Perception of English Vowel Sounds by Spanish Speakers

FACULTAD DE FILOLOGÍA GRADO EN ESTUDIOS INGLESES TRABAJO DE FIN DE GRADO Curso académico 2024/2025 Perception of English Vowel Sounds by Spanish Speakers Alumno: Mónica Fernández Grandes Tutor: Darío Barrera Pardo Contents 0. Abstract ....................................................................................................................... 1 1. Introduction ................................................................................................................ 1 1.1 Previous studies ..................................................................................................... 2 1.2 Research Questions and Hypothesis ................................................................... 4 2. Methodology ................................................................................................................ 7 2.1 Selection of minimal pairs .................................................................................... 8 2.2 Questionnaire and lab recordings ..................................................................... 14 2.3 Participants and native speakers....................................................................... 15 2.4 Data analysis ....................................................................................................... 16 3. Results ........................................................................................................................ 17 3.1 Interpretation of results ..................................................................................... 19 3.2 Discussion (Comparison with hypotheses and previous studies) ................... 21 4. Conclusion ................................................................................................................. 26 References ...................................................................................................................... 27 Appendix ....................................................................................................................... 28 1 0. Abstract This Study investigates how native Spanish speakers perceive English vowel sounds, focusing on contrasts that do not exist in Spanish, such as /iː/–/ɪ/, /ʊ/–/uː/, and /æ/–/ʌ/. The main aim is to identify which phonemes are most difficult to distinguish and how factors like proficiency level, age of acquisition, years of formal instruction, and participation in Erasmus programs influence perception. Twelve Spanish-speaking participants with varying levels of English (B1 to C2) completed a minimal pairs listening test featuring audio recordings by a native English speaker. Each participant selected the word they believed they heard from a pair of options. Results were examined through the number of correct responses and the types of recurring errors. The findings show that /ʊ/– /uː/ and /iː/–/ɪ/ were the most challenging contrasts, supporting previous research on second language phoneme perception. Higher proficiency levels generally led to better results, but other factors also played a role. Participants with early exposure to English or real-life experience abroad performed more accurately, while the number of years spent in language school did not always correlate with improved perception. These results highlight the importance of both linguistic and experiential factors in the development of phonological awareness. They also suggest that pronunciation teaching for Spanish learners of English should go beyond classroom instruction and include more exposure to authentic input and focused listening practice. Key words: English vowels, accuracy, perception, proficiency, minimal pair, exposure. 1. Introduction Learning a second language (L2) involves more than acquiring grammatical and lexical skills; it also requires mastering a new phonetic and phonological system. A key aspect of this process is speech perception, which directly influences listening comprehension and accurate pronunciation. However, research suggests that learners tend to perceive L2 sounds through the phonological filter of their native language (L1), leading to difficulties in distinguishing unfamiliar phonemes (Flege, 1995; Best, 1995). For Spanish speakers when learning English, these difficulties are particularly evident due to significant phonological differences between the two languages. English has a more complex vowel system, including distinctions in vowel length and quality (Roach, 2009). These differences often lead to misperceptions and pronunciation errors. 