levels of formal instruction in Spanish as a second language. The authors investigated the impact on languague contact on subject-agreement morphology and the null subject properties in both languages. Shipibo has no verbal morphology on the verb and licenses null subject only in the third person, unlike Spanish, which is a null subject language with rich person morphology on the verb. The authors found that bilingual speakers produced a non-standard distribution of the null subject in Spanish, indicating the attainment of a complex syntax-discursive interface. However, the results did not reveal morphological residual transfer from Shipibo in Spanish subject-verb agreement. Previous studies partially supported the claims of the Interface Hypothesis since results showed that instability at the syntax–pragmatic interface persists only under certain conditions. Therefore, other conditions should be addressed in explaining the cross-linguistic effects of language contact (Odlin, 1989).Sánchez (2003) examined syntactic and morphosyntactic changes in the bilingual Quechua-Spanish discourse of children in Peru. She analyzed CLI in acquiring functional categories, such as the representation of the direct object and word order in both systems. Quechua has an obligatory accusative case morpheme ‑ta on the direct object, and the same morpheme can license null nouns in the direct object position. In Spanish, the definite determiners also license null nouns in certain semantic and pragmatic conditions. Sánchez revealed the absence of accusative marking on direct objects in Quechua and the presence of a definite determiner marking null nouns in a non-canonical position in Spanish. These partial similarities between the Spanish definite determiner and the Quechua accusative marker ‑ta indicate the grammatical changes in both languages, allowing the emergence of new structures, thus demonstrating that partial similarities between languages in contact can undergo CLI. These findings led the author to elaborate on the Functional Convergence Hypothesis (2004), which states that different functional features present in a language contact situation can be subject to CLI. The FCH initially stated that convergence, the structural similarities between two linguistic systems in a contact situation, occurs when functional features not present in one of the languages are mapped onto features of another language. Sánchez (2004) modified her proposal, arguing that convergence was favored when the matrix of features associated with a functional category was partially divergent. She tested the hypothesis examining CLI between Spanish Aspect and Quechua evidentiality since, in Quechua, past tense is closely tied to evidentiality, whereas in Spanish, it is linked to aspect. A group of Quechua-Spanish bilingual children performed a retelling story task in the target languages. The results showed a high frequency of the use of the Quechua reportative morpheme ‑sqa to convey an indirect source of information (in other words, that a third person reported the action) and a shallow frequency of the attested form ‑rqa that could convey Spanish aspect. The evidence supporting convergence in evidentiality features within the bilingual Spanish narratives was more compelling. Bilingual participants, on average, used imperfective morphology during a task involving the transmission of hearsay information. This suggests that bilinguals associate imperfective morphology with evidentiality features.This paper seeks to build on the foundational work of Sánchez (2004), which identified significant cross-linguistic influence between Quechua evidentiality and Spanish aspect in bilingual Quechua-Spanish children. By extending these findings, the present study examines a similar phenomenon in Aymara-Spanish bilingualism, focusing specifically on the interaction between Aymara evidentiality and the Spanish aspect. In doing so, this research broadens the scope of cross-linguistic influence to include another indigenous language. It contributes to a deeper understanding of how evidentiality systems in indigenous languages shape aspectual interpretations in bilingual contexts.2.2Evidentiality in Andean SpanishEvidentiality is a linguistic category in which the primary meaning of a statement is the source of information (Aikhenvald, 2018). Evidentiality specifies the origin of an utterance content. The origin refers to the degree of evidence between the speaker and the event. The degree of evidence ranges from testimonial evidence (e.g., vision, hearing, smell, etc.) to non-testimonial evidence, such as a second source of information, an inference based on visual traces, reasoning, and general knowledge (Aikhenvald, 2018: 1).Languages differ in the way they convey the source of information. Spanish does not have an obligatory grammatical evidentiality system. However, there are several lexical expressions to mark the source of information, such as vi lo que ella hizo, “I saw what she did” (direct evidence), or me dijeron lo que ella hizo, “they told me what she did”, (indirect evidence). Sometimes the evidential concepts are encoded in mental or perception verbs (e.g., saber, know; creer, guess; escuhar, hear).Furthermore, Sánchez (2004) analyzed different interpretations of Spanish imperfective morphology. She argues that the imperfect tense is used to report events occurring outside of consciousness, such as dreams. The author offered the following example:(1)En mi suenoyomeganabaLa loteriaIn my dreamICL1SGwinPAST IMPERFthe lotery‘In my dream, I won the lottery’.In the case of Andean Spanish, several authors have shown a systematic use of evidentiality features in this variety of Spanish. Fernandez (2014) examined dialectal convergence in the Andean Spanish divergence in contact with Chilean Spanish in the north of Chile. The author presented a description of the features of Andean Spanish that differed from Chilean Spanish, one of which was the use of pluperfect to express non-experienced past tense action. The author provided the following example:(2)Medijeron que la genteloshabía discrimadoaLos chilenos1. 1S.DATtold that the people3.PL.AChave-IMP.3.S discriminatedDOMthe Chileans‘They told me that people had discriminated against Chileans.’Escobar (2011) reviewed the features of Andean Spanish that differentiated this dialect from other varieties of Spanish and described how some bilingual speakers in Bolivia and Peru, where numerous varieties of Quechua are commonplace, appeared to use the pluperfect for reported information and the present perfect to mark first-hand information. Bustamante (1991) found that the present perfect had a reportative function in Ecuadorian Spanish. These findings suggest that the evidential systems of indigenous languages in the Andean region of South America influence the restructuring of the perfective aspect in Spanish in multiple ways.To summarize, the expression of evidentiality in Spanish is mainly manifested through lexical expressions such as verbs, adverbs, modal verbs, and conjunctive locutions. Nevertheless, a vast amount of literature about Andean Spanish has found perfective/imperfective morphology to be a mark of evidentiality. Although there is no consensus about the contrast between the present perfect and the pluperfect to mark direct and indirect evidentiality, these studies suggest that functional features such as aspect and evidentiality are subject to permeability in a language contact situation. The current investigation aims to contribute to previous studies, analyzing the impact if Amymara on a less-studied variety, the case of Chilean Spanish spoken in the north of the country, examining the comprehension and production of the present perfect or the pluperfect to mark the source of information in the bilingual population.2.3Aymara EvidentialityThe Aymara language has a diverse evidential system, where different evidential categories are expressed through verbal inflections within the