Heritage Spanish in Montreal: An Analysis of Clitics in Spontaneous Production Data

Abstract

This study investigates clitic use in adult heritage speakers (HL speakers) of Spanish, with French as their dominant language. We conducted an exploratory case study using spontaneous production data from HL speakers of Spanish and first-generation Spanish immigrants living in Montreal, Canada. Data were collected through two guided production tasks, one oral and one written, to account for task-induced performance variations. Our analysis focused on clitic production, omission, function, optionality, and grammaticality. The findings reveal both similarities and differences compared to monolingual native Spanish speakers, highlighting a broad range of clitic structures produced by all participants and a potential tendency among HL speakers to favour fixed choices in optional structures. This study provides valuable insights into the production patterns of clitics in adult HL speakers of Spanish with French as their dominant language, contributing to our understanding of bilingual grammar.

1. Introduction

Clitic pronouns represent an interesting linguistic phenomenon for the examination of bilingual1 language acquisition. They embody a diverse array of functional grammatical categories and usage patterns across different languages. This complexity is evident in the convergence of auxiliary verbs in Slavic languages, interrogative particles in Austrasic languages, and pronominal and lexicalised forms in Romance languages (Camacho Taboada 1998; Bogard 2015; among others). Understanding the acquisition of clitic pronouns in bilingual contexts holds significant relevance due to the complexity of these linguistic phenomena and the potential for cross-linguistic influence.

Bilingual speakers engage with multiple linguistic systems simultaneously, and clitic pronouns can pose a particular challenge because their properties and usage often differ across languages. Clitic pronouns exhibit intricate morphosyntactic properties, such as agreement with verbs, placement restrictions and different forms depending on their function in the sentence. While similarities exist between French and Spanish clitic pronoun systems, notable disparities can also be found alongside areas of convergence.

Previous research underscores the challenges encountered by second language (L2) learners in proficiently employing Spanish clitic pronouns, notably by monolingual first language (L1)2 speakers of languages such as English or French. These challenges encompass the complexity of the Spanish clitic system as well as interference from their L1. Bruhn de Garavito and Montrul (1996) highlight the challenges with correct placement of clitic pronouns due to influence of the L1 (e.g., clitic climbing position for L1 French speakers). The acquisition of se has proven to be particularly challenging for L2 learners, as suggested by studies such as Bruhn de Garavito (1999a, 1999b); Montrul (1999); Escobar and Teomiro (2016), and García-Tejada et al. (2021). Challenges are observed not only in basic clitic pronoun placement but also in more complex constructions such as passive, impersonal, and reflexive constructions, as noted in studies by Bruhn de Garavito (1999a, 1999b) and Tremblay (2006).

The study of clitics in heritage speakers (going forward, HL speakers) has also attracted significant attention, and some of the challenges identified for L2 speakers are also identified for HL speakers, with some notable differences regarding aspects such as configurations with clitic “se” (García-Tejada et al. 2021). Research has traditionally focused on the English–-Spanish pair (Pérez-Leroux et al. 2011; Sequeros-Valle et al. 2020; Montrul 2010; Montrul et al. 2008; García-Tejada et al. 2021; among others).

This study aims to shed light on how bilingual people use clitic pronouns, providing insights into their cognitive and linguistic representation while potentially uncovering patterns of cross-linguistic influence. Specifically, we seek to explore clitic usage patterns among adult HL speakers of Spanish residing in a predominantly French-speaking environment. The pairing of these Romance languages presents an interesting area of study due to their linguistic affinities. However, despite the potential interest, scholarly attention to this specific pairing remains limited. Noteworthy exceptions include the works of DeMelo (2014); Pérez Arreaza (2017); Pato (2022) and Alba de la Fuente et al. (2018), who have contributed valuable research in this domain.

The paper begins with an introduction to HL, followed by a brief overview of clitic pronouns, with a specific focus on their usage in Spanish and French. It then describes the clitic systems of these languages, laying the groundwork for a comprehensive study on clitic usage by HL speakers. Then, the methodology section outlines the study’s objectives and process of data collection, including a description of both elicitation tasks and corpus analysis procedures. Moving on to the results and discussion, the paper examines narratives from both HL and L1 speakers, revealing similar usage patterns with some noteworthy differences. Finally, the paper concludes with a set of concluding remarks, summarising key findings and potentially opening avenues for future research.

2. Theoretical Background

2.1. Heritage Languages

HL acquisition and maintenance, occurring within bilingual and multilingual contexts, are increasingly studied, especially within immigrant communities. HLs are spoken in families where a language different from the majority language is used, significantly influencing cultural identity and social cohesion. Defining and identifying HL speakers remains complex. The term heritage language emerged in Canada in the late 1970s, initially referring to any non-official language in the country, including Indigenous languages and those of immigrant communities (Cummins 1991, 2005). However, in contemporary North American contexts, the term refers mostly to minority languages spoken by immigrants and their descendants (Montrul 2012).

According to various scholars (Potowski and Shin 2019; Valdés 2000; Rothman 2009), bilingualism in a HL context results mainly from early exposure to a minority language in the speakers’ homes. Montrul (2023) characterises HLs as sociopolitical minority or minoritised languages, acquired either as an L1 or as one of the L1s in multilingual environments. The same author explains that the relationship between the languages spoken by bilingual individuals is determined by factors such as order of acquisition, dominance, and sociopolitical status (Montrul 2016). HL speakers form a heterogeneous group, exhibiting early bilingualism with varying outcomes (Meisel 2009; Montrul 2016). They are often considered interrupted native speakers, showing proficiency in some linguistic domains but not fully developing all grammatical features (Montrul 2012).

