Critical Language Awareness in the Spanish as a Heritage Language College Classroom

Abstract

This article analyzes the critical language awareness (CLA) of Spanish university-level students who were enrolled in a 16-week Spanish heritage language (SHL) course, using CLA as an instructional approach. Students’ attitudes towards bilingualism, Spanglish, language variation, and prescriptivist grammar were measured via pretest and posttest surveys that used a four-point Likert scale of strongly agree, agree, disagree, or strongly disagree, along with a text box asking participants to explain their answers. The CLA instructional methods delivered in the course included the analysis of code-switching grammar constraints, the study of standard language ideologies and monolingual language ideologies, the analysis of stigmatized grammar features found in varieties of Spanish, and English-influenced lexicon. Ten out of fourteen items were included in a factor analysis which yielded a statistically significant change between pretest and posttest. Qualitative analysis of answer explanations showed that some students adapted their language ideologies to the new information and did not change their beliefs at a deep level. Future CLA research should identify “CLA proficiency” levels as well as how to differentiate for students who hold deeply entrenched language ideologies.

1. Introduction

The main objective of this article is to offer preliminary evidence of the effectiveness of critical language awareness (CLA) in a Spanish as a Heritage Language (SHL) class and to point out the need to identify “CLA proficiency” levels that require differentiation. Calls have been made for a social-justice oriented approach to HL instruction that deals with the challenges of linguistic insecurity, language loss, and linguistic disenfranchisement of speakers of a minoritized language (e.g., Carreira 2000; Correa 2011, 2016; Ducar 2009; Holguín Mendoza 2018; Holguín Mendoza et al. 2018; Martinez 2003; Parra 2016; Villa 2002). The field of heritage language instruction has long been preoccupied with effective approaches to teaching language arts to students who may or may not know a prestige variety (e.g., Villa 1997; Valdés 1981); these approaches are in part driven by the different motivations of HL students for taking HL courses, which include the improvement of their writing (Mikulski 2006) and to learn “correct” Spanish (Ducar 2008). However, what initially was an effort to break into the codes of power ended up being its own type of oppression, in that, historically, HL instruction had used an eradicationist or appropriateness model of teaching a standardized variety to students (Leeman 2005).

It has long been acknowledged that HL learners have different needs than L2 learners (e.g., Burgo 2017; Carreira and Kagan 2018; Kagan and Dillon 2018; Leeman et al. 2011; Potowski 2005; Torres et al. 2019), which require a tailored teaching approach. The instructional goals for heritage language education include explicit efforts in language maintenance, acquisition of a prestige language variety, expansion of bilingual range, transfer of literacy skills, development of academic skills, development of positive attitudes towards all varieties of the HL and its cultures, and development of cultural awareness (Beaudrie et al. 2014; Martinez 2016; Valdés and Parra 2018). Responding to the need for a social-justice pedagogy, CLA not only raises awareness of the issues of language and power structures in society but also builds agency in challenging dominant language ideologies that ultimately oppress speakers of minoritized languages (Beaudrie et al. 2021; Fairclough 1992). CLA is part of an ongoing “critical turn” in SHL pedagogy in which educational equity is the focus of research and praxis (Beaudrie and Loza 2022).

Studies incorporating sociolinguistic content into instruction have shown promising results (e.g., Holguín Mendoza et al. 2018; Shin and Hudgens Henderson 2017). CLA, however, goes beyond the inclusion of sociolinguistic content and calls for what Leeman and Serafini (2016) call a “critical translingual competence”. In contrast to prescriptivist pedagogies that attempt to “fix” the heritage learner, in CLA, the learner’s variety is central to instruction and students are made aware of the social and political functions of different language varieties (Correa 2016). CLA is fundamental to modern HL instruction because it cultivates educational equity and helps eradicate language disparities in education (Correa 2010; Leeman 2005, 2018; Leeman et al. 2011; Martinez and Foulis 2022). Accordingly, Beaudrie and Wilson (2022) argued that developing CLA among HL learners should be the first goal of HL instruction and that each of the subsequent curricular and instructional goals should integrate CLA in their implementation.

Recent studies investigating CLA among HL learners have found that students tend to have somewhat high levels of preexisting CLA even prior to a curricular intervention, likely due to their social and linguistic positioning as speakers of a minoritized language and culture. Gasca Jiménez and Adrada-Rafael (2021) measured HL students’ preexisting CLA with a survey developed by Beaudrie et al. (2019), finding that learners who had completed three or more courses in a university Spanish program had higher CLA, although they also held standard language ideologies and stigmatized code-switching. In a curricular intervention study, Beaudrie et al. (2021) implemented four modules of CLA over the course of a semester with intermediate-level Spanish heritage learners. On the posttest, 14 out of 17 students had increased their levels of CLA, with students signaling an awareness of linguistic diversity within one language as well as an awareness of the differential treatment of certain language varieties.

