The Distribution of Manner and Frequency Adverbs in Child Heritage Speakers of Spanish

Abstract

We investigate the acquisition of adverb placement in Spanish among school-age child heritage speakers of Spanish born and raised in the US by Mexican parents. We examine frequency and manner adverbs with negative and positive polarity and the potential role of cross-linguistic influence, dominance, and experience in the path and rate of development. Fourteen child heritage speakers of Spanish born and raised in the US and twenty-five Spanish monolingual children from Mexico completed an elicited production task. Results showed that the heritage children produced significantly fewer verb-raising structures compared to the monolingual children, leading to a higher proportion of pre-verbal adverb use and adverb-final use. The heritage children treated manner and frequency adverbs with negative and positive polarity significantly differently. We also found a strong correlation between dominance and experience in the probability of producing specific adverbial positions. In other words, common adverbial positions in English were more likely to be produced with higher dominance and experience in English; likewise, Spanish adverbial positions were more likely to be used with higher dominance and experience in Spanish. We argue for differential outcomes in child heritage grammar due to differences in the path and rate of language development as well as the role of dominance and experience in child heritage language acquisition.

1. Introduction

The present study examines the distribution of manner and frequency adverbs among child heritage speakers of Spanish. Spanish heritage speakers are usually exposed to the minority language at home and become more dominant in English after school immersion (Montrul 2008, 2016, 2022; Polinsky 2018; Sánchez 2019; Shin et al. 2023; Valdés 2000). Specifically, we examine the extent to which child heritage speakers of Spanish born and raised in the United States (US) are sensitive to adverb placement in contexts where English and Spanish diverge in their surface word order. In contrast with Spanish (1a), English disallows adverb placement in post-verbal position (2a). However, both languages allow adverb placement in pre-verbal position (1b and 2b) and sentence-final position (1c and 2c):

(1)	a.	Esteban come frecuentemente pasta.
		‘Esteban eats pasta frequently’
	b.	Esteban frecuentemente come pasta.
		‘Esteban eats pasta frequently’
	c.	Esteban come pasta frecuentemente.
		‘Esteban eats pasta frequently’
(2)	a.	*Esteban eats frequently pasta.
	b.	Esteban frequently eats pasta.
	c.	Esteban eats pasta frequently.

The flexibility in Spanish regarding adverbial placement is argued to stem from its rich morphological system (Ayoun 2005; Guijarro-Fuentes and Larrañaga 2011). Spanish is considered a verb-raising language that allows verb movement within the sentence (Camacho and Sánchez 2017; Suñer 1994; Ordóñez 1997; Zagona 2002). English, in contrast, does not possess this feature, and therefore, adverbial placement is more restricted. Given the syntactic configurations of each language regarding adverb placement and the semantic nature of manner and frequency adverbs in each language (Bosque and Gutiérrez-Rexach 2009), we expect heritage speakers of Spanish in the US to have difficulties in this domain due to cross-language interaction, proficiency in each language, and the speakers’ level of language exposure and use.

The study is structured as follows. Section 2 provides an overview of the syntax of adverb placement in both English and Spanish. This is followed by a summary of previous research, research questions, and hypotheses in Section 3. Section 4 presents the study, followed by the results in Section 5. The discussion and conclusions are presented in Section 6.

2. Adverb Placement in Spanish and English

Previous work on adverb distribution in Spanish has followed various approaches to investigate why adverbs go where they go inside a sentence. Bosque and Gutiérrez-Rexach (2009) claim that adverbs are semantically heterogeneous, an idea that resonates with authors such as Zagona (2002), who categorizes adverbs in Spanish according to their semantic features (Cinque 1999). Zagona (2002) argues that VP-adverbs are divided into five different categories, each allowing different positions that the adverb can take, as in

Table 1. Available positions in Spanish based on Zagona (2002).

	Time	Place	Degree/Extent	Manner	Quantity
S-V-Adv-O	Yes	Yes	X	Yes	Yes
S-Adv-V-O	Yes	Yes	Yes	?*	X
S-V-O-Adv	Yes	Yes	X	Yes	Yes

.

Note that Spanish seems fairly flexible in terms of adverb distribution. However, certain types of adverbs, such as degree/extent, are not allowed in the verb-adverb (3a) nor adverb-final positions (3b). This also applies to quantity and manner with the adverb-verb position (3c, 3d), although for the latter, more than ungrammatical, it appears to be rather odd.

(3)	a.	*Juliana completó casi la tarea1. (degree)
		‘Juliana almost completed the homework’
	b.	*Juliana completó la tarea casi. (degree)
		‘Juliana almost completed the homework’
	c.	*Juliana mucho quiere su muñeca. (quantity)
		‘Juliana loves her doll much’
	d.	?*Juliana cuidadosamente lava sus manos. (manner)
		‘Juliana washes her hands carefully’

Zagona (2002) does not provide a description for a ‘frequency’ adverb category. However, it is suggested that they belong to the ‘time’ category. Accounting for their semantic variation but also for their syntactic behavior, Jackendoff (1972) suggested a similar classification of English adverbs into subject-oriented, sentential, and manner adverbs, stating that each could go in the positions shown in

Table 2. Available positions in English based on Jackendoff (1972).

	Subject-Oriented	Sentential	Manner
Adv–S–V–O	Yes	Yes	Yes
S–Adv–V–O	Yes	Yes	Yes
S–V–Adv–O	X	X	X
S–V–O–Adv	Yes	Yes	Yes

.

For their adverbial positions, English does not allow adverbs between the verb and the object for either subject-oriented (4a), sentential (4b), or manner (4c).

