What does your accent say about you? The perception of Cuban and Peninsular Spanish varieties by native and non-native speakers of Spanish

2.1 Cuban and Peninsular Spanish varieties

The Cuban and Peninsular Spanish varieties were chosen in part because they contain many phonological and phonetic differences (see Hualde 2014; Lipski 1994; Martínez-Celdrán et al. 2003). Additionally, previous studies (e. g. Hualde et al. 2012; Lipski 1994) have contextualized the prestige disparity that exists between Peninsular and Cuban Spanish, or the variety spoken in the Caribbean in general. However, where the gap in the literature arises is whether these notions of prestige and stigmatization translate across not only to native speakers, but also to individuals whose first language is not Spanish or who perhaps do not speak the language at all.

2.2 Theories on accent perception

2.3 The role of categorization

However, for these theories to describe the situation accurately, listeners need to have the ability to correctly discriminate between accents within and outside their membership. For instance, Díaz-Campos and Navarro-Galisteo (2009) examined the perceptual categorization of Peninsular, Chilean, Argentinian, Colombian, Costa Rican, and Mexican Spanish varieties among Peninsular and Venezuelan Spanish speakers. The authors contest that Spaniards and Venezuelans were not very accurate in the categorization of other Spanish dialects, but the Venezuelan participants performed better since they had more exposure to the dialects tested in comparison to the Spaniards. The study concludes that individuals who have regular contact and exposure to other dialects were better able to encode cues that help identify and categorize other speaker’s accents, compared to those with less exposure. This relates to the social identity theory in that individuals could correctly categorize others based on speech and ascertain their membership within that group. It is evident that these theories help to explain different aspects as to why individuals may perceive accents the way they do, but they have not been widely applied to judgements made on varieties of Spanish. While native and heritage speakers were often aware of phonological distinctions in Spanish varieties due to their level of familiarity with Spanish accents, L2 learners had a harder time grasping some of these differences. For instance, Schmidt (2018) examined L2 speakers on their ability to pick up on certain phonological markers such as the aspirated /s/ in Venezuelan, Caribbean, and Argentinian Spanish varieties. The author found that while native and heritage participants were able to classify this sound as an aspirated /s/, intermediate L2 learners frequently ignored this aspiration cue and classified the sound as a deletion. In contrast, Chappell and Kanwit (2022) found that advanced L2 Spanish learners with phonetics training were more likely to overgeneralize speakers with /s/ elimination as Caribbean. Therefore, varying levels of familiarity and proficiency in a given language could potentially affect how individuals perceive certain accents in comparison to native speakers.

2.4 Perceptions toward the Cuban variety

The body of research that examines attitudes toward different Spanish varieties is growing. For instance, Carter and Callesano (2018) investigated how young adults living in Miami-Dade County conceptualized Cuban, Colombian, and Peninsular accents based on socioeconomic class, income, and employment. They found that the Cuban and Colombian voices were consistently rated lower on most social factors when compared to a Peninsular Spanish voice from Madrid that distinguished between /s/ and /θ/. For example, in terms of income, the Peninsular voice was always assigned to a higher income bracket in comparison to the Cuban and Colombian voices (Carter and Callesano 2018). Partially, this was explained by the fact that speakers of a Caribbean variety are often perceived negatively or as uneducated. This was further demonstrated by a follow-up study by Callesano and Carter (2019) where the participants perceived Peninsular speakers significantly higher in terms of traits associated with competence (e. g. intelligibility, confidence, etc.) when compared to Cuban and Colombian speakers. This finding is surprising given the number of participants in this study that were of a Cuban or Colombian background, contrasting the tenants posed by the social identity theory and own-accent bias. This negative perception toward the Cuban variety can in part be attributed to the high prevalence of non-standard speech found in this region.