Other studies in CLA curricular effectiveness have focused on how student agency is connected to language attitudes. Holguín Mendoza (2018) investigated Spanish HL students’ language attitudes towards “nonstandard” language (loanwords from English, “rural” variants, slang, and “standard” distractors) at the beginning and end of a university term. The students were recruited from two intermediate HL courses within a Spanish HL program that infused critical approaches into the curricula. Students indicated on a survey if they considered certain words to be “correct” or “incorrect” and if they themselves would say these words. The author found that students were more accepting of commonly stigmatized forms over time, although she acknowledged that attitudinal change does not necessarily cooccur with agency and social change. Herrera-Dulcet (2019) investigated sociolinguistically informed critical pedagogies in Spanish heritage courses to determine the impact on students’ critical translingual development and agency beyond the classroom walls. HL students in an advanced Spanish heritage course participated in focus group interviews, surveys, and follow-up interviews four months after the CLA-style curricular intervention. She found that participants grew in their awareness of their own language varieties as well as recognizing linguistic discrimination; additionally, some participants engaged with a “language expert identity” in which they challenged standard language ideologies in their local community.

Service-learning projects have also been found to change negative language attitudes and promote student agency (e.g., Leeman 2011; Leeman et al. 2011; Lowther Pereira 2015; Martinez and Schwartz 2012; Parra 2013; Pascual y Cabo et al. 2017; Villa 2010). Lowther Pereira (2015) investigated how a service-learning project built linguistic confidence, sociolinguistic awareness, and language skills in students enrolled in an advanced Spanish HL course. Choosing from among tasks such as community interpreting, tutoring Latino schoolchildren, and mentoring, students completed reflections and investigative team projects. In questionnaires and reflections, students reported feeling connected to the local Spanish-speaking community and becoming more aware of cultural and linguistic diversity as they themselves were agents of linguistic choice. Rabin and Leeman (2015) discussed how three service-learning programs (an after-school Spanish for heritage speakers class at an elementary school, a bilingual book club at a middle school, and a Spanish and ESL literacy class for adult workers) connected back to the coursework in Spanish literature courses. By integrating literary studies, the study of language ideologies, and critical service-learning, the authors argued that students’ critical consciousness stimulated them to see themselves as agents of social justice beyond the classroom walls. Crucially, critical service learning is distinguished from traditional service learning by its commitment to social justice; projects must challenge students to examine the root causes of social inequities as well as take action to remedy them (Mitchell 2008, cited in Lowther Pereira 2022). In SHL education, critical service learning involves interrogating language ideologies and power relations in linguistic choice (Lowther Pereira 2022).

As the variety of these studies show, CLA can be implemented with different tasks and measured in multiple ways. More studies are needed that offer empirical evidence of the effectiveness of CLA to bolster its theoretical and instructional design and to offer compelling evidence to HL educators who may be reluctant to challenge societal power structures in their classrooms (Beaudrie et al. 2019). This study meets the need for more investigations into the effectiveness of CLA in SHL classrooms, in addition to advancing the field’s understanding of CLA methodologies. Because CLA pedagogy can manifest in different ways based on instructor knowledge and experiences or student needs, it is important that SHL critical educators be aware of the array of instructional choices they must make as well as the range of CLA that students have when they enter the classroom. The following question guided this study: does CLA instruction increase HL students’ CLA? The principal conclusions are that students did demonstrate statistically significant increased CLA after a semester of instruction, yet some students adapted their preexisting language ideologies to fit new information. Students with deeply entrenched language ideologies need differentiated strategies in order to move beyond a surface-level acceptance of linguistic equality and bilingual linguistic practices.

2. Materials and Methods

2.1. SHL Course and CLA Curriculum Description

The course investigated in this study is an upper-division SHL class taught at a small public university in Minnesota over 16 weeks. Taught in the fall semester, it typically attracts first- and second-year students who identify as native speakers of Spanish, heritage speakers of Spanish, and students who have attended Spanish immersion K-12 programs. Students’ Spanish oral proficiency must be at least at the intermediate level or above to enroll. Active efforts are made to recruit students during freshman registration periods since the course helps students meet general education graduation requirements and also serves as an entry point for native and heritage speakers into the Spanish major and minor programs. The Spanish language proficiency of participants was not formally measured; students self-enroll or advisors assign them to the course based on the course description, which includes verbiage about requiring at least an intermediate speaking level. The students’ language use patterns at home or other contexts were not investigated beyond data collected on the survey.

This course was created to match another entry-point course at the institution that was originally designed for nonnative speakers. While the nonnative speakers’ bridge course focuses on building oral proficiency and reviewing grammar in addition to developing academic writing fluency in Spanish, the native/heritage speakers’ course teaches surface-level writing convention skills such as accent marks, spelling, punctuation, upper- and lower-case letters, as well as deeper-level literacy skills such as writing effective essays, integration of research into an argument, and citing sources. In addition to these writing skills, the native/heritage speakers’ course uses CLA-based thematic units that discuss concepts of bilingualism, code-switching, the three sociolinguistic generations, and “nonstandard” grammar forms. The course focuses on building CLA around topics affecting Spanish-speakers in the U.S., such as language ideologies, bilingual linguistic practices often referred to as “Spanglish”, and the maintenance of Spanish among immigrant families.