distribution of tenses.1 The Aymara understanding of time is distinct from European language-based models. The timeline in Western languages places the past behind the speech act (S) and the future in front of it. However, Aymara conceptualizes time so that the future is behind the speech act, and the past is in front (Miracle and Yapita Moya, 1975; Núñez and Sweetser, 2006). Then, Aymara conceptualizes time in terms of future and non-future actions. The non-future is divided into remote and present/recent past actions. The former refers to an action that occurred before the statement was completed (Ticona, 2007).In addition to tense, Aymara speakers must encode the source of information for events occurring in the past. Hardman et al. (2001) identify a distinction between personal knowledge tenses, associated with a non-remote past, and non-personal knowledge tenses, which include two remote past categories: remoto cercano (‘near remote past’) and remoto lejano (‘far remote past’). According to the authors, these tenses differ in their degree of remoteness, with remoto cercano referring to events that are relatively recent and remoto lejano indicating events that occurred further in the past. The non-remote past encodes a direct source of information, whereas the remote past tenses convey an indirect source. However, not all verbal inflections mark these distinctions consistently, suggesting a more nuanced evidential system.Figure 1Aymara tense paradigmFigure 1: Aymara tense paradigmSimilarly, Cerrón-Palomino (1994) describes two past tenses: pasado experimentado (‘experienced past’), overtly marked by the suffix ‑yä, and pasado no experimentado (‘non-experienced past’), indicated by the verbal morpheme ‑tay. According to him, this distinction is evidential: ‑tay encodes a non-experienced past, indicating an indirect source of information, while ‑yä marks a recent past (non-remote) associated with direct personal knowledge of the action. Ticona (2007) adopts the terms testimonial and non-testimonial to distinguish between recent personal knowledge and remote non-personal knowledge, respectively, using ‑yä and ‑tay to mark direct and indirect information sources.2 The following examples illustrate the verbal paradigm of evidentiality in the recent and remote past tenses:(3)Luysumanzanamanq’-ä-naLuisappleeatREC/DE3>3‘Luis ate an apple.’(4)Sapa uruthuqtayna (Ticona, 2007:42)Every daydanceREM/IE 3>33‘He danced every day!’ (the speaker is not witness of the action)2.4Spanish AspectAspect is a linguistic category that expresses different temporal properties of past events. Like evidentiality in Aymara, the Spanish aspect is overtly manifested in different past morphemes. Comrie (1976) described the grammatical aspect, or viewpoint aspect, as a distinction between perfectivity and imperfectivity. He described the perfective aspect as ‘looking at the event from the outside’ (Comrie, 1976: 17) and the imperfective aspect as ‘looking at the event from the inside’; (Comrie, 1976: 17).Spanish contains two past tense verbal paradigms, which align with the perfective/imperfective distinction. For example,(5)Jugu-éfútbolen la universidadPlayPRF.1.Ssoccerin the University‘I played soccer in college.’More recently, Bardovi-Harlig (1994; 1998; 2000) proposed the discourse hypothesis, which claims that aspects are determined by narrative structure. The author described different perspectives of the events within the narrative structure and distinguished between foregrounding and backgrounding in narratives by referring to the imperfective-perfective distinction. A background feature may have three interpretations: (1) the background verb aspect indicates that the verb occurred before the focal event; (2) the aspect of the verb indicates a prediction about the outcome of an event; or (3) the aspect could be interpreted as an evaluation of an action reported in the foreground. The foreground verb aspect, on the other hand, is in the event timeline. The properties of events in the past and their relation to aspect can be illustrated using Reichenbach’s (1947) speech parameter. The author described three different parameters of speech: a speech act (S), an event (E), and a reference point (R). In context, background and foreground clauses in the past take place before the speech act, but the event and reference points differ. Figure 1 explores this relationship.Figure 2Spanish aspect and the reference pointFigure 2: Spanish aspect and the reference pointBoth background and foreground events are located before the moment speech act (S). However, the chronological relationship between the point of the event (E) and the reference point (R) is different in the timeline. In the imperfective aspect, the reference point is before the event, unlike the perfective aspect, in which the event and the reference points are simultaneous.Here, we find a potential for convergence between the two systems, which could lead to CLI since Spanish and Aymara have two verbal morphology representations in the past tense. Such findings would be consistent with those of Sánchez (2004), who found that Quechua-Spanish bilingual children transferred evidential properties to the Spanish aspectual system and aspectual properties to the Quechua evidential system, as proposed in the present study. In addition, the distance between R and E in background events is greater than in foreground events. Similarly, the direct evidential Aymara morpheme ‑yä conveys a shorter distance between the reference point and the event, unlike the indirect morpheme ‑tay, which encodes a greater distance; therefore, the comprehension and production of both structures require the speaker to understand not only the morphosyntactic features of the structures but, also the pragmatic features linked to the reference point within the timeline.3The Present StudyThe present investigation examined the distribution of the two past verbal morphemes in Aymara, spoken in northern Chile, exploring whether the evidentiality system exhibits the value of Spanish aspects in the past narrative discourses of Aymara-Spanish bilingual children. The current research aims to demonstrate that bilingual speakers assign an evidential value when they produce and comprehend past narration in Spanish. Previous studies have demonstrated a non-standard distribution of present perfect and pluperfect in Andean Spanish. Therefore, the present work expects to find evidential marks in a high frequency of compound past forms in Aymara-Spanish bilingual speakers, which differ from the standard distribution of the simple and imperfect past in the Spanish Chilean variety (de Alba, 1998; Fernandez, 2014).I will test the Functional Convergence Hypothesis by Sánchez (2004) which states that convergence is favored when the matrix of features associated with a functional category is partially divergent. In the current study, the functional category is tense, and the divergent features are aspect (namely perfective and imperfective morphology) and verbal inflectional suffixes that mark evidentiality. Therefore, the current work proposes four research questions:Do heritage bilingual Aymara-Spanish children produce aspect morphemes with an evidential value in their past narrative discourse in Spanish?Do heritage bilingual Aymara – Spanish children comprehend aspect morphemes with evidential value in their past narrative discourse in Spanish?In response to these questions, it is hypothesized that evidentiality converges with aspect features in the Spanish discourse of Aymara-Heritage speakers in a contact situation.Do heritage bilingual Aymara-Spanish children produce past evidential morphemes with aspect value in their past narrative discourses in Aymara?Do heritage bilingual Aymara-Spanish children produce past evidential morphemes with aspect value in their past narrative discourses in Aymara?In response to these questions, it is hypothesized that aspect converges with evidentiality features in the Aymara