Interestingly, the Cuban variety may have not always had this negative connotation attached to it. In a study by Alfaraz (2002), Cubans living in Miami rated a variety of different Spanish accents, such as Puerto Rican, Dominican, and Peninsular, among others. Also included in this study was the variety of Spanish spoken in Cuba, which was divided into Cuba 1 and Cuba 2 to highlight the variety of Spanish spoken before (Cuba 1) and after (Cuba 2) 1959, which is the mark of the Cuban Revolution. As the author expected, the Peninsular variety was attributed the highest ratings, but when the two Cuban varieties were compared to each other, the pre-1959 variety was rated higher on all measures such as correctness and pleasantness (Alfaraz 2002). When asked, the participants stated that the variety of Spanish spoken in Cuba currently is ‘horrible’ and ‘sloppy,’ while others attributed the linguistic impoverishment of the post-1959 variety due to the moral and political stance of the country (Alfaraz 2002). Additionally, Díaz-Campos and Killam (2012) propose that the speech used in educational settings can oftentimes represent a form of standard speech in which anything that deviates is seen as ungrammatical. Caribbean speech is often viewed with this lens, as it contains many forms of deviated speech, such as /s/ aspiration and deletion, lateralization, and deletion of /d/, among other ‘non-standard’ speech forms.

2.5 Perceptions toward the Peninsular variety

A great majority of studies dealing with attitudinal perceptions among Spanish accents have dealt with the Peninsular variety in some way. Spain is a very linguistically diverse country, meaning that while many languages are spoken aside from Spanish, there are also many different regional accents that have been analyzed in prior studies to determine whether stereotypes occur at the regional level. As previously mentioned, Carter and Callesano (2018) found that the Peninsular accent was rated higher on most social factors such as occupation, income, and level of educational attainment. These results were consistent among all participant groups, including those that did not speak Spanish within Miami-Dade (Carter and Callesano 2018). In terms of stereotypes within Spain itself, Gallego and Rodríguez (2012) examined the attitudes towards 10 regional varieties of Spanish in Spain (e. g. Madrid, Canarian, Catalan, Andalusian, etc.) and found that Castilians preferred their own variety of Spanish when compared to other regional Spain varieties, but they exhibited no predominant negative associations toward other regional accents found in Spain. The most notable differences were found in terms of age, where both older and younger, but not middle-aged, participants demonstrated a slight dislike towards the Catalan Spanish variety (Gallego and Rodríguez 2012). For the most part, however, the Peninsular variety tends to be one of the most favourably rated in prior research. In sum, while there is regional variability in the accents found within Cuba, the differences are not as pronounced as in Spain. Furthermore, although all the accents found within Cuba, as is the case with Caribbean Spanish in general, tend to be stigmatized all together, in the case of Spain, it is primarily the Andalusian or Canarian Spanish that are stigmatized (see Hualde 2014; Lipski 1994; Martínez-Celdrán et al. 2003).

2.6 Additional factors involved in modulating accent perceptions

Besides the general trend that can be seen in Sections 2.2 and 2.3, with the Peninsular variety being perceived as the most prestigious and the Cuban accent with the most negative association, several other studies have examined a host of additional factors that could have a role in modulating accent perceptions. For instance, Winke et al. (2013) found that the background of the raters themselves can pose an effect as some individuals may favour certain accents over others based on which accent sounds most familiar to them. This is supported by Stotts (2014), who compared attitudes toward Peninsular and Mexican Spanish among individuals living in either Spain or Mexico who were learning Spanish as a second language. Stotts noted that speakers attributed a higher rating to the accent they could understand the easiest, but there was also a general tendency to rate the Peninsular accent higher, even among the participants that were most familiarized with Mexican Spanish. Additionally, in a study examining Mexican Spanish and English, the Mexican Spanish accent was found to be rated lower on a variety of social and linguistic measures, especially by female participants (Alvord and Thompson 2020). As Aguilar (2018) explains, females tend to attribute higher ratings to ‘prestigious’ and more mainstream languages and their subsequent varieties, as shown in the previous examples with Peninsular Spanish and English. These studies have examined Spanish speakers’ perceptions of Peninsular Spanish, in particular, as well as factors outside of nationality and familiarity. A small amount of research has examined dialect perceptions by speakers of other languages, which will be discussed in the next sub-section.