In the present paper, we adopt Rothman’s (2009) definition of HL, as presented in (1) below:

(1)

Heritage language (HL):

A language qualifies as a heritage language if it is a language spoken at home or otherwise readily available to young children, and crucially this language is not a dominant language of the larger (national) society… [A]n individual qualifies as a heritage speaker if and only if he or she has some command of the heritage language acquired naturalistically… although it is equally expected that such competence will differ from that of native monolinguals of comparable age. (Rothman 2009, p. 156)

Among the unique characteristics that distinguish HL speakers from other language users, Montrul (2016) identifies the following traits listed in (2):

(2)	Characteristics of HL speakers:
	a.	They have been raised in bilingual households and possess linguistic competence in two languages.
	b.	Their L1 (or one of their L1s) spoken at home is a minority language.
	c.	They are usually proficient in the majority language, usually with native or native-like proficiency.
	d.	They tend to be less proficient in the HL, whose level of proficiency ranges from minimal and receptive ability to full fluency and native proficiency.

In the broader context of bilingual acquisition studies, the exploration of HLs holds significant relevance. Understanding the acquisition and maintenance of HLs provides insights into the intricate dynamics of bilingual and multilingual development within multilingual contexts (Scontras et al. 2015; Domínguez 2009; among others). HL speakers navigate a linguistic landscape where their HL coexists with the majority language, presenting unique challenges and opportunities (Wiese et al. 2022; Polinsky 2018; Montrul 2016; Scontras et al. 2015; among others). By examining the use of linguistic phenomena in HL speech, either written or oral, our understanding of bilingual acquisition processes can be enriched (Baal and Natvig 2021; Lohndal et al. 2019; Benmamoun et al. 2013; among others). Moreover, insights gleaned from HL studies can inform educational policies and practices aimed at fostering linguistic diversity and supporting HL maintenance among bilingual populations (Carreira and Kagan 2018). Thus, the study of HLs contributes to a more comprehensive understanding of bilingual acquisition and underscores the importance of linguistic heritage in shaping individuals’ linguistic repertoires.

2.2. Clitic Pronouns

Cardinaletti and Starke’s (1999) influential work offers a comprehensive framework for understanding the structural hierarchy of pronouns, delineating strong, weak, and clitic categories based on their syntactic properties. Strong pronouns, like English she, he or Spanish él, nosotros, characterised by rich syntactic structure, contrast with weak and clitic pronouns, which exhibit progressive reduction in complexity. Clitics, like Spanish se and me, are reduced forms of full pronouns, involving less syntactic structure that results in some form of syntactic defectiveness (Cardinaletti and Starke 1999; Fernández Soriano 1993, 2016; Zwicky 1977). This reduction in syntactic structure makes clitic pronouns particularly dependent on adjacent elements in the sentence, often requiring a host to attach to, thus further distinguishing them from their strong and weak counterparts. Fernández Soriano (2016) refers to clitics as forms that resemble words but are not independent, since, instead, they depend on contiguous elements. Furthermore, Bogard (1999, 2015) describes them as constituents that expand the host word.

Clitic pronouns have morphosyntactic (person, number, gender, and case) and phonological features, they lack stress and have a deficient prosodic status, and they occur in fixed positions near verbs, among other properties (Zwicky 1977; Fernández Soriano 1993; Halpern 1998; among others). Ordóñez (2012) compiles some of them in the following list:

• Coordination: Clitic pronouns, unlike strong pronouns, cannot be coordinated, as shown in examples (3a) and (3b).

  (3)

  a.

  Nosotros

  y

  vosotros

  fuimos

  al

  cine

  We

  and

  you-PL

  go.1.PL.PAST

  to-the

  cinema

  ‘We and you went to the cinema.’

  b.

  *Los

  y

  las

  compramos

  ayer.

  3.ACC.PL.MASC

  and

  3.ACC.PL.FEM

  buy.1PL.PAST

  yesterday

  ‘We bought them yesterday.’

• Modification: Unlike strong pronouns, clitics cannot be modified, as shown in examples (4a) (adjective modification) and (4b) (adverbial modification).

  (4)	a.	*{beau; rapide; …} il	(Cardinaletti and Starke 1999, p. 151)
  		*{beautiful; quick; …} il
  	b.	*{vraiment; seulement; …} il
  		*{really; only; …} il

• Emphasis: Clitics cannot be emphasised, while pronouns can be easily emphasised, as in (5a) and (5b).

  (5)	a.	La	saludé	en	el	CINE.
  		3.ACC.SG.FEM	greet.1SG.PAST	in	the	cinema
  		‘I greeted her at the cinema.’
  	b.	Me	encontré	con	ELLA.
  		1.REFL.SG	meet.1SG.PAST	with	her
  		‘I met her.’

• Isolation: Unlike strong pronouns, clitics cannot appear in isolation as a response to a question, as shown in example (6).

  (6)	¿A	quien		llamaste?
  	To	who		call.2SG.PAST
  	‘Who did you call?’
  	a.	*La.
  		3.ACC.SG.FEM
  		‘Her.’
  	b.	A	ella.
  		To	she
  		‘Her.’

2.2.1. Clitics in Spanish

The Spanish clitic system is characterised by its complexity in morphological forms, syntactic placement, and variations across different dialects and geographical regions.

According to Nueva gramática de la lengua española (Real Academia Española and Asociación de Academias de la Lengua Española (RAE-ASALE) 2009), the Spanish clitic system organises personal pronouns based on grammatical person, case, and tonicity. Pronouns are categorised into different types based on case, including nominative (yo, tú), prepositional or oblique (mí, ti, sí, conmigo, contigo, consigo), accusative (lo, la, los, las), dative (le, les), pronouns without distinction between accusative and dative case (me, te, se, nos, os), and pronouns without a specific case distinction (nosotros, vosotros, usted, etc.). Additionally, there exist reflexive (me, te, se, nos, os) and neuter pronouns (lo).