Starting in the Fall semester of 2019, coursework included a service-learning project in which the students create and teach minilessons on Spanish written conventions (such as spelling with “b” or “v”, the difference between “porque” and “por qué”, comparing and contrasting capital letter usage in the days of the week in English and Spanish) to native and heritage speakers of Spanish at a local high school. The final semester of data collection (Fall 2020) occurred online due to the pandemic; the course was taught synchronously online, and the service-learning project was converted to a recorded video minilesson with no in-person visit to the high school.

Table 1. Key concepts delivered across all five semesters.

Theme	Key Concepts	Sample Learning Activities
Bilingualism	Definitions and misconceptions of “bilingual” Spanglish features The historical and current presence in Spanish in the US	Language autobiography Recognizing and distinguishing loanwords, semantic extensions, calques, code-switching
Language Attitudes	Definition and misconceptions of “language”, “dialect”, “register” prescriptivism versus descriptivism “standard language” versus “non-standardized language” language attitudes and discrimination language ideologies Mock Spanish	Recognizing patterns behind use of verbals Comparing and contrasting English and Spanish grammar: articles, passive voice, gerunds Distinguishing Mock Spanish from Spanglish
Heritage Language Learners	Three sociolinguistic generations Language shift Majority and minoritized languages How the learning needs of HL learners are different from L2 learners	Personal essay on Spanish use in family Personal essay on experiences with Spanish/ESL in school Examining a spoken-word poem on language identity

below outlines the CLA themes taught across every semester. Anecdotally, students often report that by the end of the course they have a different perspective on Spanish itself and plan to make special efforts to help their younger siblings or children maintain Spanish.

2.2. Participants and Instructor

A CLA survey was administered to 60 students enrolled in an SHL course taught over five semesters (2017–2020), of which 56 consented to participate in the study. Of these participants, 52 completely responded to all items on both the pretest and the posttest survey and are considered in this analysis. Effective CLA instruction and measurement rely on having a complete picture of the students since language use and language beliefs can impact instruction; as discussed further on, CLA instruction must make itself flexible for a multitude of experiences.

Table 2. Gender and age of participants.

	Gender			Age at Time of Posttest
	F	M	Other	18–19	20–21	22–23	24+
Spring 2017 (n = 9)	7	2	0	3	2	3	1
Fall 2017 (n = 15)	10	5	0	13	2	0	0
Fall 2018 (n = 9)	6	2	1	6	1	1	1
Fall 2019 (n = 14)	10	4	0	9	2	2	1
Fall 2020 (n = 5)	5	0	0	3	0	1	1
Total (N = 52)	38 (73%)	13 (25%)	1 (2%)	34 (65%)	7 (13.5%)	7 (13.5%)	4 (8%)

reports the gender and age at the time of the posttests for all participants (open text boxes in which participants typed their responses). The majority of participants were female (73%, n = 38) and mostly young adults aged 18 to 19 years old in their first or second year of college (65%, n = 34).

Participants were asked where they learned Spanish (at home, at school, with family, etc.) and what their ethnicity was. Both questions were open text boxes where participants typed their responses.

Table 3. Ethnicity and location(s) of Spanish exposure.

	Where Did You Learn Spanish?			Ethnicity
	Home, with Family	School	Both Home and School	Mexican, Mexican-American, Mexican/American, Mexicana, Mexicano	Hispanic, Latina/o/x, Hispanic/Latino	Blanca, Caucasian, White	Other
Spring 2017 (n = 9)	3	2	4	4	3	1	1
Fall 2017 (n = 15)	7	2	6	6	7	2	0
Fall 2018 (n = 9)	3	3	3	4	1	3	1
Fall 2019 (n = 14)	6	6	2	4	4	5	1
Fall 2020 (n = 5)	4	0	1	0	5	0	0
Total (N = 52)	23 (44%)	13 (25%)	16 (31%)	18 (35%)	20 (38%)	11 (21%)	3 (6%)

below reports the student’s responses for ethnicity and where Spanish was learned, as reported on the posttest. Three-fourths of participants learned Spanish at home, and many of these students (31%, n = 16) reported that they learned Spanish both at home and took classes in middle school or high school: “I grew up speaking Spanish at home, but I also have taken Spanish classes in school to learn to speak it more formally.” A quarter of participants learned Spanish only in a school setting.

As of January 2022, there were approximately 80 Spanish dual language and immersion schools in Minnesota, ranging from preschools to high schools (MAIN 2022). Some of the participants in this study attended Spanish immersion programs as children; however, most of the participants reported attending English mainstream classrooms and ESL pull-out programs, and at the middle or high school levels, they were often placed into beginner-level Spanish courses which did not meet their instructional needs. For most participants, this course was their first encounter with curricula designed specifically for their instructional needs as heritage learners of Spanish.

Previous studies have found links between students’ reported ethnicities, proficiency levels, and attitudes toward Spanish (e.g., Hudgens Henderson et al. 2020). Although this study did not examine how participants’ ethnic self-perception may have changed over the course of the semester, the posttest answers are reported here with the assumption that after a semester of SHL content, students may have a more specific ethnic descriptor for themselves. If participants reported a specific country along with the labels “Hispanic” or “Latinx” (e.g., “Mexican Hispanic”), they were coded as the country-specific label. If participants supplied only “Hispanic” or “Latinx”, they were coded in that category. About three-fourths of the participants reported an ethnicity related to Mexico and/or Hispanic/Latinx, one-fifth self-reported as White, and three participants self-reported as another ethnic descriptor.