discourse of child Aymara-heritage speakers in a contact situation.3.1MethodologyThe methodology included three groups of participation. The experimental group contained 21 heritage Aymara-Spanish bilingual children (HS) aged four to 12 years. These children live in the region of Tarapacá in the northern part of Chile, specifically in the community of Sibaya, one of the rural regions where there is a coexistence between the traditional Aymara culture and the Chilean culture (Goldstein, 2015). They used Aymara and Spanish simultaneously during their first two or three years. At the age of 4, they start pre-k schooling instruction in Spanish. From 6th to 8th grade, they have Aymara as a second language for three hours per week. After 8th grade, they must move to the biggest nearest town (Huara) to complete high school. When I conducted my research (2021), there were no Aymara L2 programs in Huara High School. The baseline of the heritage children was 15 heritage Aymara-Spanish bilingual adults (AD). The adults were the parents of the heritage children and lived in the same rural community. All the adult participants had migrated from Bolivia to Chile. In addition, 19 heritage Aymara-culture Spanish-dominant children (SP) of the same age as the experimental group participated in the study. The monolingual children lived in the capital of the Tarapacá region, Iquique, which is three hours away from Sibaya. They had been exposed to the traditional Aymara culture and knew some lexical items and phrases in Aymara. However, they did not understand or speak Aymara. Including the monolingual as a comparison group allowed us to understand the effects of the heritage language in the Spanish of the Aymara heritage speaker children, testing whether the two children groups behaved similarly (or not) when they needed to comprehend and produce Spanish sentences in the past. Table 1 includes relevant sociolinguistic features of the children’s participants for the current study.Table 1The demographic of the heritage (HS) and monolingual (SP) childrenGroupNMean ageFirst languageSecond languageDominant languageHS218.5AymaraSpanishSpanishSP197.9Spanishn/aSpanishThe following tests and tools were used:1. Modified sociolinguistic background questionnaire (Fernandez, 2014): The children’s parents completed a language questionnaire in Spanish, answering to questions about the children’s and family members’ activation, frequency, and exposure to Spanish and Aymara.2. A screening bilingual Spanish-Aymara test based on the textbook Para compartir voces: Texto de consulta para profesores de castellano como segunda lengua (Sánchez, 2014). The test was used to assess the Aymara and Spanish levels of the heritage children and adults. The maximum score was 43, and the minimum Aymara score to perform the Aymara version of the tasks was 15 points or intermediate I level.3. Experimental Items and Tasks: Spanish Production Tasks. To investigate Research Question 1 – whether Aymara heritage children mark Spanish past narratives with Aymara evidentiality – the following tasks were employed:3a) Elicited production task (EPT)This experiment attempted to elicit brief oral narratives in the Spanish past to examine whether Aymara-Spanish heritage bilingual heritage speakers produced the present perfect as a mark of direct evidentiality (Aymara experienced past tense). The previous assumption was based on Escobar (2011) and Rojas-Sosa (2008), whose findings showed evidence of a higher-than-standard distribution of the present perfect in narrative clauses in Andean Spanish speakers.To examine whether bilingual children used the present perfect instead of the simple past to narrate actions they had witnessed, the Principal investigator (PI) conducted ‘the see condition’ EPT. The interviewer told the children that she would show them pictures of Luis, a teenage boy. The PI then asked the children to tell her what Luis did yesterday. The children were shown eight pictures illustrating Luis’ actions from the previous day. The interviewer then asked the children to tell her what Luis did. For example:Picture 1: The child was shown an image of Luis waking up. The interviewer asked, ‘Can you tell me what Luis did yesterday?’Expected response: Luis ha despertado (‘Luis has woken up’).Unexpected response: Luis despertó (‘Luis woke up’).Picture 2: The child was shown an image of Luis buying tomatoes. The interviewer asked, ‘Can you tell me what Luis did yesterday?’Expected response: Luis ha comprado tomates (‘Luis has bought tomatoes’).Unexpected response: Luis compró tomates (‘Luis bought tomatoes’).If participants used the present perfect instead of the simple past tense preterit, I assumed they used it as a mark of direct evidentiality. On the other hand, if there were a preference for using the simple past tense, the Aymara evidential system would not have affected their production.3b) Retelling task.This experiment attempted to elicit past narratives in Spanish containing aspectual morphology to assess whether Aymara-Spanish bilingual children produced the pluperfect as a mark of indirect evidentiality (Aymara experienced past tense). The assumption was based on Fernandez (2014), who showed the use of the pluperfect to express non-experienced past tense actions in Andean Spanish. The PI presented a retelling story task to examine whether the bilingual children used the pluperfect, a periphrastic construction, instead of the simple past to describe an action in the past that they had not witnessed. The children listened to a Spanish recorded audio story, ‘La música en las montañas,’ after which the PI asked the children to tell her the story. For example:The child listened to (input): Todos los días Chuqu le pedía al yatiri que le mostrara el camino (‘every day Chuku asked the wise man to show the way’).Expected response (output): Todos los días Chuqu le había pedido al yatiri que le mostrara el camino (‘every day Chuku had asked the wise man to show the way’).4. Production tasks – Aymara versionThe current experiments were designed to answer research question 3 of whether Aymara heritage children produce Aymara past narratives marking the Spanish aspect. The tasks were the following:4a) EPT. This experiment attempted to elicit the production of Aymara evidential past morphemes to examine whether Aymara-Spanish bilingual children produced the non-experienced past morpheme (‑tay) as a marker of imperfective actions in the past. To examine whether the bilingual children used the non-experienced past morpheme with the imperfective value in Aymara, eight pictures of Luis as a baby appeared on the laptop computer, and the PI told the children that they knew Luis. They had seen what Luis did when he was young. Each child was administered eight trials. The interviewer then asked the children to describe what Luis did as a child. For example:Picture 3: The child was shown an image of Luis with a ball. The interviewer then told the child that he/she witnessed what Luis did as a child and asked, ‘Can you tell me what Luis did when he was a child?’ The expected answers were:– Expected response:Luis piluta- mpianata-tay-naLuis ball-INSTPLAYREM/IE3> 3‘Luis played with the ball – someone told me.’– Unexpected response:Luis piluta- mpianata- tä-naLuis ball-INSTPLAYREC/DE3> 3‘Luis played with the ball – I was a witness.’Picture 4: The child was shown an image of Luis reading a book. The interviewer then told the child that he/she witnessed what Luis did as a child and asked, ‘Can you tell me what Luis did when he was a child?’ The expected answers were:– Expected response:Luis pankliyi-tay-naLuis bookreadREM/IE3>3‘Luis read the book – someone told me.’