Regarding tonicity or cliticity, Spanish distinguishes between tonic pronouns, which are independent forms, and clitics, which lack phonetic independence and always accompany the verb they depend on morphophonologically. Furthermore, Spanish allows for clitic doubling, which means that clitic pronouns can appear alongside their corresponding full noun phrases or pronouns for emphasis or clarification. This can happen with either accusative clitics or with dative clitics. Additionally, reflexive pronouns permit doubling by adding the particle mismo.

The placement of clitics can vary, with some attached to the verb’s base (enclitics) and others preceding the verb (proclitics). There are also rules governing the sequences of clitics and variations observed in different dialects, including phenomena like leísmo, laísmo, loísmo, and other syntactic and semantic variations across geographical regions (Fernández Soriano 2016).

2.2.2. Clitics in French

Granfeldt (2014) delineates two main categories of pronouns within the French language, drawing from the framework established by Cardinaletti and Starke (1999): strong pronouns and clitics. While formal French predominantly features distinct morphologies for strong pronouns and clitics, informal registers blur this boundary, particularly regarding subject pronouns which are often considered clitics. Granfeldt (2014) highlights exceptions such as nous and vous, where both strong and weak subject pronouns exist. Notably, accusative and dative clitics in French typically overlap, except for in the third person, where the distinct dative forms lui/leur exist. Furthermore, Granfeldt (2014) observes that accusative clitics of the third person coincide phonologically with definite articles (le, la, les).

Choi-Jonin and Lagae (2015) provide an in-depth exploration of the French clitic system, characterising clitics as intermediate linguistic elements positioned between independent words and affixes. These entities exhibit distinctive phonological and syntactic properties. Phonologically, clitics lack inherent stress and amalgamate with a host word to form a cohesive prosodic unit. They encompass various elements such as unstressed pronouns, determiners, prepositions, and the negative morpheme ne. Syntactically, clitics adhere to the canonical SVO order, with subject clitic pronouns positioned before the verb, similar to lexical constituents. However, they cannot be separated from the verb by a non-clitic element, distinguishing them from weak pronouns.

Rodriguez-Mondoñedo et al. (2005) highlight that French clitics cannot move to higher syntactic positions, contributing to their distinct syntactic behaviour, nor can they be coordinated as shown in (7). While subject clitic pronouns can coordinate two verbal phrases, they must be in a preverbal position to do so, as represented in (8).

(7)	a.	Je	vois		Marie		et		Paul.
		I	see.1SG.PRES		Marie		and		Paul
		‘I see Marie and Paul.’
	b.	*Je	le	vois		et		Marie.
		I	him	see.1SG.PRES		and		Marie
		‘I see him and Marie.’
	c.	*Je	le	et		la		vois.
		I	him	and		her		see.1SG.PRES
		‘I see him and her.’

(8)	a.	Il	chante		et		danse.
		He	sing.3SG.PRES		and		dance.3SG.PRES
		‘He sings and dances.’
	b.	*Chante-t-il		et		danse?
		Sing.3SG.PRES-he		and		dance.3.SG.PRES
		‘Does he sing and dance?’

2.2.3. Contrasting Clitic Systems: Spanish vs. French

The clitic systems of Spanish and French exhibit notable differences, reflecting divergent grammatical structures and usage conventions.

Firstly, Spanish clitics predominantly serve as object pronouns, except for the impersonal clitic se. Conversely, French encompasses both subject and object pronouns within its clitic inventory (Zagona 2002; Ordóñez 2012). Secondly, French presents a broader range of clitic types compared to Spanish. Alongside accusative and dative pronouns, French includes reflexive, partitive (en), and locative (y) clitics. These additional clitics facilitate the replacement of quantity expressions and locative phrases, functionalities absent in the Spanish clitic system (Colomina Samitier 2016). Thirdly, French employs distinctive or emphatic pronouns, resulting in subject doubling (Fernández Soriano 1989). However, this construction, shown in (9), does not introduce a genuine contrast in the sentence in French:

(9)	Je	parlerai,	moi	(Fernández Soriano 1989, p. 178)
	I	speak.1SG.FUT,	1.SG
	‘I	will	speak’

Moreover, while Spanish permits clitic climbing, allowing the clitic pronoun to ascend to a higher position in the sentence for emphasis or focus, as in (10a), such syntactic manoeuvring is not permitted within French clitic syntax, as in (11) (Rodriguez-Mondoñedo et al. 2005).

(10)	a.	Juan	lo	quiso	decir.	(clitic climbing)
		Juan	3.SG.ACC.MASC	want.3.SG.PAST	say.INF
		‘Juan wanted to say it.’
	b.	Juan	quiso	decirlo.		(no clitic climbing)
		Juan	want.3SG.PAST	say.INF-3.SG.ACC.MASC
		‘Juan wanted to say it.’
(11)	a.	Jean	a voulu	le	dire	(no clitic climbing)
		Jean	want.3SG.PAST	3.SG.ACC.MASC	say.INF
		‘Jean wanted to say it.’
	b.	*Jean	le	a voulu	dire.	(clitic climbing)
		Juan	3.SG.ACC.MASC	want.3SG.PAST	say.INF
		‘Jean wanted to say it.’

Another notable distinction lies in the arrangement of clitic pronouns within clusters, particularly regarding the order of dative (DAT) and accusative (ACC) pronouns. In Spanish, clitic pronouns typically appear in a DAT + ACC order within clusters (Real Academia Española and Asociación de Academias de la Lengua Española (RAE-ASALE) 2009), as illustrated in (12a). In contrast, French follows an ACC + DAT order for clitic pronouns within clusters (Choi-Jonin and Lagae 2015), as illustrated in (12b), which is a reversed sequence compared to Spanish.

(12)	a.	Él	se	lo	da.
		He	3SG.DAT	3SG.ACC	give.3.SG.PRES
		‘He gives it to her.’
	b.	Il	le	lui	donne.
		He	3SG.ACC	3SG.DAT	give.3.SG.PRES
		‘He gives it to her.’