2 The primary goal of this study is to analyse how Spanish speakers perceive English sounds and identify the factors that influence their ability to distinguish them. Specific objectives include: • Identifying the English phonemes that cause the most difficulty for Spanish speakers. • Examining the influence of proficiency level on phoneme perception • Assessing the impact of factors such as age of acquisition and linguistic exposure on perception accuracy 1.1 Previous studies Before designing my own study, I reviewed relevant literature to better understand how Spanish speakers perceive English vowel sounds. One of the most influential studies I consulted was The Perception of English Vowels by Monolingual, Bilingual, and Heritage Speakers of Spanish and English by Amanda Boomershine (2013). This research investigates how speakers from different linguistic backgrounds monolingual Spanish and English speakers, bilinguals, and heritage Spanish speakers perceive American English front vowels, particularly those that do not exist in the Spanish vowel system, such as /ɪ/, /æ/, and /ɛ/. The study uses a similarity rating task, where participants listen to pairs of vowel sounds and rate how similar or different they sound. The results showed that monolingual Spanish speakers had more difficulty distinguishing between certain English vowels specially /i/ and /ɪ/ and /e/ and /æ/ due to the absence of these contrasts in Spanish. In contrast, bilingual and heritage speakers demonstrated more accurate perception, performing more similarly to native English speakers, particularly when they had higher proficiency in English or had been exposed to it from an early age. These findings support the idea that greater exposure to English and earlier age of acquisition contribute to improved perception of difficult vowel contrast. This study was especially relevant to my own research because it not only confirmed the phonemes that are most problematic for Spanish learners but also offered insight into how factors such as proficiency and language background can affect auditory perception. Another study that significantly informed the design of this project is The Perception of English Vowels by Spanish Speakers by Andrew R. Jeske (2012), which examines how native Spanish speakers perceive English vowels using the framework of the perceptual Assimilation Model for Language Learners (PAM-L2). The research 3 involved twelve Spanish-speaking graduate students at the University of Pittsburgh who completed three tasks: vowel assimilation, AX categorical discrimination, and forced identification. Therse tasks aimed to analyse how English vowels map onto the Spanish vowel system and whether these mappings could predict the difficulty in perceiving specific contrasts. Jeske’s findings showed that while some English vowel contrasts, such as /ɪ/-/eɪ/, were relatively easy for participants to distinguish, others especially those categorized under single-category assimilation (e.g., /i:/-/ɪ/ and /æ/-/ɑ/) were difficult to discriminate accurately. Notably, the forced identification task revealed that participants often confused similar vowel sounds, such as identifying /i:/ as /ɪ/ more than half of the time. Although PAM-L2 was helpful in some cases, the study concluded that it does not fully account for all discrimination difficulties. The author suggests that factors beyond simple vowel mapping such as exposure, mental representations, and perceptual biases may influence L2 vowel perception. This study reinforced the importance of including minimal pairs involving problematic contrast in my own research and highlighted the need to consider individual learner variables like proficiency and prior exposure when interpreting perception accuracy. A key study that informed the present research is the Perception of English and Spanish Vowels by Native English and Spanish Listeners: A Multidimensional Scaling Analysis by Fox, Flege, and Munro (1995). This study aimed to compare how monolingual English speakers and Spanish-English bilinguals perceive both English and Spanish vowels. Using a multidimensional scaling (MDS) methos, the authors analysed how each group rated dissimilarities between vowel pairs based on acoustic properties. Participants listened to 405 vowel pairings including English and Spanish vowels and rated them on a nine-point dissimilarity scale. The results revealed notable differences between the two groups. English speakers used three perceptual dimensions to evaluate the vowels: vowel height, front/back placement, and a central/noncentral distinction. In contrast, Spanish listeners relied mainly on two dimensions, with vowel height being the most prominent. Importantly, the study found that Spanish listeners’ perception patterns formed clusters around the three native Spanish vowels (/i/, /e/, /a/), suggesting a “perceptual magnet” effect, where L2 vowel sounds are perceived in relation to similar native categories. Furthermore, the researchers compared proficient and nonproficient Spanish-English bilinguals. The vowel space of the more proficient group was found to 4 resemble that of native English listeners more closely, particularly in how English vowels like /eɪ/ and /ɛ/ were perceived. This finding supports the idea that increased proficiency and exposure to the second language can reshape the perceptual vowel space, even in adult learners. This study is particularly relevant for my own research, as it not only confirms the challenges Spanish speakers face with specific English vowel contrasts, but also highlights how linguistic experience and proficiency can alter perception strategies. It reinforces the importance of analysing individual variation and including participant background in the interpretation of results. 