Highlights
-
• Heritage children show low infinitive use in subject position and with parar de.
-
• Spanish dominance and experience modulate target infinitive use
-
• Gerund overextension in subject position increases with greater English dominance.
-
• Age affects gerund overextension only in subject position among adolescents.
-
• Spanish morphosyntactic development is gradual and shaped by bilingual alignments.
1. Background
Bilingual language development is shaped by a complex interplay of linguistic and psycholinguistic factors, including developmental age, language dominance, structural complexity, crosslinguistic influence (CLI) and linguistic experience. In the case of Spanish as a minority language in the US, research has documented morphosyntactic variability across various domains, including mood, aspectual distinctions, copula selection, gender agreement, differential object marking and clitic placement (Alarcón, Reference Alarcón2011; Castilla-Earls et al., Reference Castilla-Earls, Pérez-Leroux, Martinez-Nieto, Restrepo and Barr2020; Cuza et al., Reference Cuza, Reyes and Lustres2021; Cuza & Pérez-Tattam, Reference Cuza and Pérez-Tattam2016; Dracos & Requena, Reference Dracos and Requena2022; Goebel-Mahrle & Shin, Reference Goebel-Mahrle and Shin2020; Montrul, Reference Montrul2002, Reference Montrul2009; Montrul & Sánchez-Walker, Reference Montrul and Sánchez-Walker2013; Thane, Reference Thane2024). The Bilingual Alignment Hypothesis (BAH) (Sánchez, Reference Sánchez2019) offers a useful framework to interpret these patterns by positing that morphosyntactic convergence or divergence between a bilingual’s two languages is shaped by the activation of linguistic features in each system. Rather than indicating incomplete grammars, bilingual systems are argued to be dynamic. That is, activation levels shift with exposure, dominance and usage, leading to systematic alignment patterns that shape production and comprehension. Recent findings support this view by showing strong associations between expected response and language dominance, lexical frequency and experience (Cuza et al., Reference Cuza, Shin and Sánchez2025; Hur et al., Reference Hur, Lopez Otero and Sánchez2020; Putnam & Sánchez, Reference Putnam and Sánchez2013; Shin et al., Reference Shin, Cuza and Sánchez2023; Thane, Reference Thane2024). Consistent exposure and use of Spanish, for example, are linked to higher proficiency and more accurate production among both child and adult heritage speakers. However, some studies report greater variability among older children with lower English proficiency, particularly in articles and clitic use in Spanish (Castilla-Earls et al., Reference Castilla-Earls, Pérez-Leroux, Martinez-Nieto, Restrepo and Barr2020).
Despite extensive research, the acquisition of non-finite verbal forms in child heritage Spanish has received less attention. We address this gap by examining the production of the Spanish infinitive in three syntactic contexts where English and Spanish diverge or overlap: infinitives in subject position (1a), infinitives as complements of the phrasal verb parar de (“to stop doing something”) (1b) and infinitives with the prepositional verb parar a (“to stop to do something”) (1c):
In (1a), the infinitive conveys a generic or habitual meaning. In (1b), parar de marks the cessation of an activity. In both cases, Spanish requires the infinitive, while English uses the gerund to convey the intended meaning. In (1c), parar a/para signals an interruption of one action to perform another, and in this case, infinitives are required in both languages. The difference in meaning between (1b) and (1c) is determined by the type of preposition used (de vs. a/para). Thus, English and Spanish diverge in (1a) and (1b) but align in (1c) in their use of the infinitive.
We examine whether school-age bilingual children and adolescents adhere to these grammatical forms in Spanish, compared to their parents, and if not, the extent to which CLI, Spanish dominance and experience modulate the use of the infinitive in these syntactic positions. Specifically, we examine whether the difficulties observed in adult heritage speakers (Belpoliti & Bermejo, Reference Belpoliti and Bermejo2019; Escobar & Potowski, Reference Escobar and Potowski2015; Solano-Escobar & Cuza, Reference Solano-Escobar and Cuza2023) also emerge in earlier stages of bilingual development. Alternatively, these difficulties may intensify with age and extended exposure to English in the lifespan of the speaker. If challenges do arise, which morphosyntactic structures are most vulnerable to CLI? Is the use of the infinitive more vulnerable in subject position than as complement of a preposition?
In what follows, we discuss the grammatical properties of infinitives and gerunds in Spanish (Section 2), followed by a review of previous acquisition research, and our research questions and hypotheses (Section 3). Section 4 describes the study, including participants, task and procedures. The results are presented in Section 5, followed by the discussion and conclusions in Section 6.
2. Infinitives and gerunds in Spanish and English
The Spanish infinitive form is a non-finite verb form consisting of a verb stem and a vowel suffix, as in cant-ar (to sing). In subject position, it denotes various semantic readings, including [+undetermined/indefinite] interpretations (Comer vegetales es saludable “Eating vegetables is healthy”), [+ongoing] interpretations (¿Qué hace Juan? = Estudiar “What is John doing? = “He is studying”), and [+habitual] interpretations (Nadar es mi deporte favorito “Swimming is my favorite sport”). The infinitive can be replaced by a noun (Leer es divertido or La lectura es divertida “Reading is fun”), while retaining an unmarked aspectual value (Alarcos Llorach, Reference Alarcos Llorach1994; De Miguel, Reference De Miguel1996; Hernanz, Reference Hernanz, Bosque and Demonte1999; Ramírez, Reference Ramírez2003; Yoon & Bonet-Farran, Reference Suk Yoon, Bonet-Farran, Wanner and Kibbee1991).