(4)	a.	*Life has clearly difficulties.
		‘Clearly, life has difficulties’
	b.	*John opened probably the door.
		‘Probably, John opened the door’
	c.	*Ana completed carefully her tasks.
		‘Carefully, Ana completed her tasks’

Comparing the distribution for both Spanish and English, we see that from the possible positions, the post-verbal placement is not allowed in English, but it is in Spanish. The pre-verbal and sentence-final positions are allowed in both languages, as discussed before. Syntactically speaking, this difference appears to stem from a phenomenon usually referred to as ‘verb raising’. Guijarro-Fuentes and Larrañaga (2011) found evidence that the post-verbal position that adverbs can take happens in languages where the adverb adjoins the VP (Spanish and French, for example). This may suggest that adverbs themselves are not moving and positioning themselves, but rather that other syntactic components, such as verbs, do (Koronkiewicz 2022; Pollock 1989). This process can also be called V-to-T movement (Camacho and Sánchez 2017; Guijarro-Fuentes and Larrañaga 2011; Koronkiewicz 2022; Pollock 1989), and it is defined by whether there is verb movement inside the sentence. See (5a) below.

(5)	a.	[TP Carlos [T [come +t [VP [AdvP tranquilamente] V [DP su almuerzo]]]]]
		Carlos    eats            quietly              his lunch
		‘Carlos quietly eats his lunch’

The opposite situation happens in English. As shown in (5b), the adverb “quietly” originates adjoined to the VP like in Spanish “tranquilamente” in (5a). However, in contrast to Spanish, in English, both adverb and verb stay in situ; it is T that moves and lowers to reach the verb located in VP. This process is known as ‘affix lowering’ or ‘T-lowering’ (Camacho and Sánchez 2017; Chomsky 1957; Guijarro-Fuentes and Larrañaga 2011; Koronkiewicz 2022; Larson 2010; Radford 2009).

(6)

b.

[TP Carlos [T [t -s [VP [AdvP quietly] V eat +t [DP his lunch]]]]]

This difference is not new to the English language. Studies about its evolution suggest that English displayed verb movement before the 16th century, the date by which it started to decline (Haeberli and Ihsane 2016; Roberts 1993). The authors claim it was a progressive process that took over two centuries to be completed in order to surface the syntactic features known in modern English: [+mvt] for auxiliaries (to be, to have) and [-mvt] for lexical verbs (Ayoun 2005).

English and Spanish also diverge in relation to the adverb placement in negative sentences. Zagona (2002) argues that for clausal negation, the negative adverb ‘no’ must be adjacent to the first verb in the clause (7a) and that only clitics are allowed to go between the ‘no’ and the verb; no other lexical categories are allowed (7b, 7c). This word order in English requires do-support (7e); otherwise, it will result in being ungrammatical (7d). Camacho and Kirova (2018) argue that this is the result of negation blocking the affix-lowering process.

(7)	a.	María no lee el periódico.
		‘María doesn’t read the newspaper’
	b.	María no lo lee.
		‘María doesn’t read it’
	c.	*María no frecuentemente lee el periódico.
		‘María doesn’t read the newspaper frequently’
	d.	*María not read the newspaper frequently.
	e.	María doesn’t read the newspaper frequently.

Bearing these differences between English and Spanish in mind, we investigated the placement of two types of adverbs from the categories mentioned above: manner and frequency. The selection of these two types of adverbs was motivated by their word order in both languages. Manner adverbs in Spanish are allowed in verb-adverb position (verb-raising), but they are dispreferred in adverb-verb position; this word order is more likely to occur in English due to affix lowering. Hence, if cross-linguistic influence from English were to occur in the production of this adverb type in child heritage speakers of Spanish, we would expect noticeable differences between them and their monolingual counterparts in the use of this adverb type. This claim is built upon Camacho and Kirova’s (2018) findings, suggesting a possible cross-linguistic influence effect in relation to adverb-verb placement in heritage Spanish. Furthermore, we focus on these two types of adverbs in order to narrow down the experiment on specific adverb types that have shown variability in previous work rather than including a complete spectrum of adverbs in Spanish that might not necessarily show divergences in the Spanish heritage grammar.

Table 3. Available adverbial positions for ‘manner’ and ‘frequency’ in Spanish and English.

	S–Adv–V–C	S–V–Adv–C	S–V–C–Adv
Spanish	✓	✓	✓
English	✓	X	✓

below provides a summary of the available adverbial positions accounting for both ‘manner’ and ‘frequency’ adverbs.

In what follows, we discuss previous acquisition work in Spanish relative to adverb placement.

3. Previous Research

Few studies in adverb placement/verb movement have been conducted with child populations. One of them was Alvarez (1999), who investigated the acquisition of adverbs in English and Spanish monolingual and bilingual children. The study aimed to make predictions about the usage of adverbs like casi (“almost”) and cabeza abajo (“upside down”), as well as manner adverbs ending in -mente (-ly). The author tested 42 children between the ages of 3 and 6 and presented them with three experiments: two comprehension tasks (Truth Value Judgment tasks) and one comprehension task designed as a role-play or figure manipulation task using puppets, drawings, and illustrations. Out of the total of 39 participants who took part in the study, 21 were monolingual English speakers, 8 were monolingual Spanish speakers, 10 were Spanish speakers with some knowledge of English, and 3 were excluded from the study. Although no significant differences were found between the Spanish and English-speaking participants, the study revealed that children are susceptible to syntactic positioning when interpreting sentences with adverbs. The item cabeza abajo “upside down”, for example, was interpreted as referring to the subject when in sentence-initial position but as referring to entire objects or nouns when in sentence-final position. This suggests that adverbs and adverbial positioning are present very early in life, giving relevance to the interpretation of sentences but also showing children’s understanding of complex morphological categories such as -mente (‘-ly’) adverbs.