2.7 Non-native perceptions on dialect variation

Non-native perceptions are an important consideration stemming from the ideas presented in the previous section. Specifically, L2 speakers and individuals with no knowledge of the perceived language can make dialectal judgments. When accent perception studies do include non-native speakers of a language, non-native individuals are often the speakers as opposed to the listeners, as is the case with foreign accent perception studies (e. g. Fuse et al. 2018). While accent perception research is abundant among varieties of English and the perceptions of native speakers, few studies have considered the perceptions of non-native speakers of another language. For example, Trimble (2014) demonstrated that intermediate-advanced English learners of Spanish rated varieties with an abundance of consonant deletion (e. g. Caribbean and Rioplatense) as more difficult to understand and subsequently less intelligible compared to the Spanish spoken in Mexico and the highlands of Colombia. Schmidt and Geeslin (2022) found similar perceptions attributed to the Puerto Rican variety examined in their study. The authors explained that the negative perceptions attributed to stigmatized varieties may be due to a combination of factors characterizing the L2 speaker, such as linguistic repertoire, exposure to a single variety in the classroom, etc. Notwithstanding, more studies including non-native speakers are needed to understand how these perceptions develop in the acquisition of Spanish.

2.8 Testing techniques and protocols in previous research

A great deal of previous research (e. g. Carter and Callesano 2018; Fernández-Mallat and Carey 2017) has used what is known as a matched guise technique to analyze unconscious attitudes toward a given accent. The matched guise technique, developed in 1960 by Lambert, Hodgson, Gardner and Fillenbaum, involves presenting listeners with auditory samples of a language or language variety in the form of a recorded passage. Typically, directly after playing a speech sample, individuals are asked to rate the speaker on a variety of measures (e. g. intelligence) to determine their unconscious biases toward the voice. One important note about this method is that it is typically used to measure nonlinguists’ perceptions toward a given language or language variety (Preston 2011). This method allows for the evaluation of an individual’s unconscious perceptions since simply asking a person what their outward attitudes are towards an accent may yield in their conscious attitudes, stereotypes, and individuals may not be inclined to share how they truly perceive the accent (Lambert et al. 1960). Therefore, since the individual does not know the identity of who they are listening to, the researcher can attempt to isolate the voice and evaluate perceptions much more clearly. In terms of recorded speech, samples in prior studies tended to range anywhere from 25 s to 2 min, with most hovering in the 30 s range (Carter and Callesano 2018; Fuertes and Gelso 2000). After presenting the participants with the recordings of each accent, this is commonly followed by either the administration of a survey, conducting an interview, or both to gauge perceptions toward each language or accent.

Based on previous literature, it is evident that while there is a substantial growing body of research in this area, much is still unknown regarding the factors modulating accent perceptions. While most studies have focused on the Peninsular variety, very little is known about perceptions toward Cuban accents with the latter being examined primarily within the context of Miami. Furthermore, studies that have examined accent perceptions from a comparative standpoint are scarce. Thus, this study will contribute to the existing gap in the literature not only by having native speakers rate two varieties that are found on opposite ends of the prestige spectrum, but also by including the perceptions of non-native speakers to gain further insight into how these groups formulate accent perceptions in languages that are not indicative of their mother tongue.

4.1 Participants

The present study included a sample size of 50 adult listeners, 68 % of which were female, who rated five Cuban (Havana) and five Peninsular (Madrid) female voices. This total participant pool was composed of 13 native Cuban Spanish speakers, 10 native Peninsular Spanish speakers, 13 L2 Spanish learners, and 14 monolingual English speakers, who were all recruited using the snowball sampling technique. The native Spanish speakers did not report speaking any other languages aside from Spanish, except for a few reporting low-to-intermediate proficiency in English. Both native listener groups were living in distinct parts of Spain, Canada, or the United States at the time of task completion. As for the L2 Spanish learners, this group reported speaking English as their L1, learning Spanish after the age of 16, and had studied Spanish until at least an intermediate/advanced level through a formal institution, study abroad/exchange program, or a combination of the two. The subsample of L2 Spanish learners in this study received exposure to distinct varieties of Spanish depending on where their instructor was from or if they participated in a study abroad experience offered through their institution. Study abroad trips tended to take place in Cuba, Spain, or Mexico. As for their instructors, several came from Cuba or Spain, but many also come from other countries, with a high proportion from Mexico and Colombia. Therefore, not all L2 Spanish learners would readily have access to Cuban or Peninsular input through their instructor, as their instructors may have been from other countries.