In addition to appearing in subject position, the infinitive also appears in object position (Quiero comer “I want to eat”) and is the only verbal form that can be the complement of a preposition (Aprobó el examen sin estudiar “She passed the exam without studying”; Estoy cansado de trabajar “I’m tired of working”). This diverges from English, where prepositions typically select a gerund form. Furthermore, the infinitive also appears as a predicative complement with perception verbs such as ver (to see) and oír (to hear). In these cases, both the infinitive and the gerund alternate in Spanish and English (La vi cantar/cantando “I saw her sing/singing”). The infinitive tends to highlight the event as a whole (cantar as referring to the act of singing), whereas the gerund might be used to emphasize the event as in progress. As pointed out by an anonymous reviewer, the infinitive also functions as the complement of modal verbs in Spanish and English (puede/debe/cantar “can/must/sing”).
We consider the use of the infinitive with parar a/para as a final infinitive clause, with a [+intentional] semantic value, as it expresses the intent of the subject to stop an activity to do another one (Juan paró a/para saludar a su amigo “John stopped to say hi to his friend”). This aligns with the traditional semantic meanings selected by the prepositions a and para with prepositional verbs in Spanish, as in Francisco vino a/para conocer a Santiago “Francisco came to meet Santiago.” In contrast, we consider parar de as a phrasal verb in Spanish. This is motivated by the fact that the preposition de in this specific context is not interchangeable with any other preposition in Spanish to convey the same meaning of stopping an action. Thus, it functions as a sort of verb-particle construction that bilingual children and adults would acquire as a lexical unit to express the concept of stopping an activity ([+culmination]). The gerund in English and the infinitive in Spanish with phrasal verbs like parar de or dejar de are also used to express giving up something, as in Alex paró/dejó de fumar “Alex stopped/quit/gave up smoking.”
Unlike infinitives, which may occur in subject position with an undetermined/indefinite reading, Spanish gerunds are formed with the suffix -ndo (cant-a-ndo) and cannot appear in that same syntactic environment (*Comiendo vegetales es saludable “Eating vegetables is healthy”) nor can they convey habitual meaning as the infinitive does. The gerund is primarily used to form the present continuous tense (Estoy trabajando “I’m working”), as an adjunct expressing manner (Mario salió corriendo “Mario left running”), and to express the manner of action (Caminando se quita el frío “Walking takes the cold away”). However, Hernanz (Reference Hernanz, Bosque and Demonte1999) cites cases of gerund use in Spanish with a durative meaning, as in Caminando por la Diagonal, encontrarás muchas tiendas “You will find many stores walking on the Diagonal” (p. 2201). The gerund, however, can be used as a complement of certain semi-modal verbs in both Spanish and English (Juana siguió cantando como si no hubiera pasado nada “Juana kept singing as if nothing had happened”).
Regarding English, the infinitives are formed periphrastically with the preposition to, without a dedicated vowel suffix (Álvarez, Reference Álvarez2022). In contrast to Spanish, English infinitives do not usually select undetermined interpretations (*To smoke kills) or ongoing eventualities (To study does not mean “studying right now”). Instead, English often uses the gerund for undetermined interpretations (Smoking kills), and for ongoing eventualities, the present progressive must be used (I’m studying right now). However, English can select the infinitive in subject position to express habitual interpretation (To swim is my favorite sport), but gerunds are preferred in everyday English.
Another configuration that allows the use of the infinitive in English is with final infinitive clauses, as in John stopped to play the piano, as explained earlier in the case of Spanish. Suppose that John is studying at home and then decides to take a break to play the piano. As mentioned earlier, English also allows for bare infinitives with perception verbs (I saw him leave), which parallels with Spanish (Lo vi salir). Regarding gerunds in English, they can function as the subject of the clause (Traveling is fun) or as the object of the verb (I like traveling). The choice between the two is often stylistically constrained (Arseneau, Reference Arseneau2009; Biber et al., Reference Biber, Johansson, Leech, Conrad, Finegan and Quirk1998). As the object of a preposition, English obligatorily selects the gerund form (She is tired of working).
For the purpose of our study, we focus on the use of the infinitive in subject position, and as a complement of parar de, and parar a. These syntactic configurations are interesting to examine given that English and Spanish overlap with the use of the infinitive with parar a/para (Ella paró a/para tomar agua “She stopped to drink water”) but not with phrasal verbs like parar de or as subject of the clause. Table 1 summarizes the differences between Spanish and English in relation to these syntactic configurations.
Infinitives and gerunds in Spanish and English: form/meaning connections
In what follows, we discuss previous research and postulate our research questions and hypotheses.
3. Previous research
The use of non-finite forms like infinitives and gerunds has been observed in Spanish monolingual children from early stages of language development. López-Ornat’s (Reference López-Ornat1994) corpus of a child named María shows instances of Verb+INF use (e.g., voy a senta(r) “I’m going to sit,” voy a escribi(r), “I’m going to write”) by 1 year and 11 months, as well as the gerund form by age 2 (e.g., (es)tá bu(s)cando a Antonio “He is looking for Antonio”; (es)toy t(r)abajando “I’m working”). María also produced instances of preposition A + INF (e.g., a (se) car (“to dry”), a pi (ntar) (“to paint”) by 1 year and 7 months. Hawayek (Reference Hawayek and Westgaard1997) argues that the early appearance of these structures may reflect lexicalized forms that disappear as children start developing inflection and assigning nominative case to subjects. However, A + INF is grammatical in Spanish if used with an exhortative meaning, as in A comer, muchachos (“Let’s eat, kids”). Infinitive forms have also been documented in Montes’ (Reference Montes and MacWhinney2002) corpus of Mexican Spanish in the speech of Koki, a child born in Mexico to Spanish and English-speaking parents. By 1 year and 9 months, Koki produced infinitives like arreglar (“to fix”) as well as present progressive forms such as e a bañando for se está bañando (“She/He is taking a shower”). By this age, Koki also produced target Verb+INF forms such as voy a buscarla (“I’m going to find her”).