Regarding the acquisition of frequency and manner adverbs in English, White (1991) investigated how positive or negative classroom input could enhance English language development among French-speaking children learning English as L2. The study involved 11- and 12-year-old children divided into two groups depending on the type of classroom input: group A, consisting of 82 children trained on adverb placement, and group B, consisting of 56 children instructed on question formation. The participants underwent pre-tests and post-tests, and some received a follow-up assessment one year later. Teaching materials for the group trained on adverb placement concentrated on two adverb types: frequency (e.g., often, sometimes) and manner (e.g., slowly, quickly). The results indicated that child L2 learners who received specific training in adverb placement showed no difficulties in rejecting the ungrammatical structure S-V-Adv-O in English. However, child L2 students who did not receive training on the topic failed to recognize this ungrammaticality, even with further exposure to the English language. Another interesting finding was that child L2 learners, without any instruction, instinctively preferred the S-Adv-V position for frequency adverbs and the S-V-O-Adv position for manner adverbs. This preference could not have been influenced by their native language (French), as the S-Adv-V position is not possible in French.

Ayoun (1999) examined the acquisition of verb movement phenomena in 83 English-speaking L2 learners of French. The study employed a grammaticality judgment task and a production task. In the grammaticality judgment task, participants were presented with 25 sentences and asked to rate them on a scale from 1 to 5, indicating their grammaticality. The production task involved reorganizing 49 grammatical sentences by adding an adverb or a floating quantifier. A control group of 85 native French speakers also performed the same tasks. The results suggested that parameter resetting gradually occurs in the learners’ grammar. Ayoun argues that the performance of the learners, particularly in verb movement, improves significantly over time. This improvement is likely due to parameter resetting reaching a certain level of complexity.

Ayoun’s (1999) results are along the lines of a similar study conducted by Guijarro-Fuentes and Larrañaga (2011). The authors found that L2 learners of Spanish exhibit verb placement that aligns with the parameter settings of Spanish despite their struggle to accurately mark the verb for person and number. The study involved 41 undergraduate students, L2 Spanish speakers, from various universities in the UK, ranging in age from 20 to 37. Participants underwent a placement test, which allowed them to be divided into four groups: beginner (12), low-intermediate (7), intermediate (7), and advanced (15). Additionally, a control group of 8 native speakers from Spain, aged 21 to 55, was included. The researchers administered four tasks: (1) an identification task, (2) a grammar judgment test, (3) a preference grammaticality judgment task, and (4) a production task. The findings showed a clear distinction between syntax and morphology, suggesting that L2 learners can reset the verb movement parameter in Spanish while struggling with the effective use of verb morphology. The authors also offered insights into the acquisition process, noting that although L2 learners of Spanish demonstrate recognition and production of the S-V-Adv-O structure and have access to verb movement from the early stages of acquisition, they often face challenges in employing verb raising and utilizing morphology effectively.

More studies on verb movement phenomena include Ayoun (2005) which examined verb movement among Spanish L2 learners. The study involved 15 participants (11 females and 4 males) who were enrolled in second or third-year Spanish classes at a North American university. Their ages ranged from 18 to 38 years old, with most participants (n = 12) having started learning Spanish between the ages of 14 and 16. Two participants began learning Spanish at age 5 due to having a Spanish-speaking parent at home. The study employed five elicitation tasks: (1) a pre-test to measure the participants’ proficiency independently, (2) a scalar grammaticality judgment task (S-GJT), (3) a preference/grammaticality task (PrefG), (4) a production task (ProdT), and (5) a magnitude estimation acceptability judgment task (ME-AJT). Ayoun concluded that Spanish should be regarded as a “mixed language” concerning verb movement, as it appears to be optional in specific contexts, such as past adverbs in finite contexts and pronominal inversion. This indicates that Spanish incorporates both [+mvt] and [-mvt] features, allowing for certain flexibility in this process.

More recently, Camacho and Sánchez (2017) examined the issue of verb-raising in Spanish monolingual speakers and provided concrete evidence of how this phenomenon has surfaced in Spanish. The authors tested 31 college-age students who were native Spanish speakers from Peru. Two tasks were employed to gather judgments on adverbial word order using VP-oriented adverbs, some derived finishing in -mente (‘-ly’) and some underived such as siempre “always”. In the first task, participants rated sentences on a 5-point Likert scale ranging from ‘sounds very odd/bad’ (=−5) to ‘sounds very good’ (=5). This task examined four specific adverb positions: Adv-V-O, V-Adv-O, Neg-Adv-V-O, and Neg-V-Adv-O. For the second task, participants read sentences and then selected their preferred word order among four options. The results provided support for the verb-raising approach as a crucial factor in adverb placement. This finding is significant as it explains how the V-Adv-O position is constructed. Additionally, the study delved into the role of negation in adverb placement, revealing intriguing data that demonstrated the grammatical possibility and production of the Neg-V-Adv-O position, albeit with lower ratings. In contrast, the position Neg-Adv-V was deemed ungrammatical. Finally, the study concluded that, out of the two possible positions in Spanish (V-Adv and Adv-V), the former received higher ratings and was, therefore, the preferred option.

Camacho and Kirova (2018) examined adverb placement among heritage speakers of Spanish. They tested 34 college-age heritage speakers of Spanish in New Jersey and compared their results with a group of 30 monolingual Spanish speakers from Peru, serving as a control baseline. Using acceptability judgment and selection tasks in affirmative and negative sentences, they discovered that both monolingual and heritage Spanish speakers exhibit similar acceptability patterns but with several differences. One significant finding is that heritage speakers show a particular preference for pre-verbal placement (S-Adv-V-O) in affirmative sentences, although both positions (pre- and post-verbal) are allowed. In negative sentences, both groups prefer S-V-Adv-O over S-Adv-V-O, but heritage speakers have a stronger preference for the latter. Based on these findings and thorough data analysis, Camacho and Kirova concluded that bilingual grammars operate simultaneously, and any aspect of grammar that is compatible in both languages facilitates language processing. This could explain the brief differences observed between the two groups. However, this extended language compatibility can also lead to uncertainty for heritage speakers, resulting in less accurate judgments.