– Unexpected response:Luis pankliyï-naLuis bookread – REC/DE-3>3‘Luis read the book – I was a witness.’Suppose there was a preference for the use of -tay. In that case, it is plausible that the children produced the non-experienced past morpheme with the imperfective aspectual value since the children had witnessed the action that the character had performed habitually in the past. On the other hand, if there was a preference for the use of -yä, it was likely that the participants had not mapped an aspectual value onto any evidential marker in the past narration. In other words, the action in the past was modulated by the Aymara evidentiality grammar structure.4b) Retelling task. The children listened to a recorded version of the story ‘La música en las montañas’ in Aymara. After they had listened, the children were asked to tell the story to the PI. For example:– the child listened to (input):Chukumayantiqinaphusa-ñayatt ´ataynaChukuone more timequenaPLAYNMLZTRYREM/IE3>3‘One more time Chuku had tried to play the quena, someone told me.’– Expected response (output):Chukumayantiqinaphusa-ñayattänaChukuone more timequenaPLAYNMLZTRYREC/DE3>3‘One more time Chuku had tried to play the quena, I witnessed the action.’If there was a preference for ‑yä to describe a discrete event in the past, it was plausible that the children produced the experienced past morpheme with a perfective aspectual value since they did not witness the character’s action in the past. On the other hand, if there were a preference for the use of ‑tay in these contexts, the logical conclusion would be that Spanish perfectivity was not mapped onto the evidential morpheme in Aymara. In other words, the action in the past was modulated by the source of information.5. Comprehension tasks – Spanish versionThe current experiments were designed to answer research question 2 of whether Aymara heritage children comprehend Spanish past narratives marking Aymara evidentiality. The tasks were the following:5a) Forced-choice task (FCT) – see condition. The task aimed to examine whether the children assigned evidential value to aspectual morphemes in Spanish, that is, if they selected the present perfect for assigning the experienced past tense. The PI introduced two characters (the girl and the boy) and told the children that they were Luis’ friends and that they had seen what Luis had done the previous day. The PI told the child to listen to what the girl and the boy said and asked the participant to select which character sounded best. Each child was administered eight trials with present perfect and simple past audio recordings.For example:Audio 1: Luis tomó café (‘Luis had coffee’).Audio 2: Luis ha tomado café (‘Luis has had coffee’).If there was a preference for selecting the present perfect instead of the simple past, I assumed that the Aymara-Spanish bilingual participants were assigning the present perfect tense as a mark of evidentiality. On the other hand, if there was a preference for selecting the simple past, their selection of evidential morphology in Aymara was modulated by the Spanish aspect category.5b) FCT – hearsay condition. The task aimed to examine whether children assigned evidential value to aspect morphemes in Spanish, namely the pluperfect, to assign the non-experienced past tense. The PI told the children that the same characters in the previous task were told what Luis did when he was a child. The characters then described what they had heard about Luis, and the participants needed to decide which one sounded best. Each child was administered eight experimental trials with pluperfect and imperfect past audio recordings.For example:Audio 1: Luis jugaba a la pelota (‘Luis played the ball’).Audio 2: Luis había jugado a la pelota (‘Luis had played the ball’).If there was a preference for using the pluperfect instead of the imperfect past, it could be assumed that the participants were mapping the value of indirect evidentiality onto the pluperfect. On the other hand, if there was a preference for using the imperfect past, their production was only modulated by the Spanish aspect category.6. Comprehension tasks – Aymara versionThe current experiments were designed to answer research question 4 of whether Aymara heritage children comprehend Aymara past narratives marking Spanish Aspect. The tasks were the following:6a) FCT: See condition. This experiment examined whether Aymara-Spanish bilingual children assigned imperfect aspects to the non-experienced past morpheme ‑tay. The PI told the child that the same two characters were now seeing what Luis did as a child. Then, two audios popped up. After each character’s utterances, the PI asked who said it better. Each child was administered eight experimental trials with ‑tay and ‑yä verbal forms.For example:Audio 1: Luisa mä panq liyitayna (‘Luis read a book, someone told me’).Audio 2: Luis mä panq liyi:na (‘Luis read a book, I witnessed’).If there was a preference for the ‑tay option, I assumed the children were assigned an imperfect aspect value to the non-experienced past morpheme since the speaker witnessed the character’s action. On the other hand, if there was a preference for using ‑yä, I inferred that the children had not assigned any value to aspect features in the past narration. In other words, the action in the past was modulated by the source of information.6b) FCT: Hearsay condition. To examine whether the bilingual children assigned a perfective value to the experienced past morpheme – yä in Aymara, the PI told the children that the same characters had listened to what Luis did yesterday. The characters then described the picture, and the children needed to select who said it better. For example:Audio 1: Luis manq’äna (‘Luis ate, I witnessed’).Audio 2: Luis manq’tayna (‘Luis ate, somenone told me’).If there was a preference for selecting the past experienced morpheme ‑yä, I assumed that the children were assigned a perfect aspect value to the experienced past morpheme since the child did not witness the action in the narration in the past. On the other hand, if there was a preference for the use of ‑tay, I inferred that they had not assigned any value to the aspect feature in the past narration. In other words, the action in the past was modulated by evidentiality features.3.2ProcedureThe experimental phase was conducted in a single session, following this sequence: first, participants completed a 15-minute bilingual language screening test in Aymara and Spanish. This was followed by an Aymara Elicited Production Task (EPT) lasting approximately five minutes, and then a Spanish EPT, also taking around five minutes. Next, participants engaged in Spanish Forced-Choice Tasks (FCT s) for about five minutes, succeeded by Aymara FCT s of similar duration. The session continued with an Aymara retelling task (approximately five minutes) and concluded with a Spanish retelling task (also around five minutes).4ResultsThe analysis presents the results of the tasks performed by the 56. The participants who were divided into three groups, namely the heritage children group (HS), the Spanish-dominant-cultural heritage children group (SP), and the Aymara Adults heritage speakers (AD) group. Eight experiments were conducted: two Aymara forced-choice tasks (FCT), one Aymara elicited production task (EPT) and one Aymara retelling task, and two Spanish FCT s, one Spanish EPT and one Spanish retelling task. All groups performed the Spanish version of the exercises. However, only the Aymara-Spanish bilingual participants performed the Aymara version of the tasks since a minimum score of 15 out of 43 was required in the Aymara screening test to perform the Aymara exercises. The statistical analysis of the collected data was conducted using the R programming language.4.1Aymara Comprehension Tasks: Forced-Choice Tasks (FCT)The current analysis attempted to answer research question 4 of whether the HS group comprehended Aymara evidential morphemes with a Spanish aspect value. Each participant (21 HS; 15 AD) completed two FCT comprehension tests: the see condition and the hearsay condition. The participants completed eight experimental items in each test, distributed randomly by PPT. Task responses were coded as 1 for the expected response and 0 for the unexpected response. Figure 