Regarding bilingual development, Solano-Escobar and Cuza (Reference Solano-Escobar and Cuza2023) found patterns of CLI in the production and interpretation of infinitives among adult heritage speakers of Spanish. Results showed ungrammatical gerund use (20%) and interpretation (17%) in subject position (*Nadando es mi hobby favorito “Swimming is my favorite hobby”) but very little overextension (4%) of the gerund as the object of a preposition (Luisa está cansada de *trabajando “Luisa is tired of working”). Furthermore, findings revealed a significant correlation between participants’ proficiency, language experience, and their target use of and interpretation of the infinitive in subject position. More exposure to and proficiency in Spanish predicted more accurate use and interpretation of the infinitive form.
These findings align with earlier work by Potowski and Prieto-Mendoza (as cited by Escobar & Potowski, Reference Escobar and Potowski2015) documenting strong acceptance of gerund and infinitive forms in subject position. Escobar and Potowski (Reference Escobar and Potowski2015) explored the acceptability of gerund forms among 130 Spanish heritage speakers. Results showed that the participants rated gerunds and infinitives similarly in subject position (both in main clauses and subordinate clauses with a direct object), but infinitives were preferred as object of prepositions or following copula ser.
Gerund use in subject position and after prepositions was also found in the written production of low-proficiency Spanish heritage speakers by Belpoliti and Bermejo (Reference Belpoliti and Bermejo2019). In a corpus of 200 essays, infinitives predominated (85.6%), appearing mainly as complements of conjugated verbs (60.9%) or after prepositions (31.9%), with rare occurrences in subject position (5.2%). Gerunds occurred less frequently (14%), but when they did, they often appeared in contexts where infinitives are expected. These results suggest that, although infinitives are dominant in Spanish heritage grammars, gerunds surface in nominalized contexts, reflecting ongoing variability.
Research on other Romance heritage languages also shows variability with infinitive forms. For example, Rothman (Reference Rothman2007) examined heritage speakers of Brazilian Portuguese, and found variability in their target interpretation of inflected infinitives. The speakers struggled with agreement morphology (e.g., using falar “to speak” instead of falarmos “we speak”) and with identifying the subject of the verb in embedded contexts, compared to native speakers and L2 learners. Rothman argued that these difficulties stemmed from low frequency, variable input and limited literacy. In a follow-up study, Pires and Rothman (Reference Pires and Rothman2019) investigated European Portuguese heritage speakers, and unlike their Brazilian counterparts, these speakers showed high retention and target-like use of inflected infinitives. The authors concluded that differences found with European Portuguese speakers reflected issues related to frequency and quality of the input, rather than fundamentally different acquisition processes. Heritage speakers, they argue, tend to retain the morphological forms that are robustly represented in the input.
Taken together, these findings suggest that heritage language development is strongly modulated by the relative activation of the heritage and dominant language, as proposed by Sánchez’s (Reference Sánchez2019) BAH. Sánchez (Reference Sánchez2019) accounts for this variability by proposing that stable units (well-established forms) coexist with temporary, flexible alignments that may or may not stabilize over time. Greater activation of dominant language structures (e.g., English gerunds in our study) increases the likelihood that heritage-language features align with them, whereas more heritage-language activation limits such alignment. In this view, variability in heritage grammars reflects grammatical processing that can shift toward the dominant or heritage language depending on lexical activation (Giancaspro & Sánchez, Reference Giancaspro and Sánchez2021; Perez-Cortes et al., Reference Perez-Cortes, Putnam and Sánchez2019; Putnam & Sánchez, Reference Putnam and Sánchez2013). For example, Giancaspro and Sánchez (Reference Giancaspro and Sánchez2021) found that in heritage Spanish, proficiency modulated the overproduction of possessive determiners (Se rompió #su brazo “She broke her arm”) and the omission of the clitic se (*___ Rompió su brazo “She broke her arm”) in inalienable constructions. Yet in acceptability judgments, both intermediate and advanced heritage speakers rated expected (clitic se + definite determiner) and unexpected forms (no clitic+possessive determiner) as acceptable. These results suggest that production variability reflects temporary alignments influenced by English, rather than impaired underlying grammatical representations.
Another perspective emphasizes that variability in heritage grammars may reflect protracted development due to CLI, reduced input, and minority-language dominance (Cuza, Reference Cuza2024; Flores et al., Reference Flores, Santos, Jesus and Marques2017; Thane, Reference Thane2025; Jia & Paradis, Reference Jia and Paradis2015). This view emphasizes the developmental trajectory of heritage-language acquisition among children who are still learning the language. As noted by Flores et al. (Reference Flores, Santos, Jesus and Marques2017) and others (Gathercole, Reference Gathercole2007), bilingual children may take longer to fully acquire certain linguistic properties and reach native-like levels. For instance, Flores et al. (Reference Flores, Santos, Jesus and Marques2017) found that European Portuguese heritage children in Germany aged 6–16 showed delayed development of subjunctive morphology in complement clauses. However, convergence to monolingual norms emerged over time, particularly among children with higher Portuguese exposure at home. Similarly, U.S.-born Spanish heritage children overextended possessive determiners in inalienable constructions compared to their parents, Spanish-dominant peers, and monolingual adults, but variability decreased with more Spanish input and use (Cuza & Solano-Escobar, Reference Cuza and Solano-Escobar2023). These findings illustrate how reduced input and CLI shape developmental trajectories in child heritage grammars.