To summarize, previous research provides evidence of verb-raising being the preferred option by adult monolingual speakers of Spanish, and it is fair to claim that adult L2 learners of Spanish master it eventually when they acquire the [+mvt] parameter. Adult heritage speakers of Spanish also seem to follow this path, displaying similar acceptability ratings of adverb placement to those of Spanish monolingual speakers. However, is this also the case for Spanish monolingual and heritage children born and raised in the US?

Bilingual Alignments and Heritage Language Development

In the past two decades, there has been extensive research aimed at unraveling the complexities of heritage language grammar. Scholars have sought to gain a deeper understanding of how heritage speakers acquire and utilize their heritage language (Cuza 2013, 2016; Cuza and Pérez-Tattam 2016; Hur et al. 2020; Montrul 2002, 2016, 2008, 2022; Polinsky 2011, 2018; Polinsky and Scontras 2020; Perez-Cortes et al. 2019; Putnam and Sánchez 2013; Scontras et al. 2018; Shin et al. 2023; Solano-Escobar and Cuza 2023). Recently, Sánchez (2019) proposed the Bilingual Alignment Approach while accounting for the divergences that heritage speakers often show. This approach posits that the speakers’ mind creates connections or “alignments” to store and retrieve information related to different language patterns, structures, and components (e.g., phonology, morphology, syntax). The alignments allow for the merging and inclusion of components from the two languages (they are permeable), resulting in morphosyntactic shifts and mismatches between the production and comprehension of the less dominant language as the components from the two languages come together (López-Otero 2019; Giancaspro and Sánchez 2021). The alignments may be fleeting/temporary depending on the specific patterns of language activation and use for comprehension and production purposes, as well as linguistic proficiency. This leads to transient variability in heritage language grammars at the alignment level, mostly found in production. That is, the alignments are not necessarily part of the linguistic representation but rather allow for the coexistence of stable units (monolingual-type units) and temporary units (dominant language units), which are normally accessed in online production. However, frequent and longer activation of the temporary units stored in memory may cause those alignments to stabilize/get reassembled, leading to the grammatical reconfiguration of some of the features of the heritage/less dominant language (Cuza and Sánchez 2022). Sánchez’s (2019) proposal accounts nicely for some of the difficulties often observed among Spanish heritage speakers in contact with English as the dominant language. Heritage speakers with more proficiency and higher patterns of language activation for production and comprehension purposes would be more likely aligned with Spanish morphosyntactic patterns than with English ones, reducing their level of morphosyntactic shifts. However, speakers with more proficiency and exposure to English may eventually stabilize their alignments with English at the level of internal representation, leading to the restructuring of the heritage language.

In what follows, we discuss our research questions and hypotheses, followed by a description of the participants, tasks, and procedure.

4. The Study

4.1. Research Questions and Hypotheses

The main goal of the present study was to investigate the extent to which cross-language interaction, heritage language dominance, and linguistic experience modulate the patterns of adverb placement in Spanish among English-dominant child heritage speakers of Spanish born and raised in the US. Specifically, we postulate the following research questions and hypotheses:

RQ1: To what extent are Spanish heritage children affected by cross-linguistic influence from English in their placement of manner and frequency adverbs? In other words, does adverb placement in heritage Spanish show effects of CLI from English?

H1a. 

Heritage Spanish children will show higher production of SAVO word order in both manner and frequency adverbs compared to other available orders.

H1b. 

Heritage Spanish children will show low use of verb-raising structures (SVAO word order) in Spanish.

RQ2: Do patterns of language use, exposure, and dominance influence the patterns of adverbial placement among child heritage speakers of Spanish?

H2. 

Heritage Spanish children with higher levels of Spanish exposure, use, and dominance will show higher production of sentence-final adverbial placement and verb-raising position, comparable to monolingual-like patterns. In other words, greater language dominance, use, and exposure to Spanish will lead heritage children to align more closely with the preferred responses in monolingual children (Putnam and Sánchez 2013).

4.2. Participants

A total of 37 (n = 39) participants participated in this study: 14 child heritage speakers of Spanish born and raised in the US (age range: 7;9–10;7; M = 9;3; SD = 1.28) and 25 monolingual children from Mexico (age range: 7;2–11;8; M = 9;6; SD: 1.43) participating as baseline group. All participants completed a consent form and a language background questionnaire that requested information about the children’s language patterns of language use, exposure, fluency, and general linguistic experience in both Spanish and English (Shin et al. 2023). The background questionnaire was completed by one of the parents and elicited patterns of language usage and exposure using a 5-point Likert scale ranging from “never” (=1) to “very frequently” (=5). Parents were asked to report the patterns of language exposure and usage in both Spanish and English with family members (mother, father, grandparents, and siblings) and friends. Similarly, the questionnaire elicited the children’s fluency in both languages using a scale ranging from “not fluent” (=1) to “completely fluent” (=4) (Pérez-Leroux et al. 2011).