Finally, the French clitic system encompasses the unique pronoun on, as shown in (13), which lacks a direct counterpart in Spanish. Functioning as a vague pronoun, on refers to an unspecified individual or group without specificity. Despite sharing similarities with personal pronouns, on is exclusively singular and employed solely as a subject (Grevisse and Goose 1995).

(13)	En	France,	on	aime	bien	manger.
	In	France,	on	like.3SG.PRES	well	eat.INF
	‘In France, people like to eat well.’

2.3. Clitic Acquisition by Heritage Speakers

The acquisition of clitics by HL speakers has been a subject of interest in recent linguistic research. Previous studies have identified several key aspects regarding clitic acquisition by HL speakers.

Sequeros-Valle et al. (2020) found that while HL speakers may overextend clitic-doubled left dislocation in non-anaphoric contexts during acceptability judgement tasks (maybe due to the metalinguistic nature of such tasks, vid. infra), they otherwise exhibit similar production patterns to L1 speakers. The authors propose that differences in clitic usage between HL speakers and L1 speakers may be task-induced, as an effect of the task’s metalinguistic demands.

Pérez-Leroux et al. (2011) suggest that HL speakers may exhibit variations in clitic placement influenced by exposure to English, indicating the presence of syntactic transfer effects in bilingual acquisition. Their study on bilingual children in Canada revealed distinct bilingual patterns: a tendency towards backward repositioning of clitics in the sentence structure, occasional omissions observed particularly among simultaneous bilinguals, and a reduction in the bias towards forward repositioning.

García-Tejada et al. (2021) find that HL speakers demonstrate advantages over L2 learners in the use and interpretation of the clitic se with specific verb categories in Spanish. Specifically, L2 learners encountered greater challenges when using se with change of state verbs compared to HL speakers. This difficulty stems from the absence of positive transfer from their L1, which typically lacks similar morphosyntactic structures. Conversely, both HL speakers and L2 learners found se with psychological verbs in declarative sentences easier, benefiting from analogous constructions in English. However, both groups encountered difficulties with se in interrogative contexts, particularly with psychological verbs.

Martín Gómez’s (2022) study reveals that the age of initial exposure to Spanish and the conditions of language input do not significantly influence HL speakers’ ability to produce clitic clusters in pro-clitic positions. Instead, factors such as proficiency level and amount of exposure appear more crucial in clitic climbing acquisition among bilingual populations. Despite their diverse language learning backgrounds, both HL speakers and L2 learners demonstrated proficiency in constructing these complex structures.

López Otero et al. (2023a) provide evidence of overextension of object clitics among Brazilian Portuguese-speaking HL speakers of Spanish, particularly in contexts where null clitics would typically be expected. Their study shows that HL speakers with less frequent use of their HL are more likely to use overt clitics incorrectly in cases where indefinite and non-specific antecedents are involved, as in (14).

(14)

Rosa

no

gasta

mucho

dinero

(López Otero et al. 2023a, p. 174)

Rosa

not

spend.3SG.PRES

much

money

en

ropa,

pero

zapatos

sí

los

compra.

in

clothes

but

shoes

yes

3SG.ACC.MASC

buy.3SG.PRES

‘Rosa does not spend much money on clothes, but shoes she buys.’

These works contribute to understanding how HL speakers acquire clitic knowledge in Spanish, highlighting the dynamic interplay between linguistic exposure, language dominance, and transfer effects in bilingual language development. However, despite the considerable attention given to the Spanish–English language pair—with most of the studies cited above focusing on this combination—there exists a gap in research exploring other language pairs. The scarcity of studies in this area highlights the need for further research to understand the dynamics of clitic acquisition in HL speakers with different dominant languages, particularly in underrepresented language combinations like Spanish–French.

3. Materials and Methods

3.1. Goals of the Study

This study aims to examine how HL speakers of Spanish, whose dominant native language is French, produce clitic pronouns. Specifically, it seeks to explore the patterns of clitic usage among these speakers residing in Montreal.

3.2. Sample

Two groups of participants were involved in the study:

• HL Speakers of Spanish Group (N = 10): This group comprised adult speakers of Spanish as their HL, all of whom were second-generation immigrants. Within this group, there were the following:
  • Simultaneous bilingual participants (N = 4) (balanced gender distribution; age range 21–25, mean age: 22.5 years): participants exposed to both French and Spanish from birth.
  • Sequential bilingual participants (N = 6) (2 males and 4 females; age range 18–26, mean age: 22 years): participants exposed to Spanish first and then to French between the ages of 4 and 7.

Participants in this group were speakers of diverse Spanish dialects as their mother tongue, with French being their dominant language. All HL participants had acquired Spanish from birth and none of them were schooled in Spanish or had received formal Spanish instruction during childhood. To assess their Spanish proficiency, they took an adapted version of the Diploma de Español como Lengua Extranjera (DELE, Certificate of Spanish as a Foreign Language) exam. This exam, widely used in L2 acquisition studies (White et al. 2004; Montrul et al. 2008; among others), demonstrated their advanced-level proficiency, as all participants scored 40 or above out of a maximum of 50 points (score range for the simultaneous bilingual group: 42–46, mean score: 44; score range for the sequential bilingual group: 40–48, mean score: 43.8).

Table 1. Sociodemographic data of the HL group.

Code	Age	AofB	Country	L1	Heritage Type	Gender	Score DELE
SIM001	22	0	Argentina	SP/FR	SIMULTANEOUS	F	46
SIM002	21	0	Chile	SP/FR	SIMULTANEOUS	M	42
SIM003	25	0	Mexico	SP/FR	SIMULTANEOUS	F	43
SIM004	22	0	Mexico	SP/FR	SIMULTANEOUS	M	45
SEC001	26	4	Mexico	SP	SEQUENTIAL	M	40
SEC002	18	6	Colombia	SP	SEQUENTIAL	F	43
SEC003	20	4	Mexico	SP	SEQUENTIAL	F	45
SEC004	23	5	Mexico	SP	SEQUENTIAL	F	45
SEC005	21	5	Mexico	SP	SEQUENTIAL	F	48
SEC006	24	6	Chile	SP	SEQUENTIAL	M	42

summarises the sociodemographic data of these participants, including the age of onset of bilingualism (AofB).