The monolingual English groups did not report prior knowledge of any other languages besides English. However, it is important to mention that in Ontario, French is mandatory from grade 4 until grade 9. Therefore, the latter two groups mentioned, completed their schooling in Ontario and were required to take French until the mandatory level. Individuals who had participated in an immersive French program before grade 4 or took French beyond the mandatory grade 9 level were excluded from participating in the study. Participants in the English monolingual group were still considered to be monolingual speakers since they reported that they did not recall any knowledge of French from their previous schooling and did not use the language at all in their day-to-day life. Furthermore, this group served as a control group for the study to assess whether other aspects of the speakers’ voice (e. g. pleasantness of the voice) could have skewed the results. Since this group did not know the content of the recordings, they were used as a baseline of comparison for the other groups. Some of the listeners in this group did report prior travel to various Spanish-speaking countries such as the Dominican Republic, Cuba, Mexico, and Spain. This included approximately five out of the 14 monolingual English speakers tested, for a percentage around 36 %.

As mentioned previously, the speakers consisted of five voices each from Cuba (Havana) and Spain (Madrid) Spanish, for a total of 10. All speakers were female to avoid introducing a gender bias in the listeners’ ratings. As is evident from the literature review section, there can be much accent variability within a Spanish-speaking country itself. Hence, we chose to recruit participants from Cuba and Spain that came from Havana and Madrid respectively, as these cities contain the phonological markers that are characteristic of these accents, while also limiting accent bias toward other regions within the same country. All speakers were residing in their home country, had no knowledge of other languages besides Spanish, and had never lived in another country or region within their home country. When asked informally prior to completing the recording task (see Subsection 4.2.1), all speakers reported having a university or college degree or in the process of obtaining one. Their occupations mainly consisted of office jobs, schoolteachers, housewives, and technical support. As for socioeconomic status, the speakers from Madrid all ranged around a middle-level income. For the speakers from Havana, they tended to range in a lower income bracket when compared to the Madrid speakers, due to base salaries for distinct occupations in Cuba. While in most North American and European societies there tends to be a stark difference between different occupations (e. g. the annual salary of a professor vs. that of a cashier), this distinction does not exist to this extent in Cuba, thereby explaining the socioeconomic differences observed in our speaker group.

More than one speaker from each variety was included to provide the participants with a more holistic representation of what each accent typically sounds like. Additionally, since many voices from the same geographic location can vary on individual measures such as pitch, nasality, or other defining markers, including more than one speaker from each of the varieties under study ensured that listeners rated the voices on the accent itself and not on features in the individual speakers’ voices. All participants were over the age of 18 and instructions for the tasks were provided to the participants in English or Spanish depending on the language they felt most comfortable with.

4.2 Tasks

4.3 Stimuli

All the speakers read the same prompt to ensure consistency in content between recordings (see Appendix B). Each speaker was recorded a total of three times reading the prompt. These recordings were then transferred to Audacity where the clearest sound file free of long pauses was chosen. All the sound files were cut to a similar length to maintain uniformity and consistency. The recordings ranged from 18 s to 31 s as some speakers naturally spoke more quickly than others. The stimuli used in this study were adopted from the experimental procedures found in Carter and Callesano (2018), Carter and Lynch (2014), and Callesano and Carter (2019). Similarly to these studies, these prompts were used to divert the attention away from language perceptions and focus attention on an arbitrary topic to avoid revealing the real purpose of the study to the speakers. Additionally, the passage contained instances of words containing the following letters: <c>, <z>, <s>, <d>, and <r>. These letters reflect sound segments that tend to be produced distinctly among speakers of the two varieties being examined.

4.4 Data analysis

To address the research questions in this study, nonparametric statistical analyses were used. Specifically, a series of Fisher’s exact tests were used to examine the group differences on multicategory nominal dependent variables and Mann–Whitney U tests were used to compare the groups on ordinal dependent variables. In addition to these tests, a series of chi-square tests of independence were also conducted to determine significance level.

5.1 Categorization of Cuban and Peninsular Spanish varieties

Our first hypothesis predicted that native Cuban and Peninsular Spanish speakers would outperform L2 Spanish learners and monolingual English speakers in correctly categorizing the Cuban and Peninsular varieties (Schmidt 2018). To investigate the rate of categorization, chi-square tests of independence and Fisher’s exact tests were performed to compare the percentage of speakers in each group that were able to correctly identify which of the voices were Cuban and which were Peninsular. Both tests yielded significant results, χ ² (3) = 78.38, p < 0.001 for the Cuban prompts, and χ ² (3) = 133.06, p < 0.001 for the Peninsular prompts. The effect size of the group differences as measured by Cramer’s V was large for both the Cuban (0.56) and Peninsular (0.73) speech prompts. The rates of categorization for both Cuban and Peninsular voices among the four listener groups is displayed in Figure 1.