3 shows each group’s proportion of expected responses in each task by group. The y-axis represents the total number of expected or ‘1’ responses.Figure 3Proportion of expected responses across bilingual participants in the Aymara comprehension taskFigure 3: Proportion of expected responses across bilingual participants in the Aymara comprehension taskThe results for the heritage children showed that in the see condition (right block), the mean was above chance (X= 0.53, SD = 0.50), demonstrating that more than 50 % of the participants selected the indirect morpheme ‑tay to describe a remote action that they were witnessing in the past, thus suggesting that the participants assigned an imperfect aspect value to the indirect evidential morpheme. In the hearsay condition task, less than 50 % of the heritage children selected the morpheme ‑yä to describe a recent past action secondhand information (X= 0.45, SD = 0.50), showing that more than 50 % of the participants selected the unexpected response (the selection of the indirect morpheme ‑tay), suggesting that children did not assign an aspect value to the direct morpheme ‑yä. Regarding adult heritage speakers, the frequency of ‘1’ responses was higher than in see (X= 0.88, SD = 0.32) and hear (X= 0.51, SD = 0.50) conditions compared to the HS group. An explanation of mean differences by group could be that Spanish influence has a greater impact on how heritage adults comprehend the source of information than children (see discussion).4.2Spanish Comprehension Tasks: Forced-Choice Tasks (FCT)The current analysis attempted to answer research question 2 of whether the HS group comprehended Spanish aspect morphemes with an Aymara evidential value. Specifically, the current task aimed to identify whether the participants selected the present perfect verbal form to describe an action they had witnessed in the past (see condition), as well as to identify whether the participants selected the pluperfect form to describe pictures of secondhand source of information (hearsay condition). Each participant completed two comprehension tasks, namely the see and hearsay conditions. The participants completed eight experimental items in each test, distributed randomly by PPT. Task responses were coded as 1 for the expected response and 0 for the unexpected response. Figure 4 shows each group’s proportion of expected responses in each task by group (21 HS, 15 AD, and 19 Spanish dominant children with a strong Aymara identity or SP). The y-axis represents the total number of expected or ‘1’ responses.Figure 4Proportion of expected Spanish comprehension task responses groupsFigure 4: Proportion of expected Spanish comprehension task responses groupsThe HS group performed differently across conditions. Heritage children’s score was above chance in the see condition (X=0,59; SD= 0,49), showing that more than 50 % of the heritage children selected the expected response. In other words, they preferred the present perfect rather than the simple past to describe an action they witnessed. On the contrary, they performed on chance in hear conditions (X= 0.5; SD= 0.50), showing no tendency for any choices (imperfect or pluperfect). The results could be due to a task effect (see discussion). Adults had fewer expected responses in the see condition (X= 0.4, SD= 0.49) than children. Adults tended to select the use of simple past tense to express a witnessed action in a recent past. In addition, AD expected response in the hear condition was higher (X= 0.57; SD= 0.49) than in the HS group, showing a tendency to select the pluperfect to describe a reported action in the past. Differences in conditions across heritage groups are discussed in the next section (see discussion). Finally, the SP group’s performance was below chance in see (X= 0.32; SD: 0.46) and hearsay (X= 0.30; SD: 0.46) tasks, demonstrating that Spanish-dominant children tended to select the simple past instead of the present perfect and the imperfect past instead of the pluperfect in their responses to describe actions in the past, as it was expected.4.3CLI in Production Tasks4.3.1Aymara Elicited Production Task (EPT)The current analysis attempted to answer whether the heritage children (HS) produced Aymara evidential morphemes with a Spanish aspect value. Specifically, the current task attempted to identify whether the participants elicited the non-direct morpheme ‑tay with an imperfect aspect value. Each participant completed an EPT task in which they witnessed the actions that happened long ago (see condition). The participants completed eight experimental items that were randomly distributed by PPT. Only one of the 21 HS participants and five of the 19 AD participants performed the task. The low performance was because that participant said they could not speak Aymara. They said they did not know how to describe the pictures task orally. Due to the low number of participants, the current sections provide the frequency of evidential forms elicited from the participants in each group. The number of verbal tokens was counted by group. Then, the number of evidential morphemes was counted to examine the frequency. Participants did not produce any ‑yä morpheme. Participants produced verbal forms with ‑tay morpheme or verbal forms with ‑ki morpheme. ‑ki morpheme is an aspectual- long-lasting morpheme formed by two suffixes: the suffix of incompletive aspect ‑ka and the suffix ‑i, which indicates the third person of the simple tense (Cerron-Palomino, 1994).Table 2 shows the frequency of the ‑tay and ‑ki morphemes by group in the see-condition EPT task.The frequency table shows that the most produced morpheme was ‑tay, followed by ‑ki. It is helpful to remember that the participants were told they had witnessed a character’s actions long ago. Table 3 provides examples of the expected responses for each experimental item and the participants’ most frequent responses.Table 3 shows the data that most participants produced verbal forms with indirect evidential morphemes. This behavior differed in item 3 and item 6, in which the most frequent verbal type did not attach the non-experienced past morpheme ‑tay. Conversely, the participants in item 3 and item 6 tended to elicit the aspect-durative morpheme ‑ki. The reason could be a task effect. In other words, images 3 and 6 could depict a progressive action, unlike the rest of the pictures. In addition, using the imperfective morpheme -ki could be a case of attrition of the syntactic evidentiality representation. However, further investigation needs to be done to support the previous claim.Table 2Frequency of evidential morpheme across groupsn ofn tokens (numbern ‑kin ‑tayparticipantsof verbal forms)morphememorphemeHS1826AD532626Table 3Expected responses versus frequent responsesExperimental itemExpected responseMost frequent response1Luis piluta anatatayna‘Luis played with the ball, someone told me.’piluta anatatayna‘with the ball he played, someone told me.’2Luysu wasa isi uchasitayna‘Luis put on cloth from someone else, someone told me.’Disfrasasitayna‘(he) put on a costume, someone told me.’3Luysu mapanka liyitayna‘Luis read a book, someone told me.’Libri liyiki‘Luis is reading.’4Luysu uta luratayna‘Luis built a house, someone told me.’Luis anatayna‘Luis played, someone told me.’5Luysu jarisitayna‘Luis washed himself, someone told me.’Luis wañasitayna‘Luis took a bath, someone told me.’6Luysu mansana maq’atayna‘Luis ate an apple, someone told me.’Manzana katuski‘(he) grabbing the apple’7Luysu mamapampi saratayna‘Luis walked with his mother, someone told me.’Mamallpampi sarkatayna‘He walked with his mother, someone told me.’8Luisu karta qilqatayna‘Luis wrote letters, someone told me.’Karta qilqatayna‘(he) wrote letters, someone told me.’4.3.2Spanish Elicited Production Task (EPT)The current analysis tried to answer whether the HS group produced Spanish aspect morphemes with an Aymara evidential value. Specifically, the current task attempted to identify whether the