2.

Speakers of Spanish as L1 (N = 10; balanced gender distribution; age range 24–70, mean age: 47): This group consisted of adult L1 speakers of Spanish, representing different dialects. Participants in this group had been living in Montreal for 5 years or less. They may have had French, English, or another Romance language as their L2, but all participants in this group were born and raised as monolingual speakers of Spanish in their respective countries of origin, and Spanish remains their dominant language.

3.3. Data Collection

The linguistic data for this study were obtained from an existing corpus, which has been used in previous research such as Cruz Enríquez (2019) and Cruz Enriquez and Alba de la Fuente (2024).

3.3.1. Elicitation Tasks

The data collection process encompassed two semi-guided production tasks, comprising both oral and written components. The inclusion of both oral and written production tasks in the data collection process was motivated by previous research, which has identified performance variations based on task modality and type. Typically, HL speakers acquire their HL at home, often without formal instruction (Montrul 2020). This lack of formal education tends to result in stronger performance on tasks that measure implicit knowledge, such as oral tasks, as evidenced by prior studies (cf. Bowles 2011; Montrul 2012; Montrul et al. 2008; Iranzo 2022). To gain a comprehensive understanding of their language proficiency and usage patterns, it is valuable to incorporate both oral and written tasks. Hence, we aim to examine both written and oral production. This approach not only fulfils the need to assess written production but also recognises the intrinsic significance of oral communication within the typical context of HL speakers.

The elicitation tasks consisted of a written video description task and an oral video description task. Specifically, participants were asked to narrate the plot of an 8 min segment from Modern Times (Chaplin 1936).

3.3.2. Corpora

Two corpora were compiled from the two elicitation tasks: one from HL speakers (N = 10) and another one from L1 speakers (N = 10). The total word count for the HL speakers’ corpus is 8031 words, comprising 3613 words from written tasks and 4418 words from oral tasks. In contrast, the L1 speakers’ corpus encompasses a total of 10,851 words, comprising 4644 words from written tasks and 6207 words from oral tasks. This collection of linguistic data facilitated an in-depth investigation into clitic pronoun usage.

3.3.3. Procedure

The video description tasks were conducted via a specialised webpage, crafted to adhere to established research protocols (Cruz Enríquez and Alba de la Fuente 2024; Cruz Enríquez 2019), ensuring standardised administration and data collection procedures. Upon completion of the oral tasks, participants’ spoken narratives were transcribed. The collected oral and written production data underwent systematic analysis, which involved a comprehensive examination of various coding parameters. Specifically, the data were dissected into individual clauses to facilitate a detailed exploration of clitic presence and absence. The analysis included identifying instances of clitic doubling or its absence, determining clitic functions such as accusative and dative.3 In addition, a category of pronominal verbs, encompassing several verb types, was also included in the analysis. The classification of pronominal verbs employed in this study is an elaboration of the one found in Teomiro (2017) and includes the following:

• Alternating inherent pronominal verbs:4 These verbs can occur with or without the pronominal particle and do not participate in the causative alternation. They may function as either transitive or intransitive verbs and the presence of the particle may involve changes in the argument structure of the predicate.5 Examples include encontrar(se) (to find) and llevar(se) (to take).
• Non-alternating inherent pronominal verbs: These verbs necessitate the pronominal particle and do not undergo causative alternation. For example, desmayarse (to faint) and ponerse (a hacer algo) (to start).
• Movement pronominal verbs: These verbs, except ir(se) (to go), which changes meaning, exhibit alternation and imply movement. For instance, salir(se) (to get out).
• Anticausative pronominal verbs: These verbs may undergo causative alternation, and in the resulting structures, the presence of the pronominal particle is obligatory. For example, derretirse (to melt) and romper(se) (to break).
• Non-anticausative pronominal verbs: These are intransitive verbs that do not undergo causative alternation and can occur with or without the pronominal particle without any change in the argument structure of the predicate.6 Examples include caer(se) (to fall) and morir(se) (to die).
• Consumption pronominal verbs: These verbs alternate and entail either material or psychological consumption. For instance, comer(se) (to eat) and fumar(se) (to smoke).
• Reflexive and reciprocal verbs: These are agentive verbs whose argument structure involves co-referentiality between subject and object. For example, lavarse (wash oneself) and saludarse (greet each other).

Furthermore, distinctions were made for non-paradigmatic se, discerning between passive reflexive and impersonal uses (Mendikoetxea 1999). Additionally, the presence of oblique pronouns was examined, and co-referentiality with the corresponding co-referent element was assessed. Furthermore, the optionality of clitic use and potential geographical variability within the Spanish language were considered.7 Finally, the analysis encompassed evaluating the grammaticality or ungrammaticality of each token, distinguishing between production, omission, and agreement mismatches.

Data codification was conducted by three judges according to predetermined parameters. Subsequently, a comparative analysis contrasting HL speakers’ data with those of L1 speakers was performed, offering insights into clitic usage trends among HL speakers of Spanish with French as their dominant L1. When relevant to the data analysis, the HL group was further divided into two subgroups (simultaneous HL speakers and sequential HL speakers). To assess the significance of observed trends and differences within the data, statistical tests were performed, specifically Fisher’s exact test. This test’s suitability for our study lies in its robustness with small sample sizes and its accuracy in analysing categorical data resulting from multiple classifications. Given the nature of our dataset, Fisher’s exact test serves as a reliable method to determine the significance of associations between different linguistic classifications. We performed separate statistical analyses for each category and modality (written vs. oral), as well as considering all data combined.8

4. Results

4.1. Native Speakers

The analysis of native speakers’ performance reveals a diverse array of clitic forms, as illustrated in Figure 1.9

The vast majority of clitics in the corpus are in the third person, with very few instances of first-person (15a) and second-person (15b) clitics, appearing only in the oral narratives. This distribution is expected given the nature of the task, which consisted of narrating a series of events depicted in a video.