3.1. Crosslinguistic influence on non-finite verb use in Spanish/English bilingual children
In the context of child bilingual development, the interaction between the two non-finite forms (infinitive and gerund) available in both languages introduces an additional layer of complexity to the acquisition process. As represented in Table 1, English allows both gerund and infinitive forms in subject position, while Spanish only allows the infinitive. English also differs from Spanish in requiring the gerund as the object of the verb to quit/to stop, while Spanish categorically requires the infinitive form in this context as the object of the phrasal verb parar de. The English-dominant child has to learn that Spanish only allows the infinitive form in these two contexts. Therefore, they must suppress the gerund morphological form in English and expand the infinitive form to successfully select the [+habitual] semantic value of the infinitive form in subject position and the [+intentional] semantic value of the infinitive with the phrasal verb. This form-meaning alignment is required for semantic felicity, as the gerund in Spanish does not select a habitual interpretation in subject position or an intentional interpretation as the object of the phrasal verb parar de. Given these existing differences, English-dominant children may overextend the gerund form in Spanish in both contexts, influenced by nominal characteristics of the English gerund, which is ungrammatical in Spanish. In addition to CLI, the acquisition process could also be compromised by inconsistent input (Cuza & Pérez-Tattam, Reference Cuza and Pérez-Tattam2016; Hulk & Müller, Reference Hulk and Müller2000; Van dijk et al., Reference Van Dijk, Dijkstra and Unsworth2022). The input is inconsistent as the infinitive form can appear in subject position with various semantic specifications, in addition to habitual values, as mentioned earlier. The child might not get consistent and ample input to be able to acquire the specific [+habitual] and [+culmination] semantic values of the infinitive. Regarding the prepositional verb parar a, learners might be more accurate in their use of the infinitive form, as the INF is also required in English to select a [+intentional] semantic value.
As noted previously, other factors including developmental age, Spanish dominance and patterns of language use and exposure might modulate the degree of morphosyntactic variability in heritage Spanish (Cuza et al., Reference Cuza, Shin and Sánchez2025; Gómez Alzate et al., Reference Gómez Alzate, Cuza, Camacho and Zanelli2024; Montrul, Reference Montrul2002; Montrul, Reference Montrul2008; Montrul, Reference Montrul2015; Perez-Cortes, Reference Perez-Cortes2022; Putnam & Sánchez, Reference Putnam and Sánchez2013; Sánchez, Reference Sánchez2019; Shin et al., Reference Shin, Cuza and Sánchez2023). This aligns with recent work documenting gerund overextension in subject position in adult heritage speakers of Spanish (Solano-Escobar & Cuza, Reference Solano-Escobar and Cuza2023). However, it is unclear if this pattern of overextension also occurs in contexts where the type of preposition presupposes a change in meaning, as in the cases of parar de and parar a.
3.2. Research questions and hypotheses
Building on previous work and the existing differences between English and Spanish in this domain, we postulate the following research questions and hypotheses.
RQ1: To what extent do Spanish/English bilingual children and adolescents demonstrate target use of the infinitive in subject position, as an object of the phrasal verb parar de, and as an object of the preposition with parar a, compared to their parents?
Hypothesis 1: Bilingual children and adolescents will demonstrate low non-target use of the infinitive in subject position, and as object of parar de, compared to their parents. Instead, they will overextend the gerund form due to CLI from English. However, they will show target use of the infinitive with the prepositional verb parar a, as both English and Spanish require the infinitive in this context (Solano-Escobar & Cuza, Reference Solano-Escobar and Cuza2023).
RQ2: Is the target use of infinitive forms modulated by chronological age? In other words, does the level of accuracy change with increasing age?
Hypothesis 2: Bilingual children and adolescents will show higher patterns of non-target gerund use instead of the infinitive with increasing age due to more exposure to English outside the home environment and dominant-language shift (Goebel-Mahrle & Shin, Reference Goebel-Mahrle and Shin2020; Montrul, Reference Montrul2016; Jia & Aaronson, Reference Jia and Aaronson2003; Oller et al., 2011).
RQ3: To what extent do proficiency and language experience patterns modulate the use of infinitive forms?
Hypothesis 3: Bilingual children and adolescents with greater Spanish dominance and more exposure to and use of the language will show higher rates of infinitive use than those with less experience and weaker Spanish dominance (Montrul & Sánchez-Walker, Reference Montrul and Sánchez-Walker2013; Sánchez, Reference Sánchez2019; Silva-Corvalán, Reference Silva-Corvalán2014; Thane, Reference Thane2025).
We consider these questions and hypotheses to examine the acquisition path of these forms in Spanish heritage children and adolescents in contact with English in the American Midwest. We also compare their performance with data from their parents, the primary input providers. While increasing age could, in principle, lead to specification in Spanish with continued home exposure (Thane, Reference Thane2024), older children may still show significant variability. This can depend on the structure under analysis, its frequency in the input or the elicitation methods used (Goebel-Mahrle & Shin, Reference Goebel-Mahrle and Shin2020; Silva-Corvalán, Reference Silva-Corvalán2014). For structures that are less frequent in the input, as infinitive in subject position or parar de, we would expect developmental trajectories to reflect more alignment with English as children grow older, expand their social networks and spend more time in English-dominant contexts (Montrul, Reference Montrul2008; Gathercole & Thomas, Reference Gathercole and Thomas2009; Jia & Aaronson, Reference Jia and Aaronson2003; Oller et al., 2011; Polinsky, Reference Polinsky2011). Nevertheless, stronger Spanish proficiency could potentially mitigate the shift toward English (Cuza & Solano-Escobar, Reference Cuza and Solano-Escobar2023; Dracos & Requena, Reference Dracos and Requena2022; Thane, Reference Thane2025).
4. The study
4.1 Participants
A total of 70 participants took part in the study: 28 bilingual children, 20 bilingual adolescents and 22 Spanish-dominant parents. The children and the adolescents completed the Multilingual Naming Task (MiNT) as an independent measure of proficiency in Spanish and English (Gollan et al., Reference Gollan, Weissberger, Runnqvist, Montoya and Cera2012; Sheng et al., Reference Sheng, Lu and Gollan2014). Following Sheng et al. (Reference Sheng, Lu and Gollan2014), a dominance score was calculated by dividing the number of correct responses by the total number of tokens and multiplying it by 100. Participants scoring 5% or more in one language were classified as dominant in that language, while those with a difference of less than 5% between scores in each language were classified as balanced bilinguals. Parents completed the child language background questionnaire (Pérez-Leroux et al., Reference Pérez-Leroux, Cuza and Thomas2011; Shin et al., Reference Shin, Cuza and Sánchez2023), which gathered details on age, birthplace, language exposure patterns and fluency ratings in English and Spanish. We assessed the children’s average use of both languages and their exposure to them through interactions with family members (mother, father, grandparents, siblings) and friends, using a 5-point Likert scale ranging from 1 (never) to 5 (very frequently). For the quantitative analysis, responses ranging from 1 to 3 were entered as 0, and 4 and 5 responses were entered as 1. Testing was conducted face-to-face at the children’s school and a local church.