1.2 Research Questions and Hypotheses • RQ1: Which English phonemes are the most difficult for Spanish speakers to perceive accurately? Jeske (2012) found that Spanish speakers had the most difficulty perceiving English vowel contrasts such as /i/-/ɪ/, /ɛ/-/æ/, and /ʊ/-/u/. These pairs are hard for native Spanish speakers because Spanish does not use vowel length or certain height distinctions. For example, /i/ was often confused with /ɪ/ more than 50% of the time. “1. Single-category (difficult): /i-ɪ/, /eɪ-ɛ/, and /æ-ɑ/ 2. CategoryGoodness (relatively easy): /æ-ɔ/, /ɑ-ɔ/, /ɔ-ʊ/, /o-ʊ/, and /ʊ-u/” (Jeske, 2012, pp. 42). “If we look at the four least accurately discriminated contrasts from experiment 2 (/i-ɪ/, /ɔ-ɑ/, /u-ʊ/, /ɛ-æ/) we see similar identification patterns emerge. Tables 7 and 8 show us that in three of these four contrasts, the first token was misidentified as the second more often than it was identified correctly: /i/ was labeled as /ɪ/ 54% of the time, /ɔ/ was labeled as /ɑ/ 54% of the time, and /u/ was labeled as /ʊ/ 50% of the time. What this suggests is that the native Spanish speaking participants have not yet established L2 categories that distinguish between /i/-/ɪ/, /a/- /ɔ/, and /u/-/ʊ/.” (Jeske, 2012, p. 43-44). • RQ2: Does proficiency level English correlate with improved perception of difficult phonemes? Both Boomershine (2013) and Fox, Flege and Munro (1995) showed that more proficient learners of English are better at perceiving English vowels. Fox, Flege and Munro compared proficient and non-proficient Spanish speakers and found that the vowel space of proficient speakers was more similar to that of native English speakers. “The vowel space of the proficient Spanish listeners was more 5 English like than that of the non proficient Spanish listeners” (Fox, Flege and Munro., 1995, p.1). Boomershine (2013) also predicted that bilingual speakers would be better at identifying non-native vowels, especially /æ/, /ɛ/, and /ɪ/. • RQ3: Does age of acquisition influence perception accuracy? Fox, Flege and Munro (1995) found a significant connection between age of arrival in an English-speaking country and phonemic perception. Learners who arrived at a younger age had better English vowel perception. “There was a significant negative correlation between the proficiency rankings and age of arrival in the US” (Fox, Flege and Munro, 1995, p. 4). • RQ4: Does Having spent more years in a language school influence perception accuracy? Fox, Flege and Munro (1995) noted that proficient speakers had studied English for more years than non-proficient ones (9.3 years vs 6.6 years). Although not focused on formal instruction, this suggests that more years of language study may help learners hear English vowels more accurately. “They had studied English longer in school than the non proficient listeners (9.3 vs 6.6 years)” (Fox, Flege and Munro, 1995, p. 3). • RQ5: Does having participated is an Erasmus program influence perception accuracy? Fox, Flege and Munro (1995) showed that participants who lived in the United States for longer periods developed more native-like vowel perception. This supports the idea that language immersion experiences, such as Erasmus, can improve phonemic perception. “The proficient learners had lived slightly longer in the US (4.2 vs 3.7 years)” (Fox, Flege and Munro, 1995, p.3) “The vowel space of the proficient Spanish listeners was more English like than that of the non proficient” (Fox, Flege and Munro, 1995, p. 1). Hypotheses: ● Hypothesis 1: For Spanish speakers, the most challenging minimal pairs to perceive are typically those involving vowel sounds that do not exist in Spanish 6 or are very similar from a Spanish phonetic perspective. The following pairs are likely to be the most difficult: • /iː/ vs /ɪ/ • /ʊ/ vs /uː/ These pairs involve distinctions between long and short vowels, which are not phonemic in Spanish. As a result, Spanish-speaking learners of English often struggle to hear and produce these contrasts accurately. Jeske (2012) confirmed that these contrasts are especially difficult because Spanish lacks length and many vowel quality distinctions found in English. “Spanish speakers have difficulty discriminating between instances of /i/ and /ɪ/” (Jeske, 2012, p. 37). “/u/ was labeled as /ʊ/ 50% of the time” (Jeske, 2012, p. 43-44). ● Hypothesis 2: Learners with higher English proficiency level will show greater accuracy when perceiving difficult vowel sounds. This is because more advanced learners usually have had more exposure to different English sounds, better listening skills, and more experience distinguishing subtle differences between similar phonemes. Fox, Flege and Munro. (1995) clearly showed that more proficient learners had a vowel space closer to that of native English speakers. “These results support the claim that the vowel space of the proficient Spanish listeners is more similar to the English space than is the non proficient Spanish vowel space” (Fox et al., 1995, p. 8). ● Hypothesis 3: Participants who began learning English earlier will exhibit significantly higher perception accuracy compared to those who started learning the language earlier, suggesting a critical period effect. Age of arrival was significantly related to proficiency in English phoneme perception. “The proficient listeners were somewhat younger (30 vs 38 years) and had lived slightly longer in the US (4.1 vs 3.7 years) than the non proficient listeners” (Fox, Flege and Munro. 