The heritage children were tested and recruited in the American Midwest. They all had Mexican-born parents and were exposed to both languages from birth. Testing took place in their school setting as part of an afterschool literacy program in Spanish. Their mean result for Spanish usage was 3.3/5, and for exposure was 3.4/5. As for English, their mean usage was 3.2/5, and exposure was 2.9/5. Regarding fluency, their mean proficiency in Spanish was 1.8/4 and 3.5/4 in English. The Spanish monolingual children were recruited and tested in Guanajuato, Mexico (2nd–6th graders). All of them were born and raised in Mexican families from different cities around the country (Guanajuato, Irapuato, León, San Luis Potosí). The children were tested in the school setting by the researchers. Their reported mean for English usage was reported as 2/5, and exposure was 2/5. For Spanish, parents reported 4.9/5 for usage and 4.9/5 for exposure. Finally, their English fluency was reported as ‘not fluent’ (1.9/4), opposite to their Spanish, which was reported as ‘very fluent’ or ‘completely fluent’ (3.2/4).

To determine the heritage children’s level of dominance, the Multilingual Naming Test (MiNT) was administered (Gollan et al. 2012). This test has been successfully used before as a reliable dominance measure (Hur et al. 2020; Sheng et al. 2014). The dominance score was obtained by dividing the number of correct answers by the number of tokens and then multiplying it by 100. The results in percentage were interpreted in three possible ways. If the participants showed a larger than 5% English advantage over their Spanish dominance, they were classified as English-dominant; if they showed a larger than 5% Spanish advantage over their English dominance, they were classified as Spanish-dominant. Participants showing less than 5% between-language difference in either direction were classified as balanced bilinguals. Participants completed this test with the help of the researcher, who presented each item and asked participants to name them. Scores from the MiNT reported that 93% (13/14) of the heritage children were English dominant, 7% (1/14) were balanced, and none of them were Spanish dominant. The MiNT was only taken by the heritage/bilingual children. A summary of the participant’s profile can be found in

Table 4. Summary of participants’ background questionnaire.

	Heritage Children (n = 14)	M	SD	Monolingual Children (n = 25)	M	SD
Age	(7;9–10;7)	9;3	1.28	(7;2–11;8)	9;6	1.43
MiNT	Span	41.4	12.95	SPAN only
	Eng	75.6	8.33
Span Usage		3.3/5	0.73		4.9/5	0.67
Eng Usage		3.2/5	0.59		2/5	0.47
Span Exposure		3.4/5	0.79		4.9/5	0.17
Eng Exposure		2.9/5	0.66		2/5	0.44
Self-Reported Span Fluency		1.8/4	0.59		3.2/4	0.55
Self-Reported Eng Fluency		3.5/4	0.46		1.9/4	0.82
Place of birth	USA			Mexico

below.

4.3. Tasks and Procedures

To investigate our research questions and hypotheses, we conducted an elicited production task (Castilla-Earls et al. 2020; Cuza and Pérez-Tattam 2016; Shin et al. 2023). With the aid of PowerPoint, the participants were presented with a preamble and a prompt. To help contextualize the expected response, an image was provided in addition to the verb and adverb that the participants were asked to use. Both verb and adverb were located at random places for each item to control for presentation order effects (8).

(8)
Preamble: Sergio es un chico muy tranquilo, y siempre que va a la biblioteca, lee sus libros de forma calmada, sin hacer mayor ruido. “Sergio is a very quiet kid. He always goes to the library and reads his books without making any noise.”
Prompt: ¿Cómo lee Sergio sus libros?
	“How does Sergio read his books?”
		(Here appeared an image of a child reading a book followed by the verb leer “to read” and the adverb tranquilamente “quietly”)
Possible/expected responses:
	A)	Sergio lee sus libros tranquilamente.
		“Sergio reads his books quietly”
	B)	Sergio tranquilamente lee sus libros.
		“Sergio quietly reads his books”
	C)	Sergio lee tranquilamente sus libros.
		“*Sergio reads quietly his books”

Following previous work (Alvarez 1999; Camacho and Kirova 2018; Camacho and Sánchez 2017), we tested word order patterns with VP-oriented adverbs ending in -mente (-ly): There were 5 frequency adverbs and 5 manner adverbs. Frequency adverbs included frecuentemente “frequently”, usualmente “usually”, regularmente “regularly”, normalmente “normally” and generalmente “generally”. Manner adverbs included seguramente “safely”, correctamente “correctly”, tranquilamente “quietly”, inmediatamente “immediately”, cuidadosamente “carefully”. An affirmative and negative polarity were also included, resulting in 10 items for affirmative and 10 items for negative contexts. All the adverbs and verbs tested were frequent, according to the Davies corpus of Spanish. They were also semantically transparent and interchangeable. There were 20 test items, 20 distractors, and two practice items. The distractors were part of a larger separate study examining subject pronoun distribution. All items were randomized and counterbalanced across participants.

A summary of the conditions under examination is presented in

Table 5. Structures under analysis for affirmative polarity.

Placement	Structure	Example
SVOA	S–V–O–Adv	Ella come tamales frecuentemente. “She eats tamales frequently”
SAVO	S–Adv–V–O	Ella frecuentemente come tamales. “She frequently eats tamales”
SVAO	S–V–Adv–O	Ella come frecuentemente tamales. “*She eats frequently tamales”
OTHER	Adv. Periphrasis/Other	Ella come tamales de manera frecuente. “*She eats tamales in a frequent way”

and

Table 6. Structures under analysis for negative polarity.

Placement	Structure	Example
SNVOA	S–Neg–V–O–Adv	Ella no come tamales frecuentemente. “She doesn’t eat tamales frequently”
SNAVO	S–Neg–Adv–V–O	Ella no frecuentemente come tamales. “She doesn’t frequently eat tamales”
SNVAO	S–Neg–V–Adv–O	Ella no come frecuentemente tamales. “*She doesn’t eat frequently tamales”
OTHER	Adv. Periphrasis/Other	Ella no come tamales de manera frecuente. “*She does not eat tamales in a frequent way”

. All responses were recorded for later analysis. Coding was conducted by one of the first authors (a native speaker of Spanish).