participants used the present perfect instead of the simple past to describe actions that they had witnessed. Each participant completed an EPT task in which they witnessed the actions they needed to produce (see condition). The participants completed eight experimental items that were randomly distributed randomly by PPT. The expected response was the production of the present perfect. However, participants produced six verbal forms, as shown in Table 4.Figure 5Production of verbal forms across the groupsFigure 5: Production of verbal forms across the groupsTable 4Verbal forms produced by participantsVerbal formExampleImperfect (imperf)Luis leíaimperfect progressive (imPR)Luis estaba leyendoPast periphrastic form other than imperfect progressive (periph)Luis se fue a leerPluperfect (pluper)Luis había leídoPresent perfect (PP)Luis ha leídoPresent progressive (presPR)Luis está leyendoSimple past (simplep)Luis leyóDue to the variability in participant responses, the expected response was any verbal form other than simple past. This allows us to examine a non-standard distribution of verbal production to describe discrete points in the past. Figure 5 shows the frequency of each verbal form produced by the group.The data shows that the imperfect progressive (imPR) had a high frequency within the HS (N=68) and the AD (n=41) groups. Regarding HS, the most frequent form after imPR was the simple past (26), followed by the present progressive (N=9). The expected present perfect form (PP) had a low frequency within the SP corpus (N=6), similar to the AD group (N=2). As was expected, the SP group showed a higher frequency of simple past. The most frequent form produced by the Spanish-dominant children after the simple past was the present progressive (n=18), followed by the past imperfect (n=3).4.4Aymara Retelling Task (RT)The current analysis aimed to answer whether the HS group produced Aymara evidential morphemes with a Spanish aspect value. Specifically, the current task attempted to identify whether the participants elicited the direct morpheme ‑yä with a perfective value in a secondhand narration to describe discrete points in the past. Each participant listened to a recorded Aymara narration and was then asked to retell the story to someone else. Similar to the previous task, only one of the 21 HS participants and only one of the 19 AD participants performed the task. The reason for not performing the task was the same as on EPT. Most of the participants argued that they were not able to retell the story in Aymara. The current sections analyze the most frequent type of evidential morphemes used to describe the story. The input story contained a total of 17 verbs, of which one token was a verbal form attached with a future morpheme (marked in bold), and 16 were verbal forms attached with the non-evidential morpheme ‑tay (underlined) (see supplemental material). No attested morphemes appeared in the story due to the secondhand source of information nature of the instrument.The number of tokens (verbal forms) in both participants was counted. The HS participant produced zero verbal tokens. The HS participant retold the story based on nouns, as the following example illustrates:(10) PI: ¿Le puedes contar la historia a tu hermana?‘Can you tell the story to your sister?’HCH: mmmm, no sé‘Mmmm, I don’t know.’PI: ¿y de que te acuerdas?‘What do you remember about it?’HCH: Jamp’atu, uma, yuqalla, imilla‘Frog, water, boy, girl.’Unlike the HS participant, the AD participant produced nine verbal forms, all of were evidential tokens (the ‑tay morpheme attached to the verbal form). The following example illustrates the corpus of the AH participant.(11)YatiriawisataynaukathakhisasaukaawisataynasasaukatkullakapampisarataynaWise manTo tellREM/ID3>3thatwaytellThatTo tellREM/ID3>3tellthatsister3.posCOMgoREM/ID3>3‘The wise man told about that way, that way he told. Then he went with her sister, he told.’The oral production transcription showed a high frequency of-tay morphemes in the past narration, demonstrating the use of the indirect morpheme to assign the origin of the utterance in hearsay and remote narration in the heritage language. However, the direct morpheme-yä was not used to describe a discrete point during the narration.4.5Spanish Retelling Task (RT)The current analysis attempted to answer whether the HS group produced Spanish aspect morphemes with an Aymara evidential value. Specifically, the current task attempted to identify whether the participants produced the pluperfect form with an indirect evidential value in a secondhand narration. Each participant listened to a recorded Spanish narration. The retelling nature of the task was an ideal context for bilingual children to use reportative past forms. Fifty-five participants performed the task in the same three groups: 21 participants in the HS group, 15 in the AD group, and 19 in the SP group.The input story contained a total of 26 verbs, of which 13 tokens were past imperfect forms (underlined), and 13 were past perfective forms (marked below in bold) (see supplemental material); there were no tokens of present perfect forms. A descriptive analysis of the types of verbal forms produced by the participants is presented in the current section. The forms were as follows:the imperfect past.the simple past.the past imperfect progressive.pluperfect.others. This classification included the simple present (indicative and subjunctive) and the pluperfect subjunctive. No other verbal forms were produced.The number of tokens (verbal forms) distributed in the three groups is shown in Table 5.Table 5Number of verbal forms by groupGroupImperfectSimple pastImperfect past progressivePluperfectOthersHS334941103AD426911210SP5381647The maximum number of verb forms for the three groups was similar. The most frequent verb form was the simple past, demonstrating the consistent use of the simple past to express discrete points in the three groups.The analysis also showed a high frequency of the past imperfective progressive in the HS group (n=41) compared to the AH (n=11) and HCU (n=6) groups. There was a frequent mismatch between input and output among heritage children. The mismatch was using the past progressive form in the output when the input had a simple past form. The following examples show the mismatch between simple past (input) and past imperfect progressive(output) in HS participants.4.6Input Simple Past–Output Imperfect Progressive(14) input: sopló su quena‘(he) blew his flute.’output: estaba soplando la quena‘(he) was blowing his flute.’The previous example shows that bilinguals have imperfective morphology more frequently than monolinguals in the foreground context, thus suggesting that the Aymara-heritage children preferred periphrastic forms to express simple forms. Finally, the frequency of use of the pluperfect form was lower than was expected in the HS (n=10) and AD (n=21) groups; however, the frequency of the pluperfect form in the heritage groups was higher than in the SP group (n=4).5Discussion5.1Cross-Linguistic Influence (CLI) in Comprehension TasksTwo Aymara comprehension FCT s and two Spanish FCT s were conducted to identify whether there was CLI between the Aymara evidentiality and the Spanish aspect (and vice-versa) in the past bilingual discourse of the heritage children. The assumption was based on the Functional Convergence Hypothesis (Sánchez, 2004), which states that cross-linguistic influence occurs when functional features not active in one of the languages become activated in the other when the matrix in both languages is partially divergent. The results for the FCT s partially confirmed this hypothesis.The Aymara FCT see condition task (that depicted a scenario in which two characters witnessed a remote past action) showed that more than 50 % of the participants preferred the ‑tay constructions, thus providing explorative evidence of discourse-oriented aspectual features in the comprehension of the indirect evidence morpheme ‑tay. By