The children (n = 28; age range = 6;5–11;11; M = 9.4; SD = 1.6) were born and raised in the US except two who arrived in the US at the age of 3;0 or earlier. The results of the MiNT test showed an average dominance score in Spanish of 47% and in English of 74%. Most of the children (75%, 21/28) were English-dominant, with three participants being balanced and four being Spanish-dominant. Regarding language exposure and use, their reported average Spanish use and exposure was 3.8/5. Their reported average English use was 3.4/5 and exposure was 3.3/5. Their Spanish use and exposure were reported to occur primarily with parents and grandparents, and more English with siblings and friends.Footnote 1
The adolescents (n = 20, age range = 12;0–17;11; M = 15;12; SD = 2.02) were also born and raised in the US, except three who arrived in the US at the age of 2, 4 and 9. Their average dominance in Spanish was 57% and 79% in English. The majority were English-dominant (16/20, 80%). Their Spanish usage was reported as 3.6/5 and their Spanish exposure as 3.8/5. Regarding English, their reported usage was 3.4/5, and their reported exposure was 3.2/5.
The Spanish-dominant parents (n = 22, age range = 29–53; M = 39; SD = 6.57) were first-generation immigrants to the US born in Latin America. Most of them (77%) indicated speaking only or mostly Spanish at home. At work, they reported speaking only or mostly English (38%), Spanish (33%) or both (29%). In social situations, 71% indicated speaking Spanish, 14% English, and 14% both. In general, the parents reported speaking Spanish with family members and neighbors frequently. Using a 5-point Likert scale assessing proficiency across the four language skills, the parents self-rated their Spanish proficiency at 3.7/4 and their English proficiency at 1.9/4, indicating higher proficiency in Spanish than in English (Table 2). The Likert scale for self-proficiency rating ranged from 0 (Null) to 4 (Excellent) (null, limited, adequate, good, excellent).
Participants’ information, proficiency scores and language experience
4.2. Tasks and procedures
We implemented an oral elicited production task using a question-answer and sentence-completion format (Cuza & Pérez-Tattam, Reference Cuza and Pérez-Tattam2016; Shin et al., Reference Shin, Cuza and Sánchez2023). The task was presented orally and visually with the aid of PowerPoint. Participants were presented with a story, followed by a question and a photo. They had to complete the sentence provided orally using the infinitive form of the verb, as represented below. The verb was provided in parentheses for parar a and parar de to avoid the use of “other” structures. The task consisted of 37 items: 10 items testing the use of infinitive forms in subject position, 6 items testing infinitive use with parar a, and 6 items testing the use of the infinitive with parar de. In addition, we included 12 distracters having to do with the use of the future perfect in Spanish as part of a larger study, as well as 3 items requiring the infinitive as object of a preposition de or por (cansada de trabajar “tired of working”) following previous work. However, this was not part of our research questions. All items were counterbalanced and randomized.
5. Results
5.1. Procedure for data analysis
We conducted a generalized linear model (GLM) analysis (using R Development Core Team, 2020) specifying a multinomial probit distribution to evaluate whether group membership significantly influenced the likelihood of producing the expected response rather than the gerund or “other” responses. The expected response was the use of the infinitive form, while unexpected responses included the use of the gerund and “other” responses (responses that did not match the target construction in the three syntactic contexts) (three response types). The response type (infinitive, gerund, other) was entered as the dependent variable, and group as the independent factor. For this analysis, the infinitive was coded as 1, the gerund as 2 and “other” responses as 3.
A second model was run to assess whether age, proficiency and patterns of language use modulated response type among the child and adolescent participants across syntactic context. In this model, response type remained the dependent variable, while age, proficiency and language experience were included as independent predictors. In what follows, we discuss our results by syntactic context given that the variability observed is context specific: higher gerund use in subject position, moderate with parar de and virtually absent with parar a.
5.2. Results for subject position
Results for subject position revealed that both children and adolescents produced fewer infinitives (65%) than parents (96%). Children used the gerund and “other” structures (e.g., nominalizations like la natación “swimming,” A + INF forms like a nadar “to swim,” and unrelated responses) in 18% of cases. Adolescents, in contrast, used the gerund more often than children (26% versus 18%) and produced fewer ‘other’ structures (10%). Among children’s “other” responses, 40% were A + INF, 46% nominalizations, and 15% were unrelated; adolescents showed 53%, 42% and 5%, respectively (Figure 1). The lower rate of “other” responses among adolescents, combined with their increased use of the gerund, may suggest stronger alignment with English through the overextension of the gerund as they grow older.
Proportion of use by response type per group for subject position.
A GLM analysis indicated significant group differences in the use of the gerund (β = −1.3032; z = −7.9224; p < .000; OR = 0.27) and “other” responses (β = −1.2821; z = −7.8591; p < .000; OR = 0.28) compared to the infinitive. The infinitive response was the most common in the dataset. Adolescents used the gerund more than children but produced fewer ‘other’ responses (β = −0.6333; z = −2.1470; p = .03; OR = 0.53). The odds ratio indicates that adolescents were 0.53 times less likely to produce “other” responses instead of the infinitive compared to children. Parents were highly unlikely to produce either gerunds (β = −11.8343; z = −0.2407; p = .81; OR = 0.00) or “other” responses (β = −1.8678; z = −4.94; p < .000; OR = 0.15).