In the following section, we present the results by group on both adverb types. Furthermore, we present the results on the potential role of language dominance and experience among the heritage children group.

5. Results

5.1. Manner Adverbs

Results with manner adverbs affirmative showed a preference for sentence-final position (SVOA word order), with 67% of the heritage children and 74% of the monolingual children producing this option. About 30% of the heritage speakers and 10% of the monolingual children produced a pre-verbal adverb (SAVO position). Around 16% of the monolingual children showed production of post-verbal adverbial position (SVAO), but only 1% of heritage speakers did so (Figure 1). Manner adverbs with negative polarity showed a different range of production. Position SNVAO was produced at 47% by the monolingual group and 19% by the heritage group. SNVOA was produced at 38% by the monolingual group and was highly produced by heritage speakers in this context at 27%. SNAVO was used 7% by the monolingual children and 23% by the heritage children. Finally, SANVO (an unexpected word order) accounted for 6% of production by the monolingual children and 14% by the heritage children. Other instances of unexpected production accounted for 2% in monolingual children and 17% in the heritage group (Figure 2).

A generalized linear model (GLM) analysis performed in R software version 2023.06.02 Build 561 (R Core Team 2020) was conducted with a multinominal probit distribution to examine the patterns that emerged across all different positions and groups with affirmative and negative manner adverbs. The responses were coded as “1” for SVOA, “2” for SAVO, “3” for SVAO, and “0” for other instances. It is important to remark that several GLMs were conducted in this study. To examine the role of the different covariables, two different models were conducted for each of them: one model including the covariable and one model without the covariable. This allowed us to make comparisons between both models and check whether the covariable had any effect on the dependent variable. Moreover, as an individual analysis was made for each covariable, the order of variable entry and effects of multicollinearity should not represent a concern in the analysis.

Results with the independent variable as group, and the dependent variable as the response showed a main group effect with manner affirmative adverbs (χ² = 24.61, p < 0.001) and manner negative adverbs (χ² = 36.65, p < 0.001). The groups behaved significantly differently from each other regarding the position of affirmative and negative manner adverbs (see Supplementary File S1 for complete results).

A GLM analysis was conducted to determine the role of language experience and proficiency with response as the dependent variable. The results showed a main effect with Spanish usage (χ² = 12.832, p < 0.005) and English usage (χ² = 8.029, p < 0.045) in the production of affirmative manner positions. The more Spanish usage the heritage speakers had, the more likely they were to produce sentence-final adverbial position (SVOA) and the less likely they were to produce pre-verbal adverbs (SVAO) (Figure 3). Opposite results were found for English usage. The more English usage the heritage children had the more likely they were to produce the preferred English word order (SAVO) and less likely to produce sentence-final adverbial placement (SVOA) (Figure 4). The heritage children also showed a low probability of producing the SVAO word order and OTHER structures. There was no main effect with Spanish exposure (χ² = 5.846, p < 0.119) or English exposure (χ² = 2.084, p < 0.554).

Regarding English dominance, results showed a significant main effect with manner adverb placement (χ² = 12.817, p < 0.005). The more dominant the heritage speakers were in English, the more likely they were to produce the SAVO position (preferred option in English). However, there was no main effect of Spanish dominance in the production of manner affirmative positions (χ² = 2.041, p < 0.563). This is to say, Spanish dominance played no role in the likelihood of producing one position vs. another. Regarding manner negative adverbs, results showed a significant effect with Spanish dominance (χ² = 16.960, p < 0.001) (Figure 5). The more dominant the heritage children were in Spanish, the more likely they were to produce final-sentence adverbial position (SNVOA). Opposite to that, SNAVO and SNVAO positions showed low probability in production together with OTHER structures as Spanish dominance increased.

These results were consistent with other covariables, showing a similar probability between word order type and Spanish usage (χ² = 32.662, p < 0.001) and exposure (χ² = 31.837, p < 0.001). That is, the more Spanish usage and exposure the more likely they were to produce the final-sentence adverb position. English usage (χ² = 47.434, p < 0.001) and exposure (χ² = 21.855, p < 0.001) were also significant, displaying the opposite pattern. The more usage and exposure to English, the higher the likelihood to produce SNAVO and the less likely to produce positions SNVAO and SNVOA. There was no main effect for English dominance (χ² = 2.255, p < 0.521).

5.2. Frequency Adverbs

The production of frequency affirmative adverbs among the monolingual children showed a different pattern of positioning compared to that of manner adverbs. For instance, the most preferred positions by the monolingual children in this context were SAVC with 46%, followed closely by SVOA with 44%. Heritage children showed a similar trend in these two positions (50% for SAVO and 46% for SVOA). About 10% of the monolingual children also showed production of SVAC word order (Figure 6). Frequency negative adverbs seemed to have a balanced production across all positions. However, there was an unexpected use of SNAVO word order, which was the preferred answer by both the monolingual children (25%) and heritage children (33%). Monolingual children used the unexpected SANVO and SNVOA word orders at 22%, and SNVAO was used at 20%. Similarly, SANVO was produced at 21% by heritage children, followed by SNVOA (14%) and SNVAO (13%). OTHER instances of production accounted for 19%, a percentage that increased in comparison to the monolingual speakers (10%) (Figure 7).

Results from a GLM analysis showed significant differences between groups in the production of frequency adverbs with affirmative polarity (χ² = 9.918, p < 0.019). Nonetheless, this was not the case for frequency-negative adverbs, where no significant differences were found (χ² = 6.169743, p = 0.1868) (see Supplementary File S2 for complete results).