Figure 1:

Categorization of Cuban and Peninsular Spanish varieties by all four listener groups.

Not surprisingly, the native Cuban group was able to identify Cuban speakers better in comparison to the other three groups. However, the percentage of native Peninsular speakers who correctly recognized the Cuban Spanish variety was not significantly different from the monolingual English group. The lowest and significantly different group compared to the other three listener groups was the rate of recognition of the Cuban Spanish variety in the monolingual English group.

As for the Peninsular Spanish variety, both groups of native speakers recognized it at a significantly higher rate compared to the other two groups. Additionally, the L2 Spanish learner group recognized this variety at a higher rate compared to monolingual English speakers.

5.2 Perceptions of all listener groups towards Cuban and Peninsular Spanish varieties

5.2.1 Perceptions regarding socioeconomic status (SES) attributes

Based on the accent prestige theory (Fuertes et al. 2002) in addition to results from Callesano and Carter (2019), Carter and Callesano (2018), Carter and Lynch (2014), Stotts (2014), and Winke et al. (2013), our second hypothesis predicted that native Cuban and Peninsular Spanish speakers’, L2 Spanish learners’, and monolingual English speakers’ perception of Cuban and Peninsular Spanish varieties would differ from each other. This hypothesis was tri pronged. Specifically, the first part of the hypothesis predicted that whereas native Cuban and Peninsular Spanish speakers would attribute higher levels of education, income, and employment to their own varieties, L2 Spanish learners and monolingual English speakers would attribute higher levels of these factors to the Peninsular variety (H2. [a]).

The subsample of native Cuban and Peninsular Spanish listeners was used to explore whether these groups attribute their own varieties with higher levels of education, income, and employment (Figure 2). A series of chi-square tests of independence were conducted to address this question. The results of these tests and effect size measures (Cramer’s V) are reported in Table 1.

Figure 2:

Perceived SES attributes of Cuban and Peninsular Spanish varieties among native Cuban and Peninsular listener groups, divided per variable.

Table 1:

Perceived SES attributes of Cuban and Peninsular Spanish varieties among native Cuban and Peninsular listener groups.

SES attribute	Native Cuban Spanish speakers		Native Peninsular Spanish speakers
	χ2-test (p-value)	Cramer’s V	χ2-test (p-value)	Cramer’s V
Education	3.32 (0.200)	0.16	19.32 (<0.001)	0.44
Income	8.71 (0.006)	0.26	5.07 (0.054)	0.23
Job occupation	12.47 (0.013)	0.31	24.63 (<0.001)	0.50

As can be seen from Table 1, in the native Cuban group, there were significant differences (at a level of 0.05) in the perception of income and job occupation between Cuban and Peninsular speakers. However, the effect size of those differences, as measured by Cramer’s V, was small. Contrary to Hypothesis 2(a), Cuban speakers attributed lower levels of income to their own variety of Spanish (Figure 2). In addition, they attributed the CEO job occupation less often and the dancer occupation more often to their own variety of Spanish. There was no difference in attribution of educational level to the two Spanish varieties among this group of listeners.

However, in the Peninsular Spanish group, Hypothesis 2(a) was partially confirmed. Specifically, Peninsular speakers were perceived as having higher levels of education and were attributed the CEO job occupation more frequently and cashier and coffee shop jobs less frequently compared to Cuban speakers. The effect size of these differences, as measured by Cramer’s V, was moderately strong. No significant differences in perception of income were found for this group of listeners.

The subsample of L2 Spanish learners and monolingual English listeners was used to explore whether these population groups attribute higher levels of education, income, and employment to the Peninsular variety in comparison to the Cuban variety (Figure 3). A series of Fisher’s tests was conducted to address this question. The results of these tests along with effect size measures (Cramer’s V) are reported in Table 2.