Bonferroni-adjusted t-tests looking at the differences between groups showed significant differences in their use of the infinitive between children and parents (β = −0.3134, t = −9.7470, p = .002; OR = 0.73) and adolescents and parents (β = −3107, t = −8.5320, p = .005, OR = 0.73) but no differences between children and adolescents in their use of infinitives (β = −0.0027, t = −0.0610, p = 1.000; OR = 0.99) or gerunds (β = −0.0809, t = −2.0910, p = 1.000; OR = 0.92). Parents behaved significantly differently from children (β = 0.1754, t = 7.549, p = .01; OR = 1.19) and adolescents (β = 0.2563, t = 8.28, p = .006; OR = 1.29) in showing zero use of gerund forms. The odds of producing the gerund were 19% higher for children and 29% higher for adolescents compared to parents. These results support H1, which predicted that both children and adolescents would overextend the gerund relative to parents in subject position.
An individual analysis within children and adolescent groups was conducted to identify the number of participants per group who produced infinitive forms in subject position. For this analysis, participants were classified depending on the number of infinitive forms they produced as high-achievers (7–10 instances of 10), mid-achievers (4–6 instances of 10), low-achievers (1–3 instances of 10) and zero-production (0 instances of 10) (Table 3).
Individual analysis of infinitive production in subject position by children and adolescents
The individual results confirmed the group results in showing similar production of the infinitive among children and adolescents (50% and 55%). However, adolescents showed more variability than children in that 15% of them did not produce any instances of infinitive verb use, and 30% were in the medium or low range.
5.3. Results for parar de
Regarding parar de, children produced 46% of infinitives, 6% of gerunds, and 49% of “other” responses. In contrast, adolescents showed higher infinitive use (73%) and gerund use (13%), but fewer “other” responses (14%). Parents were almost at ceiling in infinitive production (99%) (Figure 2).
Proportion of use by response type per group for phrasal verb parar de.
Among children’s “other” responses, 81% involved preposition omission (Paró *Ø tocar el piano “He stopped playing the piano”) and 16% involved substituting de with a (Paró #a bailar salsa “He stopped to dance salsa”), which altered the intended meaning; 3% were unrelated. Adolescents also showed more omission (76%) than substitution (24%).
The gerund was significantly less likely than the infinitive across all groups (β = −2.1327; z = −6.0520; p < .000; OR = 0.12). Results from Bonferroni t-tests showed no significant differences between children and adolescents in infinitive (p = .1004) or gerund use (p = 1.0000). However, both groups differed significantly from parents in infinitive use (children: β = −0.5319; t = −13.45; p < .000; OR = 0.58; adolescents: β = −0.2590; p = .0267; OR = 0.77), supporting H1. To examine the group results more closely, we conducted an individual analysis. In this analysis, participants were classified as high-achievers (4–6 instances), low-achievers (1–3 instances) and zero-production participants (0 instances) (Table 4).
Individual analysis of infinitive production in the parar de condition by children and adolescents
As shown in Table 4, only 43% of children produced a high number of infinitives, compared to 75% of adolescents. Although adolescents used more gerunds than children, they also produced more infinitives and fewer “other” responses.
5.4. Results for parar a
Regarding parar a, children produced 56% infinitives, 1% gerunds and 43% “other.” Adolescents showed higher infinitive use (84%), 1% gerunds and 15% “other” (Figure 3).
Proportion of use by response type per group for prepositional verb parar a.
The “other” responses among children involved preposition omission (73%), substitution (11%) and unrelated responses (15%). For adolescents, omission accounted for 28%, substitution for 11% and unrelated responses for 61%. Although “other” responses were more frequent than gerunds in both groups, the difference was not significant (β = −0.2589; z = −1.6390; p = .1012; OR = 0.77). Adolescents were significantly less likely than children to produce “other” responses compared to infinitives (β = −1.4658; z = −4.8749; p < .001; OR = 0.23). Bonferroni tests showed that adolescents outperformed children in infinitive use (β = −0.2842; t = −5.5670; p = .05; OR = 0.75), while no significant differences were found for gerunds (p = 1.000). In addition, parents significantly outperformed children in their use of the infinitive (β = −0.3971; t = −9.2980; p = .003; OR = 0.67), while no significant differences were found between adolescents and parents (p = .8914). Thus, H1 was only partially confirmed. An individual analysis confirmed the group results (Table 5).
Individual analysis of infinitive production in the parar a condition by children and adolescents
As shown in Table 5, adolescents showed higher use of infinitive forms compared to children, suggesting specification of the infinitive form as a complement of parar a in adolescence. As suggested by an anonymous reviewer, it is possible that children and adolescents were more target-like in this context because prepositions in Spanish are never followed by gerunds, so the preposition itself may be inducing the use of the infinitive.
5.5. The role of age, dominance and experience across contexts
In subject position, age did not significantly predict the choice between gerund and infinitive use among the children (β = 0.0013, z = 0.15; p = 88, OR = 1.00) but it did among adolescents (β = 0.0298, z = 3.8048, p < .000, OR = 1.03). Each additional month increased the odds of gerund use by 3%. Thus, H2 was partially supported. Spanish dominance was negatively associated with gerund use among children (β = −0.0652, z = −5.1889, p < .000; OR = 0.94) and adolescents (β = −0.0750, z = −5.3458, p < .000; OR = 0.93). Children and adolescents with stronger Spanish dominance were significantly less likely to use the gerund. Conversely, higher English dominance significantly increased the probability of using the gerund among children (β = 0.0734, z = 3.2162, p < .000; OR = 1.07) and adolescents (β = 0.0969, z = 3.8233, p < .001; OR = 1.10).
Spanish use and exposure were also significant predictors among children (use: β = −1.6732, z = −3.00, p = .003, OR = 0.19; exposure: β = −1.7583, z = −2.44, p = .015, OR = 0.17). Increased Spanish use and exposure significantly reduced the probability of using the gerund over the infinitive. Conversely, increased use of English significantly raised the odds of using the gerund instead of the infinitive (β = 2.0182, z = −3.06; p = .002; OR = 7.52). Among adolescents, Spanish use (β = −4.3791, z = −5.1571, p < .000; OR = 0.0125) and exposure (β = −6.9361, z = −5.2825, p < .000; OR = 0.0010) were also significant predictors, with more Spanish experience sharply reducing the likelihood of gerund production. By contrast, increased English exposure substantially raised the odds of selecting the gerund over the infinitive (β = 8.8678, z = 5.8056 p < .000; OR = 7099.83). These results support H3, which predicted that children and adolescents would show more target-use of the infinitive with increased Spanish dominance and experience.