Regarding dominance in Spanish, a separate GLM analysis showed a significant main effect (χ² = 19.581, p < 0.001). The production of the SAVO position decreased with higher Spanish dominance, and the SVOA position increased with higher dominance. This suggests an association between Spanish dominance and sentence-final adverbial use in Spanish (Figure 8). Dominance in English showed a similar trend (χ² = 10.106, p < 0.017) but with opposite probabilities of production. As English dominance increased, the production of pre-verbal placement (SAVO) also increased (preferred position in English), but the production of final adverb placement (SVOA) decreased (Figure 9):

Similarly, for Spanish usage (χ² = 24.588, p < 0.001) and exposure (χ² = 16.308, p < 0.001), the SVOA position kept showing an upward trend, and the SAVO a downward trend. This contrasted with English usage (χ² = 36.018, p < 0.001) and exposure (χ² = 21.948, p < 0.001), where SAVO showed an upward trend and SVOA showed a downward trend. This means the more Spanish usage and exposure, the more probability there was for the heritage children to produce the SVOA position; however, the more English usage and exposure, the more likely they were to produce the SAVO position.

For frequency negative adverbs, there were some specific positions that seemed to be favored despite no significant differences between groups. This was the case with Spanish usage (χ² = 12.577, p < 0.005), which increased the production of SNVOA and lowered the production of other positions available (Figure 10). Similarly, as Spanish exposure increased (Figure 11), the more likely the heritage children were to produce the SNVOA position (χ² = 13.901, p < 0.003). English usage (χ² = 15.547, p < 0.001) showed an increase in the production of SNAVO and a decrease in the production of SNVAO and SNVOA, as shown in previous patterns. English dominance (χ² = 0.540, p < 0.909) and English exposure (χ² = 5.704, p < 0.126) did not show effects for this context.

5.3. Results for Adverbial Word Order

A generalized linear mixed effects model with binomial distribution was also conducted in order to examine the variation between groups and adverbial word order, more specifically, in the likelihood of producing a specific adverbial placement in different contexts. Consistent with prior findings, differences in adverb production emerged for both groups; manner adverbs were the type with more distinctive variation in production in both affirmative and negative (

Table 7. Adverbial distribution by group (manner adverbs).

Adv. Position	Manner Adverbs	Estimate	SE	Z Value	Pr (>|z|)
S-Adv-V-O	(Intercept)	−1.5136	0.6914	−2.1890	0.0286
	Monolingual Children	−1.6571	0.8230	−2.0140	0.0441
S-V-O-Adv	(Intercept)	1.2540	0.6509	1.9270	0.0540
	Monolingual Children	0.3521	0.7738	0.4550	0.6490
S-V-Adv-O	(Intercept)	−5.2250	1.3090	−3.9910	0.0001
	Monolingual Children	2.8470	1.2660	2.2490	0.0245
S-Neg-Adv-V-O	(Intercept)	−0.8235	0.5470	−1.5060	0.1322
	Monolingual Children	−1.8521	0.7314	−2.5320	0.0113
S-Neg-V-O-Adv	(Intercept)	−1.0546	0.3190	−3.3060	0.0009
	Monolingual Children	0.5152	0.3811	1.3520	0.1764
S-Neg-V-Adv-O	(Intercept)	−1.4781	0.3074	−4.8090	0.0000
	Monolingual Children	1.3660	0.3558	3.8400	0.0001

), and frequency adverbs were the ones with the least differences for both polarities (

Table 8. Adverbial distribution by group (frequency adverbs).

Adv. Position	Frequency Adverbs	Estimate	SE	Z Value	Pr (>|z|)
S-Adv-V-O	(Intercept)	0.0854	0.7878	0.1080	0.9140
	Monolingual Children	−0.3549	0.9857	−0.3600	0.7190
S-V-O-Adv	(Intercept)	−0.3035	0.6559	−0.4630	0.6440
	Monolingual Children	−0.1574	0.8164	−0.1930	0.8470
S-V-Adv-O	(Intercept)	−4.2340	1.0070	−4.2040	0.0000
	Monolingual Children	1.9920	1.0520	1.8930	0.0583
S-Neg-Adv-V-O	(Intercept)	0.4730	0.8778	0.5390	0.5900
	Monolingual Children	−0.5945	1.0999	−0.5410	0.5890
S-Neg-V-O-Adv	(Intercept)	−2.7385	0.7775	−3.5220	0.0004
	Monolingual Children	0.6681	0.8461	0.7900	0.4297
S-Neg-V-Adv-O	(Intercept)	−2.6746	0.7049	−3.7940	0.0001
	Monolingual Children	0.6830	0.7483	0.9130	0.3614

).

To sum the results for manner adverbs with the positions in Table 7, there is significant evidence to confirm a difference between the heritage children and the monolingual children in the likelihood of producing the pre-verbal positions S-Adv-V-O (Z-Wald = −2.0140, p < 0.044) and S-Neg-Adv-V-O (Z-Wald = −2.5320, p < 0.011). This is to say, there is more probability for the heritage group to produce these orders in comparison to the monolingual group. Additionally, significant evidence was also found for the verb-raising positions S-V-Adv-O (Z-Wald = 2.2490, p < 0.024) and S-Neg-V-Adv-O (Z-Wald = 3.8400, p < 0.0001), for which the values suggest that the monolingual children have a higher likelihood of producing this word order compared to the heritage children.

As for frequency adverbs, Table 8 below shows a moderate significance with the verb-raising position S-V-Adv-O (Z-Wald = 1.8930, p < 0.058), indicating that monolingual children were likelier to produce such word order than the heritage children. Apart from that word order, the analysis for the other positions and this adverb type does not show any more significant differences in production.