In the decade of the 2010s, Hispanics made up more than half of the population growth in the United States. The Upper Midwest states like Wisconsin, Michigan, and Ohio have seen population growth driven primarily by Hispanics, while North Dakota and South Dakota have seen the fastest growth in the Hispanic population since 2010 (Krogstad 2020). Minnesota, where the study took place, has also seen an increase in people of Latino descent, with Mexican Minnesotans making up the third-largest ethnicity group after Whites and African Americans (Gutierrez and Richmond 2021). In line with these region-wide trends, Mexico was the most frequent country-specific descriptor related to ethnic identity for these participants.

I was the researcher-instructor of the course. I am a White female nonnative speaker who speaks a Mexican-influenced variety of Spanish. While I do not consider myself a heritage speaker of Spanish, I have had a heritage-like experience growing up surrounded by Spanish as a child in Puerto Rico and New Mexico, with Spanish-speaking family members. As a college-age adult, I decided to become proficient in Spanish to communicate with my Mexican stepfamily, and I have often suffered from linguistic self-consciousness and insecurity even after achieving “advanced” proficiency. I acknowledge that my social identity as a native-English-speaking, nonHispanic White woman means that I embody the racial and linguistic power structures in the U.S., and the first week of the semester is often tense with the expectation that yet another outsider will tell them how to speak their language. To set the tone of the class, I share my language autobiography (Aparicio 1997) with my students to position myself not as a “corrector” of their Spanish but as an ally in the maintenance and reclamation of Spanish. While I have not personally experienced raciolinguistic prejudices that my students have faced, I share my own challenges with a bilingual identity in a language that is perceived as “not mine”. The idea of “belonging to a language” surfaces later in our discussions of language shift and Hispanic/Latinx identity.

2.3. Survey Instrument

To measure the change in CLA at the start and the end of the course, participants were given an electronic survey instrument consisting of 30–40 questions based on Shin and Hudgens Henderson (2017), which asked participants to agree or disagree with attitudinal statements on a 4-point Likert scale (strongly agree, agree, disagree, or strongly disagree), as well as True/False items that measured sociolinguistic knowledge. This analysis is based on 14 language attitude questions that were consistently delivered to students across all five semesters and were divided into themes of Prescriptivism, Linguistic Stereotypes, Monolingual and Monoglossic Ideologies, Linguistic Self-Awareness, and Language-based Prejudice (see

Table 4. Survey themes and items.

Key Thematic Area	Items
Prescriptivism	1. Formal language has more clarity than informal language. 5. Formal language is grammatically superior to informal language. 9. Some grammar rules/patterns are inherently superior to others. 11. The grammar we learn in school is the grammar we should use all the time.
Linguistic Stereotypes	6. The English spoken in England is more proper than the English spoken in the U.S. 12. The Spanish spoken in Spain is purer than the Spanish spoken in Mexico.
Monolingual and Monoglossic Ideologies	7. Some bilinguals don’t speak either of their languages well. 8. A person is not bilingual unless he/she speaks both languages fluently. 13. Spanglish does not follow any grammar rules or patterns.
Linguistic Self-Awareness	2. All dialects follow specific grammatical rules or patterns. 4. I have an accent when I speak my native language. 14. I speak a dialect of my native language.
Language-based Prejudice	3. It’s ok to think someone is unintelligent because of the way they talk. 10. Some people sound unintelligent because of how they talk.

). With each item, participants were presented with a text box in which they were asked to explain their answers.

The thematic area of Prescriptivism was chosen to measure whether students believe there is a “correct” way to speak and if that perception centers around “formal” language and the school context (cf. Milroy and Milroy 1999). The Linguistic Stereotypes items measure student perceptions of prestigious varieties in both their named languages. The Monolingual and Monoglossic Ideologies subset measures students’ perceptions of bilingual speakers, given the prevalence of the myth of the “balanced bilingual” as well as “semilingualism” (cf. Grosjean 1985; MacSwan 2000). The Linguistic Self-Awareness items measure students’ recognition of their own sociolinguistic positioning, and the Language-based Prejudice items measure students’ willingness to judge and accept the judgment of others based on speech.

Shin and Hudgens Henderson (2017) used these fourteen survey items (among others) for participants enrolled in an advanced Spanish grammar course, consisting of both Spanish as a second language students and native/heritage speakers of Spanish (of Mexican or New Mexican descent). The course in that study covered advanced grammar topics from a sociolinguistic perspective in which students examined authentic discourse, focused on grammatical concepts, and reviewed common language attitudes toward the feature under investigation. Over the course of a semester, the authors found a statistically significant change in sociolinguistic concepts (represented by items 2, 4, and 14 in the Linguistic Self-Awareness theme in this study) as well as language attitudes (represented by the remaining items of Prescriptivism, Linguistic Stereotypes, Monolingual and Monoglossic Ideologies, and Language-based Prejudice in this study). The native and heritage Spanish speaker participants in that study increased their appreciation of the varieties of Spanish as well as their own linguistic self-esteem, while the second language learners learned not to be judgmental towards nonstandardized varieties. The authors concluded that grammar instruction should incorporate sociolinguistic variation and challenge linguistic discrimination.