1995, p.3). ● Hypothesis 4: Participants with a greater number of years attending a language academy will demonstrate higher perception accuracy of English vowel sounds, likely due to increased exposure, structured practice, and corrective feedback over 7 time. Although not directly studied, Fox, Flege and Munro. (1995) found that more proficient learners had more years of English instruction. “They had studied English longer in school than the non proficient listeners” (Fox, Flege and Munro, 1995, p.3). ● Hypothesis 5: Participants who have participated in an Erasmus program will achieve higher accuracy in the perception of English vowel phonemes, as living abroad may provide more extensive and naturalistic exposure to the target language. Longer time spent living in an English-speaking country was associated with more native-like vowel perception. “The proficient listeners were somewhat younger (30 vs 38 years) and had lived slightly longer in the US (4.1 vs 3.7 years) than the non proficient listeners” (Fox, Flege and Munro. 1995, p.3) 2. Methodology To carry out this study, a listening perception test featuring minimal pairs of problematic sounds was designed. Participants will listen to words and will have to indicate which word they have heard. The test was created using the Genially platform, focusing on the perception of English vowel sounds. The questionnaire consisted of a series of audio recordings, each containing the pronunciation of a single word. Participants were asked to listen carefully to each recording and select the word they believed had been pronounced from a set of two similar options. A total of 12 participants were individually invited to complete the questionnaire in person. Arranging the sessions personally allowed me to supervise the process directly, ensuring that participants followed the instructions correctly. Additionally, I made every effort to create a relaxed and natural atmosphere during the sessions to minimize participant anxiety, as nervousness could have negatively influenced their performance and, consequently the reliability of the results, being present also enabled me to observe and record the amount of time each participant took to complete the task, providing further data for analysis. The use of minimal pairs was specifically chosen to highlight subtle differences in vowel sounds that are often challenging for Spanish speakers to perceive accurately, the structure of the questionnaire and the method of administering it were designed to create a controlled yet natural testing environment, aimed at capturing genuine perceptual abilities. After collecting their answers, the results were analysed with 14 Learners may focus on the shared lip rounding but miss the significant differences in length, height, and tenseness. 2.2 Questionnaire and lab recordings The questionnaire was made on an app called “Genially”. This software provides users with the ability to design a wide range of interactive content, such as images, infographics, presentations, microsites, catalogues, and maps. Each of these can be enhanced with various interactive features and animations to make them more engaging and dynamic. The tool is fully accessible through the web, meaning there is no need to download or install any software. The free version is typically sufficient for most users’ needs, allowing them to create a variety of interactive project without any cost. However, the paid version offers additional templates and more advances functionality, including the option to save the final presentation directly to your desktop, which can be vey useful for offline user or further editing. I chose this platform because it allowed me to insert voice recordings, which was essential for the type of activity I wanted to do. By using this feature, I was able to integrate listening comprehension exercises and other audiobases tasks, making the questionnaire more interactive and engaging. Additionally, Genially provided a visually appealing and user-friendly interface, which helped ensure that participants could navigate the quiz easily. This combination of audio and visual elements contributed to a more effective and dynamic learning experience. The recordings of the words used in this study were made in the phonetics laboratory at the University of Seville, a facility I was able to access thanks to my BA thesis advisor, Darío Barrera Pardo. Since, as students, we are not authorized to reserve the laboratory, it was necessary for my advisor to make the reservation, as only faculty members with proper authorization are permitted to do so. I contacted him via email, outlining the availability of my foreign participant, and together we arranged a suitable day for