6. Discussion and Conclusions

The first research question posed in our study aimed at analyzing if heritage children displayed any potential cross-linguistic influence effects from English. Overall, we found that the word order patterns in negative and affirmative sentences with manner and frequency adverbs were similar among both groups. However, some key differences emerged from the data in relation to the probability of using certain word orders vs. others. For example, the heritage children showed a higher probability of pre-verbal adverbial use compared to the monolingual children with affirmative (S-Adv-V-O) and negative (S-Neg-Adv-V-O) manner adverbs (Figure 1 and Figure 2) as well as with negative frequency adverbs (Figure 7). Results also suggest that the largest difference in position preference occurred in negative sentences with manner adverbs. In this context, the heritage children used other strategies most frequently followed by the adverb-final option, compared to monolingual children, who used S-Neg-V-Adv-O most frequently, also followed by the adverb-final option.

Since heritage children used the order S-Adv-V-O in affirmative sentences more frequently than monolingual children, H1a was confirmed. Following similar trends found by Camacho and Kirova (2018) with heritage adults, these results can be accounted for in relation to cross-linguistic influence from English into Spanish for affirmative sentences. English favors pre-adverbial position, and that was the word order pattern that heritage children produced the most. However, this needs to be taken with caution since this explanation would not account for its negative counterpart. Our results suggest that verb-raising among monolingual children is only the preferred option for negative sentences with manner adverbs. For heritage children, the percentage of pre-verbal position also rises in that context, but it is not the first option. The difference between heritage and monolingual children cannot be a result of direct, linear CLI from English, as that would require do-support. It is possible that indirect CLI results from converging in the preferred Adv-V order in English but without the syntactic mechanism of do-support. Hence, our first hypothesis can only be partially confirmed as heritage children do demonstrate knowledge and use of the S-Adv-V-O order, but it might not be directly due to a language transfer effect from English in all cases. In future work, it would be interesting to examine bilingual children whose other language is not restricted to pre-verbal adverbs to examine more carefully the role of cross-linguistic influence on this specific domain.

Regarding the verb-raising position (S-V-Adv-O), we found a low rate of production among the heritage children compared to the monolingual children. Camacho and Sánchez (2017) argued that this position is preferred by monolingual Spanish speakers as it is computationally less costly to produce compared to the pre-verbal position, which needs extra operational processing. This finding could account to a certain extent for the results found with the monolingual children and their slightly higher production of the verb-raising position compared to the heritage children, but this leaves the question as to why the heritage children do not choose to use it if it is less costly. Assuming that the [+mvt] Spanish feature is already developed by the heritage children, we would expect them to show a preference of production for the verb-raising order; nonetheless, the operationally more costly order (pre-verbal) is still more produced. Two rather appealing ideas come out of these results and reasoning. One of them confirms Ayoun’s (2005) claims that Spanish is a “hybrid” or “mixed” language with respect to verb movement, meaning that it is rather flexible in its parameter settings and, therefore, allows several adverb orders. This claim would help explain both sides as verb movement would be seen as an optional process but might appear simplistic and arbitrary considering the high variability of production found in the study. There is no doubt that adverbs are syntactically variable in their positioning, and with this syntactic variation, the likelihood of a semantic interface playing a role should not be completely disregarded. As seen for monolingual children, the patterns of production for S-Adv-V-O for frequency adverbs are significantly different from those for the same position with manner adverbs. Possibly, a computationally less or more costly account for choosing verb-raising may not only be restricted to a syntactic procedure but also to a semantic constraint that computes this process depending on the type of adverb and the semantic information they possess. So far, this is no more than simple theorizing, but what is certain is that research on adverb placement would benefit from a future semantic point of view as well.

The last appealing idea is regarding the overall language environment. Given the community characteristics of the participants (high exposure to English), the verb-movement feature might still be difficult to process in heritage grammars. In fact, our second research question inquired about the extent to which patterns of language use, exposure, and dominance influenced adverbial placement, as would be predicted by Sánchez’s (2019) bilingual alignment hypothesis. The results support our expectations. Both experience and dominance played an important role in the production of one position vs. another, at least with the tested adverb types. We can see how the sentence-final position (S-V-O-Adv) and its negative counterpart (S-Neg-V-O-Adv) were more likely to be produced by heritage speakers with higher Spanish dominance and higher patterns of language use and exposure. Likewise, the production of pre-verbal adverbial position (S-Adv-V-O) increased with higher dominance in English, use, and exposure. These results add support to Sánchez’s (2019) approach, suggesting a strong association between morphosyntactic variability and the level of linguistic dominance, language use, and exposure in heritage language acquisition (Hur et al. 2020; Putnam and Sánchez 2013). The higher the patterns of language dominance, use, and exposure in one specific language, the more likely the heritage speakers align with such language. Heritage speakers, through their dynamic and individual patterns of language activation, tend to gravitate toward aligning with their more dominant language, leading to morphosyntactic variability. However, we can only argue for variability at the level of bilingual alignments as far as production is concerned. Future research testing both production and receptive knowledge (preference or interpretation) would be necessary to examine the extent to which the English-like alignment has led to grammatical restructuring/reassembly as far as adverbial placement is concerned in heritage Spanish. Furthermore, future research would benefit from examining older Spanish heritage children (adolescents) to evaluate the extent to which these patterns remain or are somewhat overcome with increasing age and exposure to the heritage language (see Daskalaki et al. 2023 for recent discussion in relation to developmental rate vs. differential path in child heritage speakers).

Our results also raise the need for a more nuanced analysis of raising in Spanish since we find that patterns vary substantially depending on the type of adverb and whether the context is affirmative or negative. For example, the highest preference for raising among both groups happened in negative contexts with manner adverbs, compared to negative sentences with frequency adverbs and affirmative sentences with both types. Future research would benefit from examining the receptive knowledge that child heritage speakers of Spanish have in relation to adverb placement. This is a limitation of the current study and an important issue to examine in future work with child heritage speakers and their potential variability regarding adverbial placement.