Figure 3:

Perceived SES attributes of Cuban and Peninsular Spanish varieties among L2 Spanish learner and monolingual English speaker listener groups, divided per variable.

Table 2:

Perceived SES attributes of Cuban and Peninsular Spanish varieties among L2 Spanish learners and monolingual English speaker groups.

SES attribute	L2 Spanish learners		Monolingual English speakers
	χ2-test (p-value)	Cramer’s V	χ2-test (p-value)	Cramer’s V
Education	6.16 (0.058)	0.22	8.60 (0.013)	0.25
Income	3.71 (0.175)	0.17	7.54 (0.022)	0.23
Job occupation	21.02 (<0.001)	0.40	9.60 (0.046)	0.26

As can be seen from Table 2, in the L2 Spanish learner group, significant differences were only found at the level of perceived job occupation for Cuban and Peninsular speakers. More specifically, a significantly higher proportion of Peninsular speakers were perceived as having a CEO job in comparison to Cuban speakers for the same variable.

However, in the monolingual English group, significant differences were observed in the perception of all three socioeconomic factors: educational level, income, and job occupation. Specifically, Peninsular speakers were perceived as having higher levels of education and income in comparison to Peninsular speakers. At the level of occupation, Cuban speakers were perceived as having a cashier job occupation more often than Peninsular speakers.

5.2.2 Perceptions regarding personality attributes

The second part of our second hypothesis (H2. [b]) predicted that whereas native Spanish speakers will attribute more positive associations with speech characteristics (i. e. likeable, intelligible, pleasant, friendly, correct, attractive, warm, funny, trustworthy, and confident) to their own varieties, L2 Spanish learners and monolingual English speakers will rate the Peninsular variety more positively on these measures.

A series of Mann–Whitney U tests was conducted with the subsample of native Cuban and Peninsular speakers to investigate whether they have more positive associations on speech characteristics to their own varieties of Spanish. The results of significant tests are reported in Table 3. Mean ranks for comparison of the two groups of speakers on each attribute are presented in Figure 4. Nonsignificant differences are indicated in grey.

Table 3:

A comparison of perceived speech attributes for Cuban and Peninsular varieties of Spanish among native Cuban and native Peninsular listener groups.

Speech attribute	Native Cuban	Native Peninsular
	Z (p-value)	Z (p-value)
Confident		−3.26 (0.001)
Correct	−2.38 (0.018)	−2.96 (0.003)
Friendly	−2.71 (0.007)
Likeable	−2.52 (0.012)
Pleasant	−2.29 (0.022)
Trustworthy	−2.70 (0.007)	−2.55 (0.011)
Warm	−2.33 (0.020)

Figure 4:

A comparison of perceived speech attributes for Cuban and Peninsular varieties of Spanish among native Cuban and native Peninsular listener groups, divided per attribute.

Hypothesis 2(b) was supported to some extent in both groups of speakers. Specifically, native Cuban listeners rated the Cuban Spanish variety as more correct, friendly, likeable, pleasant, trustworthy, and warm. In contrast, native Peninsular listeners ranked their own variety higher on confidence, correctness, and trustworthiness.

The subsample of L2 Spanish learners and monolingual English listeners was used to explore whether these population groups attribute the Peninsular variety more positive speech characteristics (i. e. likeable, intelligible, pleasant, friendly, correct, attractive, warm, funny, trustworthy, and confident). A series of Mann–Whitney U tests was conducted to address this question. The results of significant tests are reported in Table 4. The mean ranks for each attribute in the two groups are displayed in Figure 5. Nonsignificant differences are indicated in grey shade.

Table 4:

A comparison of perceived speech attributes for Cuban and Peninsular varieties of Spanish among L2 Spanish learner and monolingual English speaker listener groups.

Speech attribute	L2 Spanish learners	English monolingual speakers
	Z (p-value)	Z (p-value)
Confident		−2.35 (0.019)
Friendly	−2.42 (0.016)
Intelligible	−2.05 (0.041)	−2.66 (0.008)
Likeable	−2.87 (0.004)
Pleasant	−2.18 (0.029)
Trustworthy	−1.46 (0.144)
Warm	−2.89 (0.004)

Figure 5:

A comparison of perceived speech attributes for Cuban and Peninsular varieties of Spanish among L2 Spanish learner and monolingual English speaker listener groups.