the recording session. The recording took place in a soundproof room specifically designed and equipped for this type of word. Once the recording was completed, I saved the files onto my computer, which allowed me to later analyse the material and divide it into segments to be used in the questionnaire that would be send to the participant. Link to the quiz: https://view.genially.com/67c5f80411b68b7c082e10bd/interactivecontent-perception-of-english-sounds 15 2.3 Participants and native speakers This study was carried out with 12 participants, one girl, who is a native speaker of English, who recorded the audios and the rest of them are all native speakers of Spanish, both male and female, aged between 17 and 23. Their experiences with English have been different, some has studied in private language schools, other had participated in Erasmus programs, and some has only learned English at primary and secondary schools, which has likely contributed to the development and reinforcement of their listening and speaking abilities over time. Despite these differences, all participants have an Official English certificate ranging from B1 to C2, providing a reliable measure of their language skills, although these certificates came from different institutions, such as Cambridge, Trinity and Aptis. The native speaker is a girl who is 22 years old. She is from Cambridge and now she is currently participating in the Erasmus program at the University of Seville, which lasts from the entire 2024/2025 academic year. We met during the first semester, as we were both enrolled in the Sociolinguistics course. Bellow, I present a table that include the following information: gender, age, English proficiency level, the type of language examen taken, participation in an Erasmus program, whether the participant has foreign family members and whether they have attended a private language school, if so, the age at which they began studying English and at what age they ended is shown. This table provides a comprehensive overview of the factors that may influence the participants’ perception of English vowel sounds. Table 1: Information about the participants Participants Gender Age Proficiency Certificate Foreign family Erasmus Starting age lang. School 1 Female 17 B1 Cambridge No No 12-16 2 Male 23 B1 Cambridge No No 11-14 3 Male 23 B1 Cambridge No No 11-14 16 Table 1: Information about the participants 4 Male 20 B2 Aptis No No 12-14 5 Female 21 B2 Cambridge No No 18 (4 months) 6 Female 21 B2 Cambridge No Yes 8-17 7 Male 21 C1 IELTS No Yes X 8 Female 21 C1 Cambridge No Yes 5-18 9 Female 20 C1 Cambridge No No 12-16 10 Male 21 C2 Cambridge No Yes 12-15 11 Female 21 C2 Cambridge No Yes X 12 Male 21 C2 Cambridge No No 12-17 2.4 Data analysis The analysis of the data collected in this study was carried out through several approaches. The percentage of correct answers was calculated individually for each participant, this allowed for a general overview of the participants’ perceptual abilities regarding English vowel sounds. In addition, results were compared between different groups based on specific characteristics such as Englich proficiency level, participation in an Erasmus program, attendance at a private language school, and the age at which participants began studying English. These comparisons aimed to identify whether these factors had a noticeable influence on perception accuracy. Furthermore, the analysis focused on identifying which specific English phonemes presented the greatest difficulty for Spanish speakers, by observing the minimal pairs that generated the most frequent 17 errors. This helped address the first research question concerning the most challenging phonemes. Since the questionnaire sessions were conducted in person, it was also possible to observe and record the approximate time each participant spent completing the task. This information was used to consider whether quicker response times correlated with higher or lower accuracy rates, as well as to explore whether participants with higher proficiency or more international experience demonstrated greater efficiency. Finally, relationships between different factors were observed. For instance, tendencies such as participants who started learning English at a younger age, or participants who has participated in an Erasmus program displaying greater accuracy with difficult sounds like, were noted and discussed. The use of simple tables helped to organize the data visually, facilitating clearer interpretation and discussion of the results. 