contrast, the Aymara FCT hearsay condition (that depicted a scenario in which two characters heard what someone else did in a recent past action) showed that less than 50 % of the participants selected the direct evidential morpheme ‑yä, suggesting that the direct evidential morpheme does not encode a foreground aspect feature. Two explanations can be drawn.On the one hand, in the heritage speakers’ morphology, perceptually salient elements are better maintained and are more readily noticed (Polinsky, 2018). The morpheme ‑yä is phonetically weaker than the indirect morpheme ‑tay, thus resulting in a possibility of attrition; however, it is impossible to confirm a case of attrition based on the current results. Another explanation could be that the verbal morpheme ‑yä tends to overlap with the third-person nominal morpheme ‑na. Therefore, children could comprehend the morpheme ‑yä as a mark of third person singular instead of as a mark of a direct source of information. Following Lightfoot (2010), the input from the baseline generation contributes to the changes in the grammar of heritage speakers. Adult heritage speakers performed similarly to children in the same condition. Children received non-standard input from their parents, which resulted in ambiguity in their mental representations of the target structure. The ambiguity forces heritage speakers to (economically) reanalyze the constructions (Scontras et al., 2015), thus selecting the less ambiguous option or the indirect morpheme ‑tay.An interesting result displayed by the FCT Aymara tasks was that the proportion of expected responses by adult heritage speakers was higher than that of children in both conditions. In other words, higher scores in the tasks mean more influence from the Spanish aspect in the Aymara evidentiality representation. The former interpretation led us to think that children have a more stable Aymara representation of the source of information than their parents. It is meaningful to remember that all children’s parents are Bolivian migrants, and the children were born in Chile. Schools in Chile include Aymara L2 courses at elementary and intermediate levels (see participants). Three significant educational reforms took place in Bolivia during the twentieth century. The 1905 reform established a national education system, the 1955 reform broadened educational coverage and promoted a homogenous national culture, and the 1994 reform, which is currently in effect, restructured the educational system, encouraged popular participation, and promoted intercultural education (Taylor, 2003). The mean age of the parents was 35; therefore, it is highly likely that the adult heritage-speaking participants had yet to have formal Aymara education, unlike their children in Chile.The Spanish FCT see condition revealed that the bilingual heritage children tended to select the present perfect in the see condition instead of the simple past. Regarding the use of present perfect, there is abundant evidence of the use of compound forms with evidentiality value in Andean Spanish in Peru, Bolivia, and Ecuador (Sánchez, 2014; Ruiz, 2017; Laprade, 1981; Babel, 2009; Dankel, 2017). Therefore, the current study provides more evidence regarding using Spanish compound forms to mark evidentiality, adding comprehension data from the Aymara language in northern Chile.The Spanish FCT also revealed no tendency to use imperfect or pluperfect in the hearsay condition. In other words, participants did not perceive a difference between the two stimuli. These results could be a task effect. Fernandez (2014) examined dialectal convergence in the Andean Spanish in northern Chile. He described the features of Andean Spanish that differed from Chilean Spanish, namely the use of a lexical phrase formed with pluperfect with evidential value at the end of the utterance (me había dicho que / ‘they had told me that.’). The design of the current task did not include a lexical phrase to mark the indirect source of information. Instead, the task included the selection of an imperfect versus a pluperfect sentence (Luis se bañaba v.s. Luis se había bañado/ ‘Luis used to have a bath.’ v.s. ‘Luis had taken a bath.’) Based on that, children may not notice a relevant difference between the two experimental options, suggesting that both options satisfy the input perception.The results of the heritage adults differed from those of the experimental group. In the see condition, adults tended to select the simple past over the present perfect. One possible explanation is that responses could be based on parental perceptions regarding the use of the present perfect linked to a Bolivian variant. Fernandez (2014) assessed how Spanish-speaking migrants from the Bolivian Andes to San Pedro de Atacama (Chile) adapted their speech to Chilean Spanish. The author found a new preference for the simple past over the past perfect to express aspect, which was increasingly marked depending on the length of stay in Chile. Therefore, there is the possibility of a perception of stigmatization among Bolivian Aymara immigrants in Chile for using the present perfect as a strong marker of the Bolivian Spanish variety.In summary, the bilingual comprehension data in the current analysis partially support the hypothesis that the aspect and evidentiality features undergo CLI due to the convergence and divergence of the functional matrix tense in both target languages. On the one hand, it was possible to interpret that the participants had a mental representation of the aspectual background features encoded in the Aymara reported form -tay. However, confirming that Aymara-heritage children were assigned an aspect value to the direct evidentiality morpheme ‑yä was impossible. Furthermore, the present study clearly showed the influence of Aymara in the Spanish discourse of the Aymara-heritage children when they comprehend the direct source of information; however, the results were not compelling for the indirect source of information. Therefore, further reformulation of the methodology design is necessary to explore the target structure.5.2CLI in Production Tasks: EPT sThis analysis attempted to determine whether Aymara-heritage speakers undergo CLI between the Spanish aspect and Aymara evidentiality (and vice versa) in their bilingual oral past narrations. Two EPT s were conducted: an Aymara version performed by the Aymara-Spanish bilingual heritage participants (children and adults) and a Spanish version performed by the three groups (heritage-bilingual population and Spanish-dominant children).The Aymara EPT results showed that most participants produced verbal forms by attaching the indirect evidential morpheme ‑tay. As expected, none of the participants produced the direct evidential morpheme ‑yä. Although there was a low participation rate (only one of the 21 bilingual and five of the 19 bilingual adult participants), the experiments revealed CLI between the two target structures. In other words, Aymara-Spanish bilinguals demonstrated background aspects associated with indirect evidential morphology. The previous results supported previous descriptive studies on the relationship between evidentiality and aspect across languages. Forker and Aikhenvald (2018) described several languages with aspect-evidentiality morphology overlapping. For example, in Wanano, aspect and evidentiality are fused into verbal suffixes that convey visual evidence +perfective semantic value and morphemes that express visual evidence + imperfective semantic value. In Kashaya Pomo, the perfective and imperfective aspect feature is overtly manifested in suffixes that combine visual evidentiality features. If these categories can interact within languages, there is a possibility that the morphosyntactic exponents of Aymara evidentiality can interact with the Spanish aspect at the semantic level due to the long-lasting contact situation between the two languages.The Spanish EPT showed that the present perfect had a