These survey items were used to measure the CLA of SHL students in a CLA-oriented course which was specifically designed to meet the instructional needs of SHL students. Similar attitudinal items were used in a survey instrument developed by Beaudrie et al. (2019) to measure CLA in SHL instructional contexts. Their survey items also alluded to the linguistic stereotype of the “purity” of peninsular Spanish (item 1 in their survey, People from Spain speak the purest form of Spanish), with a particular focus on attitudes related to monolingual and monoglossic ideologies surrounding bilingual linguistic practices (such as I don’t like it when people code-switch because it is not a proper way of speaking a language) and prejudices towards U.S. Spanish (such as I believe Spanish-speaking Hispanics in the U.S. don’t speak correct Spanish). Their survey was found to have high internal validity and reliability for measuring CLA, suggesting that these thematic areas are appropriate for CLA measurement in the SHL context.

2.4. Procedure and Analysis

The survey was administered via the online course platform D2L Brightspace in the first week of the semester (before hearing my language autobiography) and the 14th week of the semester. Students received course credit for completing the survey at both timepoints. The data consisted of the semester in which participants took the pretest and posttest, a pseudonym for matching purposes, and the quantitative and qualitative pretest and posttest responses to each item. Quantitative responses were forward-coded (strongly agree = 4, strongly disagree = 1) if they were items that participants with high CLA should agree with (items 2, 4, and 14). Items were reverse-coded (strongly agree = 1, strongly disagree = 4) if they were items that participants with high CLA should disagree with (items 1, 3, 5, 6, 7, 8, 9, 10, 11, 12, and 13).

An exploratory factor analysis was carried out to identify potential factors in the CLA survey using the pretest quantitative responses. A principal component analysis was performed to identify the possible number of initial factors. Subsequently, an exploratory factor analysis with principal axis was used to identify the factors. The principal axis factoring method was used along with a varimax rotation. After ascertaining the loadings of all items on the identified factors, items with loadings less than 0.4 in absolute value on any factor were dropped, and the exploratory factor analysis was resubmitted. The factors and loadings from the second exploratory factor analysis were then examined. From the second analysis, factors with too few or nonsensical loadings were dropped and items with loadings greater than 0.4 in absolute value on the remaining factors were retained. Cronbach’s alpha was then used to determine the interitem reliability of the retained factors and items. The items in each factor were then averaged to create factor scores. These scores were then compared (pretest-to-posttest) with paired t-tests. In addition, paired t-tests were used for each item individually to assess whether attitudes on the items changed pre-to-post semester. Bonferroni corrections were used to adjust these tests for multiple comparisons. All statistical analyses were conducted in R (R Core Team 2021). Factor analysis was carried out using the psych package (Revelle 2021), and joint tests for interactions between pretest/posttest by construct were carried out using the nlme (Pinheiro et al. 2021) and emmeans (Lenth 2021) packages.

3. Results

3.1. Factor Analysis

A scree plot of a PCA on all 14 items suggested that four factors sufficiently explained the variability in the 52 pretest survey responses. After performing an exploratory factor analysis with four factors, items 11 and 13 did not load higher than 0.4 on any factor in absolute value. A second exploratory factor analysis omitting those items with four factors resulted in item 4 with no absolute loadings higher than 0.4. One of the factors in the second analysis had high positive loadings of both items 2 and 10. As this pairing did not make sense conceptually, this factor was eliminated. The resulting factor analysis yielded three constructs consisting of 10 of the 14 items. The identified constructs and loadings are shown in

Table 5. Factor loadings of the 10 items on the three factors by exploratory factor analysis and interitem reliability of each factor as measured by Cronbach’s alpha.

Item	C1	C2	C3
1 * Formal language has more clarity than informal language.		0.54
3 * It’s ok to think someone is unintelligent because of the way they talk.			0.49
5 * Formal language is grammatically superior to informal language.		0.54
6 * The English spoken in England is more proper than the English spoken in the U.S.	0.74
7 * Some bilinguals don’t speak either of their languages well.			0.42
8 * A person is not bilingual unless he/she speaks both languages fluently.			0.58
9 * Some grammar rules/patterns are inherently superior to others.		0.4
10 * Some people sound unintelligent because of how they talk.	0.44
12 * The Spanish spoken in Spain is purer than the Spanish spoken in Mexico.	0.63
14 I speak a dialect of my native language.	−0.61
Cronbach’s alpha	0.72	0.52	0.44

* Reverse-coded.

Construct 1 included four variables, while constructs 2 and 3 each used three. Cronbach’s alpha for the three constructs was 0.72, 0.52, and 0.44, respectively, indicating high to moderate internal reliability. These results indicate that the majority of the survey items were effective for measuring CLA.