3. Results Table 2: Results of the quiz Proficiency level Participant Words tested Correct answers Incorrect answers Errors Duration (min) B1 1 36 27 9 3-4-5-6-815-23-2635 7:42 B1 2 36 35 1 11-12 6.45 B1 3 36 31 5 5-6-11-1518-21 8:38 B2 4 36 29 7 5-6-7-812-18-36 7:17 B2 5 36 30 6 12-18-2530-31-36 9:10 B2 6 36 32 4 1-3-7-8 10:25 18 Table 2: Results of the quiz C1 7 36 31 3 1-7-18 6.45 C1 8 36 36 0 6:40 C1 9 36 33 4 1-2-7-8 6:42 C2 10 36 35 1 20 7:21 C2 11 36 33 3 5-7-20 9.33 C2 12 36 34 3 5-7-19-20 8:46 Table 3: Percentage of correct answers 1 2 3 4 5 6 7 8 9 10 11 12 75% 97% 86% 81% 83% 89% 92% 100% 89% 97% 92% 92% Table 4: Most common errors in each level B1 heat–hit (/hiːt/ – /hɪt/) [/iː/ vs /ɪ/] but–bat (/bʌt/– /bæt/) [/ʌ/ vs /æ/] luke–look (/luːk/ – /lʊk/) [/uː/ vs /ʊ/] B2 fool–full (/fuːl/ – /fʊl/) [/uː/ vs /ʊ/] cap–cup (/kæp/ – /kʌp/) [/æ/ vs /ʌ/] cook–cork (/kʊk/ – /kɔːk/) [/ʊ/ vs /ɔː/] heat–hit (/hiːt/ – /hɪt/), beat–bit (/biːt/ – /bɪt/), and peak–pick (/piːk/ – /pɪk/), all of which involve the vowel contrast [/iː/ vs /ɪ/] 19 Table 4: Most common errors in each level C1 caught–cot (/kɔːt/ – /kɒt/) [/ɔː/ vs /ɒ/] fool–full (/fuːl/ – /fʊl/) [/uː/ vs /ʊ/] C2 caught–cot (/kɔːt/ – /kɒt/) [/ɔː/ vs /ɒ/] fool–full (/fuːl/ – /fʊl/) [/uː/ vs /ʊ/] 3.1 Interpretation of results In this section, the results obtained from a minimal pair listening quiz are analysed in detail. The analysis focused on the most frequent errors, the recurring difficulties across levels, and the patterns observed within each group of three participants per level. Special attention is given to the vowel contrasts that proves problematic not only at lower levels (B1-B2) but also at more advanced stages (C1-C2). In some cases, even the advances participants had trouble with certain sounds, especially those that do not exist in Spanish. First, the results will be analysed separately by proficiency group (B1, B2, C1 and C2) in order to identify the most frequent errors within each level. For each group, attention will be given to the mistakes that were shared among the three participants of the same level. After this individual analysis, a general comparison will be made to highlight which vowel contrasts were most problematic across all levels. This will help identify the most common error made by Spanish learners of English, regardless of their level of proficiency. In this analysis, it will also be considered whether the participant has attended private lessons at a language academy and for how many years, as well as whether they have taken part in an Erasmus programme. These factors may have an influence on their listening and pronunciation skills. As shown in the table in section 2.3 none of the participants has any foreign family background, which helps isolate the influence of formal education and study abroad experience on their performance in the minimal pairs task. Finally, the duration of the test for each participant will also be discussed, as this is considered an interesting aspect to examine. The time taken to complete the quiz may reflect factors such as confidence, familiarity with the sounds, or hesitation when distinguishing between similar vowel contrasts. Analysing this data can provide further insight into how learners at different proficiency levels approach listening 20 tasks involving minimal pairs. In addition, some specific behaviours observed during the test sessions will be described, as I was present with each participant individually. These observations offer valuable qualitative information that may help explain certain results or patterns identified in the data. In the first group, which is B1, there are three participants. They did not all make the same mistake, but there were some common errors. Participant 1 and Participant 3 both had trouble with heat-hit (/hiːt/ – /hɪt/) [/iː/ vs /ɪ/] and but-bat (/bʌt/– /bæt/) [/ʌ/ vs /æ/], while Participant 2 and Participant 3 both made mistake with Luke-look (/luːk/ – /lʊk/) [/uː/ vs /ʊ/]. It seems that participant 1 who is younger and who started later at the language school made more mistakes than the other two participants. An interesting fact about Participant 2 and Participant 3 is that they are twins, which means they have received the same education, yet their results are different. This point will be discussed in more detail in section 3.2. In the B2 level, the word that all three participants misidentified was fool and full (/fuːl/ – /fʊl/), indicating difficulty with the vowel contrast /uː/ vs /ʊ/. Additionally, Participant 4 and Participant 5 both struggled with cap and cup (/kæp/ – /kʌp/) [/æ/ vs /ʌ/], as well as cook and cork (/kʊk/ – /kɔːk/) [/ʊ/ vs /ɔː/]. On the other hand, Participant 4 and Participant 6 made mistakes with heat and hit (/hiːt/ – /hɪt/), beat and bit (/biːt/ – /bɪt/), and peak and pick (/piːk/ – /pɪk/), all of which involve the vowel contrast /iː/ vs /ɪ/. We can see that Participant 6 has spent the most time in a language school, in contrast to Participant 5, who only attended a four-month intensive course. Interestingly, participants 4 who studied at a language school from the age of 12 to 14 made more mistakes. This could be because their preparation was focused on the Aptis English certificate, which is generally considered to be a lower-level qualification compared to the Cambridge examens. Nonetheless, we observe that the participant with the fewest errors is the one who has spent the most time in a language school and the only one in this group who went on an Erasmus program. In the C1 group, Participant 8 made no mistakes at all. First, it is important to highlight that Participant 7 and Participant 9 both made the same error with caught and cot (/kɔːt/ – /kɒt/), which involves the vowel