(15)	a.	[…]	que		él	lo				pagaría
			that		he	3SG.ACC.MASC				pay.3SG.COND
		me		imagino.					[…]
		1SG.REFL		imagine.1SG.PRES
		‘[…] that he would pay for it, I imagine’ [ESL1011_Oral:32]
	b.	[…]	aprovecha				y	escápate			ahora.
			seize.2SG.PRES				and	escape.2SG.PRES-2SG.REFL			now
		‘[…] seize the opportunity and escape now […]’ [ESL1016_Oral:23]

The most frequently used clitic is se, with 190 productions and 47 omissions. This high frequency is anticipated, as se appears in a wide range of syntactic configurations, including various types of pronominal verbs.

Additionally, the data showcase clitics used in a variety of syntactic configurations, represented in Figure 2.

Only one instance of ungrammaticality was identified among the data analysed, specifically an agreement error, reproduced in (16), which shows a mismatch between le, singular, and unos niños, plural.

(16)	[…]	se	fumó		uno		y	le	dio
	3.REFL		smoke.3SG.PAST		one		and	3SG.DAT	give.1SG.PAST
	el	resto	a	unos		niños.
	the	rest	to	Indef.PLURAL		kids
	‘[…] he smoked one and gave the rest to some kids.’ [ESL1001_ESC:11]

All instances of omission are grammatical and appear in the following configurations: datives (in the context of clitic doubling) (17) and pronominal verbs. In the latter, omission appears in contexts where the clitic can be optionally included without affecting grammaticality. These contexts include inherent alternating pronominal verbs (18), movement pronominal verbs (19), non-anticausative pronominal verbs (20), and consumption pronominal verbs (21):

(17)

a.

[…]

vio

lo

ocurrido

y

Ø = le

see.3SG.PAST

the

happen.PARTICIPLE

and

Ø = 3SG.DAT

alertó

del

robo

al

dueño.

alert.3SG.PAST

of-the

theft

to-the

Owner

‘[…] she saw what happened and alerted the owner of the theft.’ [ESL1010_ESC:3]

b.

[…]

y

una

mujer

que

pasó

por

allí

and

one

woman

that

pass.3SG.PAST

by

there

Ø = lo

vio

todo.

Ø = 3SG.ACC.MASC

see.3SG.PAST

all.

‘[…]and a woman passing by saw it all.’ [ESL1010_ORA:8]

(18)	[…]	enseguida	lo	record	y
		inmmediatly	3SG.ACC.MASC	remember.3SG.PAST	and
		Ø = se	quedó	pensativa.
		Ø = 3.REFL	stay.3SG.PAST	Thoughtful
	‘[…] she immediately remembered and thought about it.’ [ESL1002_ORA:39]

(19)	[…]	en	eso	llegó	el	camión	y	Ø = se
		in	that	arrive.2SG.PAST	the	truch	and	Ø = 3SG.REFL
	subió.
	get.3SG.PAST-in
	‘[…] then the truck arrived and he got in.’ [ESL1011_ORA:36]

(20)	[…]	y	los	dos	Ø = se	cayeron	al	suelo.
		and	the	two	Ø = 3.REFL	fall.3PL.PAST	to-the	ground
	‘[…] and both of them fell on the ground [ESL1008_ESC:2].’

(21)	[…]	le	encendió	y		Ø = se		lo
		3SG.DAT	lit.3SG.PAST	and		Ø = 3.REFL		3SG.ACC.MASC
	fumó		exhalando		bastante		humo.
	smoke.3SG.PAST		exhale.GERUND		quite		smoke
	‘[…] he lit it and smoked it, exhaling a lot of smoke.’ [ESL1008_ESC:12]

Inherent alternating and movement pronominal verbs showed an almost even distribution between production and omission, with omission rates of 46% and 45.4%, respectively. Non-anticausative and consumption pronominal verbs exhibited even higher omission rates (84% and 29%, respectively), albeit with a limited number of occurrences in the corpus, leading to their exclusion from detailed analysis.

Regarding clitic doubling, a total of 38 instances were identified, predominantly occurring in accusative or dative contexts. Overall, a low rate of omission (seven cases) was observed in doubling contexts, with slightly higher rates noted in dative contexts, which were produced more frequently overall, as shown in Figure 3.

Finally, the patterns of production and omission observed in both oral and written contexts exhibit remarkable similarity, as illustrated in Figure 4. Results from Fisher’s exact test indicate no significant differences between oral and written production of accusative and dative clitics (p = 0.6618) or pronominal verbs (p = 0.1923) among the L1 group.

4.2. Heritage Speakers

Similarly to L1 speakers, HL speakers exhibit a variety of clitic forms, as illustrated in Figure 5.

Also, as in the group of L1 speakers, the majority of clitics in the corpus are in the third person, with very few instances of first-person (22) and second-person (23) clitics, appearing only in direct speech contexts. This distribution aligns with the nature of the task, which involved narrating a series of events depicted in a video, as previously explained.

(22)	[…]	y,	caballerosamente,				dijo:		«yo	me
		and,	chivalrously,				say.3SG.PAST		«I	1SG.REFL
	robé			el	pan,	no		ella.»
	steal.1SG.PAST			the	bread,	not		her»
	‘[…] and, chivalrously, he said: «I stole the bread, not her»’ [SEQ007_ESC:11]

(23)	[…]	tú	también	te	puedes	escapar.
		you	too	2SG.REFL	can.3SG.PRES	escape.INF
	‘[…] you too can escape.’ [SIM003_ORA:32]

Se is the most frequently used clitic, with 271 productions and 37 omissions. This high frequency is expected, as se appears in a wide range of syntactic configurations, represented in Figure 6.