3.2. CLA before and after Instruction

All three constructs saw significant increases in CLA over the course of the semester.

Table 6. CLA factor analysis pretest to posttest.

	Pretest		Posttest		p Value
	Mean	SD	Mean	SD
Factor 1	2.7	0.5	2.9	0.5	0.003
Factor 2	2.3	0.5	2.6	0.6	0.012
Factor 3	2.9	0.5	3.0	0.4	0.003

contains the mean CLA score for each construct on the pretest and posttest. Paired-sample t-tests performed on the pretest and posttest scores showed significant changes for factor 1 [t (51) = 3.37, p = 0.003], factor 2 [t (51) = 3.03, p = 0.012], and factor 3 [t (51) = 3.61, p = 0.003]. Cohen’s effect size, indicating the relative size of the effect of an intervention, for factor 1 (d = 0.47), factor 2 (d = 0.42), and factor 3 (d = 0.50), suggest medium-sized effects. Similar to previous research (Beaudrie et al. 2019; Gasca Jiménez and Adrada-Rafael 2021), these participants had somewhat high levels of CLA in the pretest (an average of 2.61 out of 4 across the three factors). The statistically significant change on the posttest (an average of 2.86 out of 4 across the three factors) indicates that the CLA-infused curriculum was effective at increasing students’ CLA.

Out of 52 students over five semesters, 42 (81%) increased their CLA from pretest to posttest. This number is a similar proportion to Beaudrie et al. (2021), who found that 14 out of 17 (82%) participants increased their CLA. These findings suggest that approximately 20%, or one-fifth, of students do not respond to CLA instruction. Of the ten students who did not increase their scores in the current study, six students’ scores remained stable while the scores of the other four decreased. It is possible that these students need a differentiated approach that takes into consideration their life experiences with language ideologies; more research is needed into why some students do not respond to CLA instruction.

3.3. Individual Survey Items

Eleven out of fourteen items showed positive CLA growth from pretest to posttest (see

Table 7. Average CLA of 52 undergraduate students averaged across 5 semesters, by construct and for each individual item.

Item	Pretest		Posttest		Difference	p Value
	Mean	SD	Mean	SD
1. Formal language has more clarity than informal language	2.2	0.7	2.4	0.9	0.2	0.326
2. All dialects follow specific grammatical rules or patterns. **	2.6	0.6	2.8	0.7	0.2	0.23
3. It’s ok to think someone is unintelligent because of the way they talk.	3.4	0.6	3.4	0.7	0	1
4. I have an accent when I speak my native language. **	2.7	0.9	2.6	0.9	−0.1	1
5. Formal language is grammatically superior to informal language.	2.4	0.6	2.7	0.8	0.3	0.359
6. The English spoken in England is more proper than the English spoken in the U.S.	2.6	0.8	3.1	0.7	0.5	0.001
7. Some bilinguals don’t speak either of their languages well.	2.5	0.8	2.7	0.7	0.2	0.218
8. A person is not bilingual unless he/she speaks both languages fluently.	2.4	0.8	3	0.7	0.6	<0.001
9. Some grammar rules/patterns are inherently superior than others.	2.4	0.7	2.7	0.8	0.3	0.054
10. Some people sound unintelligent because of how they talk.	2.5	0.8	2.8	0.8	0.3	0.123
11. The grammar we learn in school is the grammar we should use all the time. **	2.7	0.7	2.9	0.6	0.2	0.014
12. The Spanish spoken in Spain is purer than the Spanish spoken in Mexico.	2.9	0.6	3.1	0.7	0.2	0.056
13. Spanglish does not follow any grammar rules or patterns. **	2.7	0.7	3.2	0.8	0.5	0.001
14. I speak a dialect of my native language.	2.9	0.7	2.8	0.8	−0.1	1

** Excluded from factor analysis.

). Among the specific items, items 6, 8, 11, and 13 showed statistically significant changes. It should be noted that items 11 and 13 were eliminated from the factor analysis reported above due to low factor loadings, despite being statistically significant on their own. One item showed no change (item 3, It’s ok to think someone is unintelligent because of how they talk), and the two items measuring Linguistic Self-Knowledge (4, I have an accent when I speak my native language and 14, I speak a dialect of my native language) both showed a slight negative change. These results indicate that the CLA curriculum implemented here did not increase participants’ linguistic self-awareness, in that they did not recognize that they themselves have “accents” and speak regional or social varieties.

4. Discussion

This study is a preliminary analysis of the measurement of CLA in a university-level SHL course. The participants comprised 52 students enrolled in an upper-division entry-level SHL course which taught writing skills within CLA-oriented thematic units. The results indicated that the course successfully increased participants’ CLA as measured along three factors related to the equality of language varieties, prescriptivism, and monolingual and monoglossic ideologies. Furthermore, three of the four survey items eliminated from factor analysis showed a positive change from pretest to posttest, including item 13, which relates to the patterned nature of Spanglish.