The results in Table 4 indicate that the L2 Spanish learner listener group ranked Peninsular Spanish as more intelligible compared to Cuban Spanish (Figure 5). However, they ranked Cuban Spanish significantly higher compared to Peninsular Spanish on measures of friendliness, likeability, pleasantness, and warmth. The English monolingual listener group ranked Cuban Spanish significantly lower on confidence and intelligibility.

6.1 Categorization of Spanish varieties

Our results generally confirm Hypothesis 1 since the two groups of native Spanish speakers (i. e. native Spanish speakers from Cuba and Spain) did outperform the L2 Spanish learners and monolingual English speakers in correctly categorizing the Cuban and Peninsular varieties as distinct with a higher rate of accuracy. However, categorization largely depended on the speaker that was heard, proving to be a difficult task even for native Spanish speakers. This is especially true for the Peninsular group in this study who correctly categorized the Peninsular voices with high levels of accuracy but not the Cuban voices. These findings corroborate the results from Díaz-Campos and Navarro-Galisteo (2009), who found that native speakers of a language sometimes have difficulty pinpointing a speaker’s origin. The authors also found that Peninsular speakers had the greatest amount of difficulty stating where speakers from other varieties of Latin American Spanish were from, similarly to the current study. One explanation for the difficulty of this group and the L2 Spanish learner group in identifying distinct Spanish voices is a lack of exposure to different varieties from Latin America, compounded by the fact that many Latin American dialects exhibit similar phonological and phonetic variants.

6.2 Perceptions of socioeconomic status (SES)

We predicted that native Spanish speakers, L2 Spanish learners, and monolingual English speakers would differ in their perceptions of Cuban and Peninsular varieties of Spanish. Specifically, H2. (a) predicted that whereas native Spanish speakers (Cuban and Peninsular Spanish) would attribute higher levels of education, income, and employment to their own varieties, L2 Spanish learners and monolingual English speakers would attribute higher levels of these factors to the Peninsular variety. The results generally supported that these groups differed from each other in their perceptions of Cuban and Peninsular Spanish, but only on certain measures (e. g. some speech characteristics, which will be further discussed in Section 6.3). With respect to the first part of the second hypothesis, the results partially support H2. (a), as we found that the native Peninsular listeners rated their own variety higher on level of educational attainment and associated Peninsular speakers as belonging to more prestigious job occupations (e. g. “CEO of a company”). This is consistent with the tenets of the accent prestige theory, social identity theory, and own-accent bias, which state that individuals who recognize someone as belonging to the same language variety as they belong will rate them higher on certain social and personal factors. However, these higher ratings did not transcend across all SES attributes, as there were no significant differences found at the level of income for this listener group.

Contrary to these theories is the performance observed by the native Cuban listeners. This group rated their own varieties lower on the SES attributes analyzed. In some instances, this group rated the Peninsular variety higher, as was the case at the level of occupation. There are a few explanations that could account for these specific results. First, the speakers in this study were individuals who were born and raised in Havana, Cuba, did not speak any other languages, and had not lived anywhere else at the time of testing. However, the corresponding listener group were individuals who were born in diverse regions of Cuba and were living elsewhere for a period longer than five years. Due to the current political landscape of Cuba, many individuals who immigrate to other countries may disassociate and not identify as strongly with their Cuban culture, which could result in the unconscious lower ratings found here. Secondly, many individuals who speak Cuban Spanish are aware of the stigmas of their variety which may have translated to their subsequent lower ratings. Lastly, most of the Cuban listeners in this study were well aware of the economic situation in Cuba. By this we mean that even for jobs that would be paid higher wages (e. g. doctor, lawyer, etc.) in countries such as Canada and the United States, they would not earn such high wages in Cuba. Many Cubans in general are aware of these economic disparities and know that a higher education may not translate into a higher income, helping to explain the results obtained for this variable, specifically. In sum, a combination of the factors mentioned, in conjunction to biases existing at the regional level within Cuba, could help to explain the low SES attribution results obtained for the Cuban variety on behalf of speakers belonging to this same variety.