shallow frequency across the heritage groups; however, there was a high frequency of the past imperfect progressive verbal form in the Aymara-heritage children compared to the Spanish-dominant group. An explanation for the high frequency of the past progressive form in the heritage population may be understood as a cognitive requirement for fulfilling a morphemic gap due to the complex Aymara morphemic agglutinative structure. The use of compound forms in Andean Spanish to mark evidentiality (see the authors mentioned above) is straightforward. Therefore, it can be inferred that bilinguals who speak Spanish and Quechua or Aymara need to fill the morphological gap with a functional item to surface the source of information in their mental representations. Following Torres (2013) and her analysis of mirative constructions in Spanish and Albanian, I suggest a syntactic agreement relationship between aspect and evidentiality encoded in the aspect phrase and interpreted in the TP domain. This agreement relationship must meet + imperfective and + indirect evidentiality at the left peripheral of the Aspect phrase domain to trigger the imperfective progressive auxiliary in tense. Since no other authors have suggested using different compound verbal forms to express evidentiality, the results of the Spanish production task cannot be considered conclusive regarding the use of past progressive forms to mark evidentiality.In summary, the results of the two production EPT tasks presented partial evidence of functional convergence between two partially divergent categories: Aymara evidentiality and Spanish aspect. The Aymara task results indicated the emergence of a background value in indirect evidential morphology. On the other hand, the Spanish task results showed that the direct evidentiality features might be associated with Spanish morphology but not with the present perfect morphology. This paper proposes the use of the imperfect progressive form as a restructured marker of evidentiality due to the frequency of the form in the participants’ input. However, further studies are required to support this proposition.5.3CLI in the Production Tasks: Retelling Task (RT)The data for the Aymara RT s did not reveal marks of aspect in the evidentiality morphology. In other words, the participant did not produce the attested evidential morpheme ‑yä to describe discrete points in the past. The task results revealed that the attested form was not used in secondhand narration, thus rejecting the hypothesis concerning the use of direct evidentiality morphology with perfective aspectual value in the population. However, the task demonstrated a strong use of the morpheme ‑tay to express the second source of information in the adult population. On the other hand, the data for the Spanish RT task did not show the use of pluperfect as a marker of indirect evidentiality; however, there was a high frequency of past imperfective forms to retell discrete points in the past. Based on the EPT Spanish results (low frequency of present perfect form and high frequency past imperfective forms) and the Spanish FCT results (more than 50 % of the participants selected the present perfect option), there was an asymmetry between the comprehension and production of some grammatical features in the heritage population. The previous assumption was based on the Differential Access Hypothesis (Perez-Cortes et al., 2019), which claimed that, while some features of heritage grammars were difficult to lose in the individual mental representation, these features became more challenging to access for production due to the lack of activation and the constant competition with other grammatical structures. This account could explain the absence of the present perfect as a marker of evidentiality and the high frequency of the progressive form in the participants’ Spanish production.Regarding evidentiality, abundant data show a restructuring of the grammatical representation of the attested morpheme through the present perfect form. The present perfect form is more difficult to access due to its low frequency of use in Chilean Spanish. Therefore, the speakers used a more dominant grammar source with a similar morphological structure (imperfect progressive), which allowed them to restructure their bilingual representation of the direct Aymara evidentiality category. In other words, a bilingual representation of the Spanish morphology encodes Aymara’s evidentiality. However, the target speakers chose a more dominant linguistic option that was present in the input, which was the past progressive form.In sum, the Aymara RT tasks could not find evidence of using the Aymara direct morpheme to describe perfective actions in Aymara. On the other hand, the Spanish version of the task did not display a high frequency of pluperfect form to express an indirect source of information in Spanish past narrations. However, the tasks allow us to identify a strong mental representation of the indirect source of information within the Aymara past discourse. Furthermore, the Spanish RT task evidenced a high frequency of past imperfective forms describing discrete points in the past, suggesting that the verbal form could manifest direct evidentiality, it is necessary to extend the methodology to other participants with similar linguistic features (for example, Quechua speakers in the north of Chile).6ConclusionsThe present work has demonstrated that partially divergent structures are susceptible to CLI, extending Sánchez’s findings (2004). The analysis of the interaction between Aymara evidentiality and the Spanish aspect not only confirmed the presence of cross-linguistic influence similar to that observed in Quechua-Spanish bilinguals but also revealed new insights specific to the discussion on functional features that may undergo cross-linguistic effects. The current investigation demonstrated more evidence in favor of convergence in evidentiality features in the bilingual Spanish narratives in production tasks and emerging evidence of using Spanish compound forms to mark evidentiality. In addition, the current study employed both comprehension and production tasks, providing a more comprehensive approach to assessing bilingual performance. Differences in results across the comprehension v.s. production tasks contribute to understanding universal principles that may govern the heritage language (HL) acquisition by adding data from Aymara, which had not been previously analyzed regarding CLI. The differences between the children and the adults in the CLI tasks also point to interesting conclusions, particularly for the sociolinguistic field. The analysis suggested that preferences in adult comprehension responses were conditioned by their migrant status. Moreover, the analysis allowed for the inference that Aymara lessons in Chile facilitate the maintenance of the HL by validating the ancestral language through formal education.Thus far, there has yet to be an investigation of indigenous heritage languages in Chile, and sociolinguistic research in Chile has a secondary place in the studies concerning the Aymara people, who have received much less attention than other communities. Consequently, the current work focused on Aymara people in northern Chile as early bilinguals who acquired the ancestral and socially dominant Spanish languages early on. Therefore, the relevance of this thesis extends beyond academia, as it also has educational implications. The current project is the first step toward understanding how children acquire Aymara as an HL and Spanish as a second language, thus paving the way for developing an appropriate curriculum for the target population.Supplemental Materialhttps://github.com/jmarkovitsr/CLIAymaraSpanish to review all materials and statistics scripts used in the current research.

Journal

Journal of Language Contact: Evolution of Languages, Contact and Discourse – Brill

Published: Jul 1, 2025

There are no references for this article.