In terms of clitic combinations, we found several instances in both the HL and L1 groups. These were typically combinations of se + accusative, as in “se lo llevó”, ‘he took him away’. We did not find any non-target usages of this type of structure in the data, so, in this regard, our results do not evidence cross-linguistic influence from French, as HL speakers do not behave differently than native speakers with regard to clitic clusters.10

There are more instances of ungrammaticality among HL speakers compared to L1 speakers, though the overall number of such cases remains low. No significant differences were found between the production of accusative and dative clitics in L1 and heritage speakers (p = 0.2778). However, a significant difference was observed in the production of pronominal verbs (inherent alternating, inherent non-alternating, and movement pronominal verbs) between the two groups (p < 0.01).

Ungrammatical productions, which are more frequent than ungrammatical omissions, primarily consist of number agreement errors like (24), similar to those identified in the L1 speakers’ data.

(24)	[…]	empezó			a		mirar		cosas,	le
		start.3SG.PAST			to		loo.INF		things,	3SG.DAT
	regaló		a	unos		niños		también
	gift.3SG.PAST		to	ones		kids		too
	‘[…] he started looking at things, he also gifted some children’ [SIM001_ORA:17]

Grammatical omissions appear predominantly in dative doubling contexts (25) and pronominal verbs, including inherent alternating (26), movement (27), non-anticausative (28), and consumption (29) pronominal verbs:

(25)	[…]	y	comenzó	a	regalár-Ø = se-los
		and	start.3SG.PAST	to	gift-Ø = 3.REFL-3PL.ACC.MASC
	a	unos	niños.
	to	ones	kids
	‘[…] and he started to gift them to some children.’ [SEQ001_ORA:16]

(26)	[…]	para	que	ella	pudiera	escapar-Ø = se.
		for	that	she	can.3SG.PAST.SUBJ	escape.INF-Ø = 3.REFL
	‘[…] so that she could escape.’ [SEQ006_ESC:18]

(27)	[…]	la	chica	también	Ø = se	subió	al	carro
		the	girl	too	Ø = 3.REFL	get in.3SG.PAST	to-the	wagon
	‘[…] the girl also got on the wagon.’ [SEQ005_ESSC:10]

(28)	Entonces,	los	dos	Ø = se	cayeron	del	auto.
	Then,	the	two	Ø = 3.REFL	fall.3PL.PAST	of-the	car
	‘Then, both fell out of the car’ [SEQ003_ORA:25]

(29)	[…]	para	fumar-Ø = se	un	cigarro.
		to	smoke.INF-Ø = 3.REFL	one	cigarrete
	‘[…] to smoke a cigarette.’ [SEQ006_ESC:12]

As illustrated in Figure 6, most omissions are grammatical, with only a few instances of ungrammatical omission, primarily in accusative (30) and dative (31) doubling contexts, and infrequently in alternating inherent pronominal verbs (32):

(30)	[…]	intenta	arrestar-Ø = los	a	los	dos.
		try.3SG.PRES	arrest.INF-Ø = 3PL.ACC.MASC	to	the	two
	‘[…] tries to arrest them both.’ [SIM002_ORA:14]

(31)	[…]	y	Ø = se	lo	dijo	al	señor.
		and	Ø = 3.REFL	3.SG.ACC.MASC	say-3SG.PAST	to-the	man
	‘[…] and he told the man.’ [SIM001_ESC:2]

(32)	Entonces,	el	policía	Ø = se	llevó	al	señor
	Then,	the	policeman	Ø = 3.REFL	take.3SG.PAST	to-the	man
	en	lugar	de	La	mujer.
	in	place	of	the	woman
	‘Then, the policeman took the man instead of the woman.’ [SEQ001_ES:8]

In the analysis of pronominal verbs, a notable contrast with L1 speakers is observed in configurations that allow for optionality, specifically in inherent alternating, movement, and non-anticausative pronominal verbs (p < 0.01). HL speakers exhibit fewer cases of omission compared to L1 speakers, with omission rates of 17.85% for inherent alternating verbs, 25% for movement verbs, 20% for non-anticausative verbs and 46% for consumption verbs.

Similarly to the L1 group, several instances of clitic doubling were identified among heritage speakers (31 cases), see Figure 7, occurring in either accusative or dative contexts. Overall omission rates in doubling contexts remain low, comparable to those in L1 speakers (nine cases), with higher omission rates for dative clitics, likely due to their higher overall production. Ungrammatical cases in doubling contexts are infrequent and predominantly consist of agreement errors.

The patterns of production and omission in oral and written contexts among HL speakers are largely similar, as depicted in Figure 8. Fisher’s exact test results show no significant differences between oral and written production of dative and accusative clitics within the heritage speaker group (p = 0.9009 for sequential HL speakers, p = 1 for simultaneous HL speakers). Additionally, no significant differences were found in the overall production of pronominal verbs between oral and written contexts (p = 0.1923 for simultaneous HL speakers, p = 0.5718 for sequential HL speakers). These findings align with those of Alba de la Fuente et al. (2018), who reported no significant differences in task type among L2 learners and HL speakers of Spanish with French as their dominant language in Quebec.

Comparing alternating and non-alternating inherent pronominal verbs, significant differences in usage patterns emerge (p < 0.05), indicating distinct tendencies between L1 and HL speakers in both production and omission. L1 speakers demonstrate a higher incidence of omissions compared to HL speakers. Both L1 and sequential HL speakers show disparities in the use of these verbs across oral and written contexts (p < 0.001).

Furthermore, significant differences between the sequential and simultaneous HL speaker subgroups in alternating contexts were observed. Sequential HL speakers rarely omit in written language but do so more frequently in oral language. In contrast, simultaneous HL speakers exhibit more omissions in written language and fewer in oral language, as shown in Figure 9. This pattern differs from that of L1 speakers, where omissions were similar across modalities with no significant differences (vid. Figure 2). The difference between HL subgroups is significant (p < 0.001), but due to the limited data and small group sizes, further explanations remain speculative.