Regarding the L2 Spanish learner and monolingual English groups, we predicted that these individuals would rate the Peninsular variety higher on measures related to SES. The results obtained partially confirm this claim since the monolingual English group did rate speakers from the Peninsular variety as more likely to have a CEO position in addition to belonging to a higher income and education bracket. However, significant differences were found only at the level of job occupation for the L2 Spanish learner group, who rated the Cuban and Peninsular Spanish varieties more often with ‘cashier’ and ‘CEO’ positions, respectively. We see that the two non-native groups pattern more similarly to each other than to the native listener groups. This demonstrates that among groups with lower levels of exposure to Spanish, the prestige factor associated with Peninsular Spanish transcends proficiency. One reason why we might be seeing these trends, particularly among the L2 Spanish learner group, is due to the amount of unconscious exposure these learners receive toward Peninsular Spanish not only through academic means, but also through the heightened popularity of TV shows and movies available on streaming platforms in comparison to what is currently available on these viewing methods for Cuban Spanish.

One important point to mention is that while the Peninsular variety was rated higher on certain measures of SES across all four groups, there were hardly ever any significant findings at the level of income, with the exception of the monolingual English group. As a reminder, the voices in the present study were female to control for the variable of gender which could have potentially influenced the perceived income level attributed by the listeners. This is supported by previous accent perception studies such as Carter and Callesano (2018), who included male voices and found that their listener group tended to rate most of the speakers as pertaining to an upper-middle level income bracket or higher. However, the potential role of gender in relation to perceived SES is outside the scope of this study and would add another layer of complexity that goes beyond judgements and biases towards speakers from a specific language variety.

6.3 Perceptions regarding personality attributes and difficulty

It was predicted that whereas native Spanish speakers would attribute more positive associations with speech characteristics (i. e. likeable, intelligible, pleasant, friendly, correct, attractive, warm, funny, trustworthy, and confident) to their own varieties, L2 Spanish learners and monolingual English speakers would rate the Peninsular variety more positively on these measures (H2. [b]). In general, the results partially supported this hypothesis as the Peninsular Spanish listeners did rate the Peninsular voices higher on certain measures (i. e. attractiveness, confidence, correctness, and trustworthiness), but the same was not true for the native Cuban listeners as they did not rate the Cuban voices higher on any of the personality traits. A similar explanation as provided in the section prior could motivate why we see similar results with this group. As for the L2 Spanish learners and monolingual English listeners, we can also only say that the results partially supported the claim that these groups would rate the Peninsular variety higher on measures of personality. While the L2 Spanish learners did in fact rate the Peninsular variety higher than the Cuban variety on some measures such as intelligibility, this group also ranked the Cuban variety higher on traits dealing with friendliness and likeability, among others. Regarding the monolingual English group, significant results were found only at the level of confidence and intelligibility, in that higher confidence and intelligibility were associated with the Peninsular variety. This finding highlights the fact that, even for listeners who do not speak Spanish, underlying characteristics of the Peninsular accent may result in higher ratings and, perhaps, the stigma attached to the Cuban variety may be transferred.

Difficulty was tested last. We had predicted that native Spanish speakers would find their own varieties less difficult to understand (H2. [c]). We found that the results supported our initial claim. This is illustrated by the native Cuban and Peninsular listeners rating the opposite set of speakers as harder to understand. Regarding the other two groups of listeners (i. e. L2 Spanish learners and monolingual English speakers), we originally predicted that these groups would find the Peninsular variety easier to understand, following with previous studies such as Stotts (2014). However, our results reject this initial hypothesis as there were no significant findings for any of the groups in the perceived difficulty of each variety. One explanation for this is that all of the speakers in our study were female. While the speakers did produce speech sounds that are characteristic of both Spain (e. g. distinction between /s/ and /θ/) and Cuba (e. g. /s/ aspiration, deletion of /d/, lateralization of /r/ to [l], etc.), these occur with a much lower frequency among females than among males in the case of Cuba, since the productions of /s/, /d/, and /r/ in Cuba are considered to be stigmatized. Studies such as Díaz-Campos and Killam (2012) and Aguilar (2018) contextualize that female speakers tend to use more “prestigious” language forms and may be more aware of these stigmatized language forms when speaking, thereby making a conscious effort to minimize the use of these ‘nonstandard’ speech forms. This could explain why we did not find significant results for either of these groups regarding difficulty of understanding.