Regarding parar de, age was not a significant predictor for either group (children: p = .84; adolescents: p = .41), contrary to H2. However, Spanish dominance again correlated with reduced gerund use (children: β = −0.0477, z = −2.06, p = .04, OR = 0.95; adolescents: β = −0.0969, z = −4.15, p < .000, OR = 0.90). Children with stronger Spanish dominance significantly reduced the likelihood of using the gerund instead of the infinitive. A similar pattern was observed among adolescents. In both groups, higher Spanish dominance was associated with lower odds of gerund overextension. Regarding Spanish use, we found a strong correlation among children (β = −3.1960, z = −2.1948, p = .03; OR = .04) and adolescents (β = −7.7864, z = −4.800, p < .000; OR = .0004). Furthermore, significant associations were observed between Spanish exposure and gerund use among children (β = −1.8928, z = −0.8944, p = 0.4; OR = .1506) and adolescents (β = −8.5356, z = −3.3850, p < .000; OR = .0002). Increased Spanish experience, consistently reduced the probability of gerund use, providing further support for H3 in the context of parar de.
Regarding parar a, age revealed no significant association with the likelihood of using the gerund form rather than the infinitive among children (p = .14) or adolescents (p = .67), in contrast to what was predicted in H2. Additionally, there was no significant interaction between gerund use and Spanish dominance (children: p = .94; adolescents: p = .24), use (children: p = .17; adolescents: p = .80) or exposure (children: p = .27; adolescents: p = .32). Thus, H2 and H3 were not supported.
To summarize, age, dominance and language experience significantly predicted infinitive use in subject position and dominance and experience also predicted use with parar de. No significant predictors emerged for parar a. These results suggest that CLI and proficiency interact in distinct ways across syntactic environments as far as target infinitive use is concerned.
6. Discussion
This study examined the use of Spanish infinitive forms among school-age bilingual children and adolescents across three different contexts. Our findings reveal clear, context-dependent variability shaped by CLI, proficiency, and language experience. As predicted in H1 (low infinitive use in subject position and with parar de but not with parar a), both child and adolescent groups underproduced infinitives in subject position (65% each) and with parar de (46% and 73%, respectively), compared to near-ceiling use among parents. Adolescents diverged from children in relying more on gerunds (subject position: 26% versus 18%; parar de: 13% versus 6%), suggesting stronger English alignment as age increased. Gerund overextension was most pronounced in subject position, suggesting higher alignment with English. In contrast, difficulties with parar de diminished with increased proficiency and age. These differences in target-like responses across these two contexts may stem from differences in input salience between English and Spanish. Structure frequency and salience are known to shape bilingual children’s acquisition patterns, with more frequent and salient forms in English being more likely to be overgeneralized (Gathercole, Reference Gathercole2007). In English, the gerund in subject position is highly salient and frequent, compared to stop+INF constructions, which may lead to its overextension into Spanish. The prominence of gerund forms in subject position in English may affect the full specification of the [+habitual] semantic value associated with the infinitive in subject position, more so than the specification of the [+culmination] semantic value associated with the infinitive as a complement of parar de, which appears to be resolved with age. This pattern aligns with broader findings in bilingualism research showing that CLI is often mediated by the relative frequency and salience of structures across a child’s two languages (Baker Martínez & Shin, Reference Baker Martínez and Shin2023; Montrul, Reference Montrul2008; Paradis, Reference Paradis2010; Silva-Corvalán, Reference Silva-Corvalán1994).
The results with subject position support the BAH by showing a shift toward English-like alignments with increasing age, higher English proficiency and more exposure among adolescents. The adolescents overextended the gerund with increasing age, English dominance and exposure. Increased English exposure raised the odds of selecting the gerund. We would like to argue that in this specific context, both children and adolescents have aligned with English morphosyntactic features in using the gerund at the level of processing. Future research should examine interpretation patterns to determine whether these alignments become stable and reflect grammatical restructuring. The use of “other” structures in subject position was related primarily to A + INF constructions and nominalizations. Although A + INF is attested in early monolingual development, it typically disappears with increased inflectional development and is absent in adult heritage speakers (Solano-Escobar & Cuza, Reference Solano-Escobar and Cuza2023). Its presence in our data was limited to children and adolescents with low proficiency in the language, suggesting protracted development that resolves with added proficiency and experience.
For parar de, participants showed more target-like use with increasing Spanish dominance, use and exposure in both children and adolescents. Participants with higher Spanish use, exposure and dominance were less likely to use the gerund. Age, however, was not a significant predictor for either group. These results also support the BAH. It seems as if the culprit constraining target response with parar de is dominance and language input, not necessarily age. Given the lack of associations with age, it seems more plausible to account for the results on the basis of bilingual alignments, as it is clear that they are transient in nature rather than stable as far as the infinitive with parar de is concerned. Variability seems to resolve (although not completely) with added experience and proficiency (Perez-Cortes et al., Reference Perez-Cortes, Putnam and Sánchez2019; Sánchez, Reference Sánchez2019). Participants with stronger Spanish dominance and more exposure were less likely to overuse the gerund, consistent with previous research on heritage language acquisition linking language exposure and proficiency to structural outcomes (Paradis, Reference Paradis2010; Putnam & Sánchez, Reference Putnam and Sánchez2013; Sánchez, Reference Sánchez2019). Children’s ‘other’ responses with parar de often involved preposition omission (81%) (i.e., Paró (*de) tocar el piano), likely reflecting limited awareness of the phrasal nature of parar de (recognizing the need for the preposition de as part of the verbal phrase), low Spanish proficiency and English influence. With parar a, children demonstrated low infinitive use due to “other” responses rather than gerund overextension. Adolescents, however, exhibited a high level of target infinitive use in this context. These results support H1, which predicted no overextension of the gerund with parar a.