These findings suggest that CLA involves student perceptions of “correct” grammar, the communicative equality of all language varieties, and the evaluation of bilingual linguistic practices. Similarly, Beaudrie et al. (2019) found that awareness of language variation, language and power structures in society, and the intrinsic value of bilingual language practices were crucial to CLA development. The area that did not show growth in this study was participants’ linguistic self-awareness; after participating in a semester-long CLA-oriented course, many students still did not recognize that they themselves speak varieties of their languages which entailed having a particular pronunciation pattern. Future studies in CLA should investigate how instruction can include students’ self-reflection on their own linguistic identity and idiolects.

To be clear, the survey items reflected in this study did not measure behaviors connected to CLA, only attitudes. Behavioral study items may be more difficult to create and measure; in fact, the behavioral items in the CLA survey of Beaudrie et al. (2019) did not perform as well as the attitudinal items. The authors recommended that behavior be included since it is a critical aspect of agency in language maintenance, social justice, and advocacy. Future studies should consider how to measure the correlation between attitudinal change and behavioral change, especially for CLA courses that implement a service-learning component. Indeed, CLA-related behaviors may take on more relevancy if students recognize their own linguistic identities on a deeper level, beyond just being speakers of two languages.

Similar to the findings of Gasca Jiménez and Adrada-Rafael (2021), the participants in this study had somewhat-high levels of CLA prior to instruction. In their study, they noted that their participants demonstrated standard language ideologies and an avoidance of code-switching. The authors suggested that curricula for SHL students focus on these two areas of highest need. The results presented here corroborate that conclusion. However, the inclusion of qualitative explanations for survey responses allows for a deeper thematic analysis of student responses. Examining qualitative data is important to contextualize the agreement or disagreement with survey items and capture nuances that a Likert scale may obscure.

These nuanced belief systems were evident in students’ perceptions of “correct” language and who is considered a “true” bilingual. For example, items 1, 5, and 9 constituted the second factor and all dealt with the supposed superiority of formal language and grammar. The change from pretest to posttest on this factor was statistically significant, suggesting the CLA curriculum had an impact on students’ perceptions of the superiority of standardized varieties, which are preferred in school settings. However, some students maintained the stance that formal language was a type of neutral speech that everyone should learn. On her posttest, Juana (Fall 2018) described formal language as a type of lingua franca that allows communication with a broad range of people:

I agree [that formal language is grammatically superior to informal language] because, yes formal language does make it easier to talk to others, but at the same time, no everyone was taught the formal way of speaking Spanish, so you might need to learn the way Spanish is spoken where you are at.

The view that school-based language was a lingua franca between regional and social groups was also evident with item 11, The grammar we learn in school is the grammar we should use all the time, which is an explicitly prescriptivist viewpoint that views school-based language as superior. One participant, Sofia (Spring 2017), did not demonstrate growth in CLA with this statement but demonstrated a more nuanced approach on the posttest. At the beginning of the semester, she wrote “I agree because what we learn in school is by the textbook and is proven to be correct”. On her posttest, she continued to defend the superiority of school-based language, assuming that its correctness is why it is taught in schools, but acknowledged that school does not teach all grammar forms that might be acceptable:

I agree to an extent because the grammar we learn in school should be the most correct that would be why they teach it. Although sometimes not everything is taught in school. So if there is a grammar change that wasn’t taught in school, it’s okay to mix both grammar taught in school and also out of school.

It is important to point out here that Sofia trusted her school and her educators to teach her what was “correct”. It is precisely this trust that makes CLA so critical for all educators to implement; unfortunately, that trust is misplaced when educators continue to teach that one variety is superior to another or that the variety that has political and economic power is at the same time “neutral”.

Future CLA must make special efforts to not only establish the linguistic equality of all language varieties but also to take aim specifically at the view that standard varieties used in formal contexts are somehow neutral in terms of power imbalances. Speakers of minoritized varieties carry the burden of learning a standardized variety in school, while students who speak a variety more closely aligned to school-based language have no obligation to learn the variety of their peers (Siegel 2006). CLA instruction must push students to ask themselves: once we accept that haiga is just as linguistically valid as haya, why would we continue to proscribe haya in formal contexts, and who benefits from that? If there is no communicative breakdown with the use of haiga instead of haya, what is the true barrier to using haiga in formal contexts? CLA instruction must push students in different ways to ask themselves these questions that interrogate their assumptions.

Thus, despite the growth of Factor 2 and item 11 between the pretest and posttest indicating a higher level of CLA, some participants still maintained the view that formal, school-based language was somehow neutral, albeit not “superior”. Future research should consider how to problematize this view; it may be that inculcation in standard language ideologies is not easily disrupted by a single course incorporating CLA. To be effective, CLA should be embedded throughout a university program and at all levels of K-12 instruction (Beaudrie and Wilson 2022; Holguín Mendoza 2018; Hudgens Henderson and Hackman 2022).