No additional significant differences were detected between the sequential and simultaneous HL subgroups.

5. Discussion

The results indicate a largely similar pattern of clitic use between L1 and HL speakers, with both groups producing a variety of clitic forms across various syntactic configurations. Consistent with the nature of the task, third-person clitics were the most frequent in both groups. Despite the inverted order of clitic pronouns within clusters in Spanish and French—where Spanish typically follows a DAT + ACC order and French uses an ACC + DAT order—our research did not identify any significant challenges or difficulties arising from this contrast between the two languages. However, significant differences emerged when comparing the production and omission ratios of clitics with inherent alternating pronominal verbs between L1 and HL speakers (p < 0.001). HL speakers exhibited less variability, manifested as fewer instances of clitic omission, in these contexts, suggesting a more cautious approach to avoid errors. Regarding clitic doubling, HL speakers aligned closely with L1 speakers, indicating that clitic doubling does not pose a particular challenge for HL speakers.

Our data do not show evidence of cross-linguistic influence in structures where the two systems do not overlap, such as clitic climbing and word order, in our sample. As indicated above, the only divergences between the two groups surfaced in configurations where optionality is allowed.

In summary, our findings reveal striking similarities between HL and L1 Spanish speakers across most aspects examined. Significant deviations emerged primarily in contexts allowing for optional usage, specifically with inherent alternating verbs. This suggests a cautious approach among simultaneous HL speakers, who demonstrate a reluctance to risk errors in speech, contrasting with their performance in writing. Remarkably, oral narratives by simultaneous HL speakers exhibited no omissions of alternating inherent pronominal verbs, contrasting starkly with L1 speakers, where approximately half of these clitics were omitted. Conversely, written narratives by simultaneous HL speakers showed noticeable omissions in these types of pronominal verbs. Moreover, simultaneous HL speakers displayed no omissions with alternating inherent pronominal verbs in oral narratives, whereas a few ungrammaticalities were found with non-alternating inherent pronominal verbs, suggesting a slight propensity for confusion in this category of clitics. In contrast, while sequential HL speakers frequently omitted alternating inherent pronominal verbs in their written samples, no omissions were observed in oral contexts for this group, highlighting a reverse pattern compared to simultaneous HL speakers.

All things considered, these results align with previous research findings, such as those of Thomas (2012), studying optionality in the context of clitic climbing in L1, L2, and HL speakers of Spanish. The study suggests that L1, L2, and HL speakers exhibit similarities and differences in their handling of optional grammatical features. Particularly, Thomas (2012) reveals that all groups, including monolingual and bilingual people, tend to limit optional grammatical features in their speech. In this sense, the author reports a shared tendency among all speakers to use unmarked linguistic strategies when faced with optional structures, to varying degrees. It also reveals a distinction between the performance of monolingual and bilingual people. In this sense, Thomas (2012) argues that while bilingual people may exhibit less variability compared to monolingual people in certain aspects, they might still show more variability overall due to the influence of multiple languages and the challenges associated with learning an L2.

Our findings are thus consistent with prior research highlighting HL speakers inclination towards less optionality compared to L1 speakers. This suggests a preference for specialised linguistic constructions or lexical items determined by context or personal inclination. This inclination reflects HL speakers’ preference for straightforward language structures, possibly leading to the avoidance of ambiguous or non-salient material and potential restructuring within their language, as discussed by Polinsky and Scontras (2020). Furthermore, these results could be explained, following insights from Camacho and Kirova (2018) and López Otero et al. (2023b), emphasising that HL speakers of Spanish tend to favour maintaining a consistent linguistic structure across both their languages when faced with situations requiring a choice between two options. This preference might emerge, as argued by Martín Gómez (2022), from a desire to minimise the cognitive effort required for language processing and to alleviate the mental strain associated with constantly toggling between different grammar rules.

6. Conclusions

This study reveals subtle differences in clitic pronoun usage between the HL and L1 Spanish speakers that made up our corpus. The results indicate comparable patterns in clitic pronoun use, suggesting native-like proficiency among our HL participants. Despite differences compared to the L1 group, the HL group demonstrates a strong command of clitic pronouns with more systematic preferences in optional contexts, showing more uniformity than the L1 participants. No significant phenomena related to clitic combinations or movement were observed, indicating that while clitic pronouns are complex, this group of HL speakers generally did not encounter major difficulties with them. Furthermore, despite differences between Spanish and French clitic systems, we did not find evidence of interlinguistic influence in our samples.

While our results provide valuable insights, it is crucial to interpret them within the context of the study’s limitations. Our research focused on a specific limited sample and research context, which may restrict the generalisability of the findings. Additionally, the semi-spontaneous nature of the task employed in our study could have influenced participants’ discourse, potentially leading to adjustments in their language production, such as permitting avoidance of structures that evoke insecurity among HL speakers. Thus, caution must be exercised when extrapolating these findings to broader populations or contexts.

In conclusion, our research adds to the discussion on HL acquisition and bilingual language use by examining the usage of clitic pronouns in Spanish. We have suggested insights into the proficiency levels of our HL speakers and the underlying factors shaping their linguistic competence. Moving forward, future investigations should expand beyond corpus-based research to include experimental studies. This complementary approach can provide deeper insights into potential avoidance strategies and the complexities of clitic pronoun usage in HL grammars. Embracing these alternative methodologies will help overcome the limitations of this case study, leading to more robust conclusions. This shift will allow for a comprehensive exploration of our research constraints and facilitate a deeper understanding of HL development dynamics. Additionally, exploring aspects such as ethical datives and the interpretation of non-paradigmatic se could offer valuable insights into the intricacies of HL clitic usage.