Regarding H2, predicting a role of age in the choice of gerund versus infinitive, it was only partially confirmed among the adolescents in subject condition. The odds of using the gerund among adolescents in subject position increased by 3% with each additional month, suggesting a gradual shift toward more target gerund use, reflecting increased alignment with English. There was no significant age effect among children in subject condition or among either group with parar de or parar a. Infinitive use with parar de and parar a actually increased among adolescents (parar de: 73%; parar a: 84%) compared to children (parar de: 46%; parar a: 56%). These findings suggest that Spanish heritage children begin to overcome challenges associated with using the gerund as a complement of a preposition as they grow older, reflecting a developmental trajectory toward more target-like usage patterns. This trend is consistent with previous findings documenting minimal variability in the use of the infinitive as a prepositional complement (Solano-Escobar & Cuza, Reference Solano-Escobar and Cuza2023). These findings also suggest that the acquisition of the Spanish infinitive, much like other morphosyntactic structures, follows a non-linear and gradual trajectory, with parar de and parar a being mastered progressively with age, while the infinitive in subject position remains underspecified (Cuza & Solano-Escobar, Reference Cuza and Solano-Escobar2023; Flores et al., Reference Flores, Santos, Jesus and Marques2017; Liceras, Reference Liceras, Rojo, Vásquez Rojas and Torres-Cacoullos2023; Zobl, Reference Zobl1990; Zobl & Liceras, Reference Zobl, Liceras, Bamman, Magnitskaia and Zaller2006). Hypothesis 3, which predicted higher level of infinitive use with stronger Spanish skills and experience, was fully supported among children and adults in subject position and with parar de. Higher Spanish dominance and experience correlated with lower gerund use, supporting previous work linking dominance and experience to morphosyntactic shifts in child heritage Spanish (Cuza et al., Reference Cuza, Shin and Sánchez2025; Pirvulescu et al., Reference Pirvulescu, Pérez-Leroux, Roberge, Strik and Thomas2014; Shin et al., Reference Shin, Cuza and Sánchez2023).
To summarize, Spanish infinitive use among bilingual children and adolescents varied by context, proficiency, and language exposure. Both groups underproduced infinitives in subject position and with parar de, with adolescents showing higher levels of gerund overextension, reflecting stronger English influence. Overextension was most pronounced in subject position, while parar de use improved with higher Spanish dominance and input, suggesting that variability is constrained by language exposure rather than age. Parar a showed high target-like infinitive use, consistent across age groups. Overall, stronger Spanish dominance and experience predicted more target-like infinitive use, supporting the role of input and proficiency in bilingual morphosyntactic development. Teaching approaches to heritage speakers should emphasize the use of the infinitive in subject position and provide explicit contrasts between gerund and infinitive use.
Taken together, these results support the view that the target use of infinitive forms in child heritage Spanish is highly context dependent, varying across subject position, parar de, and parar a, with adolescents showing stronger English-like alignment in subject position. Importantly, variability in parar de was driven more by Spanish dominance and exposure than by age, underscoring the important role of lexical proficiency and language experience in shaping morphosyntactic development (Putnam & Sánchez, Reference Putnam and Sánchez2013; Sánchez, Reference Sánchez2019). The study also provides evidence of non-linear acquisition in that some structures are gradually mastered with experience, while others remain underspecified. Overall, these findings highlight the influence of input salience, dominance and context in bilingual acquisition, offering a more nuanced view of how heritage and dominant languages interact over time.
7. Conclusions
We investigated the use of infinitive forms among Spanish/English bilingual children and adolescents raised in the US. Our findings showed overextension of the gerund instead of the infinitive in subject position, lower with parar de and minimal with parar a. Bilingual children and adolescents with lower Spanish proficiency and less exposure to Spanish were more likely to overextend the gerund than their higher-proficiency peers, supporting the view that language proficiency and activation plays a critical role in shaping morphosyntactic patterns (Sánchez, Reference Sánchez2019).
The findings deepen our understanding of the acquisition of infinitive forms in child heritage Spanish. The overextension of the gerund with parar de highlights areas of protracted development, where younger bilingual children initially diverge from the parental norm but seem to realign or get closer to the parental norm with increasing age and exposure to the language. Additionally, the fact that parar a was largely unaffected by CLI suggests that not all infinitival constructions are equally vulnerable in a language contact situation, aligning with findings from adult heritage speakers (Solano-Escobar & Cuza, Reference Solano-Escobar and Cuza2023).
Future research should explore comprehension alongside production to determine whether English-like alignments reflect performance pressures or deeper grammatical restructuring (Putnam & Sánchez, Reference Putnam and Sánchez2013). Examining interpretation alongside production would offer a more complete picture of participants’ internal grammatical representations and the extent to which restructuring has occurred. It would also be valuable to compare heritage speakers with adult L2 learners to determine whether the observed patterns are specific to early bilingual development or extend to other bilingual populations. In doing so, we might uncover instructional effects among L2 learners, particularly regarding the use of infinitives in subject position or the interaction of prepositions with infinitives and gerunds in contexts such as parar de and parar a. However, this remains an open question, as the fine-grained syntactic and semantic distinctions explored in the present study are rarely taught explicitly or implicitly in typical second language instruction.
Overall, the study underscores the central role of proficiency and language experience in bilingual grammatical development. While early bilinguals show variability in subject infinitives, continued exposure and increasing proficiency promote more consistent use of infinitives with parar de and parar a, aligning their patterns more closely with those of the Spanish-dominant parents.
Acknowledgments
We are deeply grateful to all the families and children who participated in this study. We also thank Luis Pardo Granados and Julian Rodríguez Herrera for their valuable comments and feedback and Leonardo Puertas Bustos for his assistance with the statistical analyses.
Open access