In the same way that some students adapted their standard language ideologies to new information (i.e., the standard is not superior but still a neutral tool for advancement), some students also adapted their bilingual proficiency ideologies (i.e., many bilinguals are dominant in one language or another depending on the context, but they can’t be true bilinguals unless they can speak those languages well). Upon entering the course, students had somewhat positive attitudes towards bilingual linguistic practices. This pre-existing CLA was captured in item 13: Spanglish does not follow any rules or patterns. This item was eliminated from factor analysis but had a statistically significant change from the pretest to posttest on its own. On the pretest, the majority of students (n = 34, 65%) rejected the idea that Spanglish does not follow any rules or patterns. On the posttest, this proportion grew to 87% of respondents (n = 45). One student who showed growth in the understanding of Spanglish was Maite (Fall 2019). At the beginning of the semester, she did not see the grammatical or lexical patterns because it was just “a mixture”:

I said I agree that Spanglish does not follow any grammer rules or patterns because someone is just speaking a mixture of both Spanish and English at the same time. Spanglish is not a real language and due to that it does not follow any grammar rules or patterns.

On her posttest, she specifically named some of the grammatical and lexical patterns studied in class:

I said I strongly disagree with the fact that spanglish does not follow any grammar rules or patterns because spanglish does have many different forms and patterns such as prestamos lexicos, calcos, [extensiones] semanticos and more [.]

By the end of the semester, most participants recognized that Spanglish was not just random language mixing.

This positive perspective toward Spanglish did not mean that all students accepted code-switching speakers as “valid” bilinguals. Two items measured participants’ perceptions regarding the proficiency of bilinguals. Item 7, Some bilinguals don’t speak either of their languages well, tackled the myth of semilingualism, while item 8, A person is not bilingual unless he/she speaks both languages fluently, took on the myth of the balanced bilingual, or the idea that someone is not truly bilingual unless they are equally proficient in both languages. These two items loaded together in the factor analysis and demonstrated positive change from pretest to posttest. In their responses to these items at the start of the semester, most participants referenced language dominance and the ability to speak “properly” as possible explanations for why someone might not speak their languages well. Thus, most students both before and after the intervention were accepting of Spanglish, and many students changed their perspectives on how to define a bilingual. However, many students specifically defined a bilingual as someone who could speak both languages well, and some students continued to cling to this idea even at the end of the semester. For example, Augustina (Fall 2019) defined a bilingual as someone who could have “full conversations”:

In order to be bilingual, you need to know both languges [sic] very well. You aren’t considered bilingual if you aren’t able to have full conversations in both languages.

Although students accepted the idea that bilinguals are usually not balanced in both languages, some continued to demand that strong oral proficiency in both languages was a defining feature. Therefore, the curriculum was not completely successful in dislodging the view that “true” bilinguals are able to speak fluently in both languages. This CLA curriculum was successful at getting students to question their previously held beliefs, but not to the point where they were willing to abandon these beliefs entirely. That is, the pre-existing language ideologies morphed into a form that was perhaps less harsh but ultimately could still be just as harmful.

CLA instruction must take aim at the ideology of who is a “true” bilingual. Once Spanglish and code-switching are accepted as valid linguistic practices, and it is recognized that bilinguals may have a dominant language according to context, students must be interrogated to ask themselves how anyone can ever truly be a “balanced bilingual” (Grosjean 1985; Shin 2013). The rejection of this myth is a crucial step in the recovery of linguistic self-esteem for many heritage language learners; once they recognize that no bilingual is ever truly balanced, CLA instruction can connect this realization to the development of students’ awareness of their own accents and idiolects. This deeper discovery of their own linguistic identity can lead to a more nuanced perspective of themselves as bilinguals and speakers of their heritage language. Therefore, CLA must reach inward (to personal linguistic identity) at a deeper level to connect with the outward (to societal linguistic identities) on a deeper level, and instruction must continue to cycle the awareness of societal forces back towards the students’ own individual practices and identities.

Indeed, the findings reported here position differentiation in CLA instruction as the next area to operationalize after instruction and assessment. Just as language proficiency levels have been mapped and described, and instructional methods and assessments have been created for each proficiency level (cf. WIDA and ACTFL), there appear to be “CLA proficiency” levels that depend upon students’ prior experiences and current belief systems. Some students progress smoothly from one level of awareness to the next, while others appear to make progress but don’t quite achieve a deeper change. Identifying and describing these “levels” as well as establishing effective interventions or accommodations for students at emerging CLA proficiencies could be the next logical step to making sure that one-fifth of our students are not left with the same (or more harmful) language ideologies as before.

5. Conclusions

CLA instruction can take many different forms and include many different types of activities; researchers continue to investigate how to operationalize and assess CLA in coursework (e.g., Serafini 2022). This study contributes to the research of CLA in the SHL classroom by offering guidance in terms of thematic areas to study and measurement of the effectiveness of CLA instruction, as well as pointing out the need to interrogate language ideologies at a deeper, personal level among students. Survey instruments must include space for students to explain their answers, so teacher-researchers do not assume that change from a pretest to posttest means deep attitudinal change. This study found evidence that some students merely adapted their language ideologies to the new information and continued to hold beliefs that could have harmful consequences in society. For these students, one course in CLA was not quite enough, and indeed, different students will have different CLA needs. Along with operationalization and assessment, future directions for CLA instruction should include the identification of “CLA proficiency” levels and methods of differentiation for students who have deeply entrenched language ideologies.