contrast /ɔː/ vs /ɒ/. Additionally, Participant 9 also mixed up fool and full (/fuːl/ – /fʊl/), showing difficulty with the /uː/ vs /ʊ/ contrast. Although Participant 8 achieved a perfect score, it is interesting to mention that she listened to the recording several times and took longer to respond, which suggests that 21 she found it somewhat challenging and had to pause to think carefully. Participant 8 stands out for having made no mistakes, which could be attributed to having started learning English at the early age of five. Participant 7, although never enrolled in a language school, shares with Participant 8 the experience of having taken part in an Erasmus programme. In contrast, Participant 9 is the one who made the most errors in the group; unlike the others, he has not participated in any international exchange, though he has attended a language school. It is also worth mentioning that bot Participant 7 and Participant 8 are currently studying for a university degree in English Studies. Finally, in the C2 group, Participant 10 made only one mistake, which happens to be the same error shared by all three participants: caught and cot (/kɔːt/ – /kɒt/), involving the vowel contrast /ɔː/ vs /ɒ/. In addition, Participants 11 and 12 both made the same error with fool and full (/fuːl/ – /fʊl/), showing difficulty with the /uː/ vs /ʊ/ vowel distinction. The results show that the participant who achieved the best result in this group is the one who both took part in an Erasmus programme and attended a language school. The most common errors among all participants occurred with the minimal pairs fool [fu:l] and full [fʊl], as well as between heat [hiːt] and hit [hɪt]. Nine participants made mistakes with fool and full, often confusing the vowels. This confusion highlights the difficulty learners face with similar vowel sounds and emphasizes the importance of focusing on both the perception and production of these vowels. Although I recorded the time each participant took to complete the quiz, I couldn’t see any clear correlation between the participant’s level and the time spent on the quiz. However, I did observe that the participants who took more time were the ones who, in order to check how the word is correctly pronounced, they tried to pronounce it themselves before deciding which word to select. This suggests that those who were more involved in actively checking the pronunciation were likely to spend additional time on the task. 3.2 Discussion (Comparison with hypotheses and previous studies) The following section discusses the results obtained from the perception test administered to twelve Spanish-speaking participants with varying levels on English proficiency. The data is interpreted in relation to the research questions and hypotheses presented earlier. 22 Phonemes Most Difficult to Perceive (RQ1) The results reveal that certain English vowel contrasts are particularly problematic for Spanish Speakers, supporting Hypothesis 1. Among the most frequently misidentified pair are: • /ʊ/ vs /u:/ (full-fool), with errors recorded across all proficiency levels, including C1 and C2. • /ɪ/ vs /i:/ (hit-heat), which appeared in the responses of B2 and C1 participants. These findings suggest that vowel contrast involving length, especially those absent in Spanish phonology are more likely to result in perceptual errors. Influence of Proficiency Level (RQ2) A clear correlation can be observed between higher English proficiency and improved phonemes perception, confirming Hypothesis 2. Participants at the C1-C2 level scored between 33 and 36 correct answers out of 36, with several producing only one or zero errors. For example: • Participant 8 (C1) achieved a perfect score (36/36). • Participant 10 (C2) made only one mistake (caught-cot) In contrast, B1-level participants made between 5 and 9 errors, indicating that phonological perception improves significantly with language development and exposure. Influence of Age of Acquisition (RQ3) Hypothesis 3, which proposes a positive effect of earlier exposure to English, is partially supported. For example: • Participant 8 (C1), who began learning English at age 5, achieved a perfect score. • Participant 6 (B2), who started between at age 8, made only 4 errors. • Participant 1 (B1), who started later (at age 12), had 9 errors. While the pattern indicates that early starters often perform better, other factors such as overall proficiency level and recent practice also contribute to the outcome. 23 Years Attending a Language school (RQ4) Hypothesis 4 suggest that participants with more years of attendance at a language school will show higher perception accuracy of English vowel sounds. This could be attributed to factors such as increased exposure to the language, structured practice, and the opportunity to receive corrective feedback over time. However, as the analysis in the previous section shows, while some participants with more time in language schools performed better, there were also cases where the results did not necessarily align with the amount of time spent in formal education. This suggests that other factors, such as the quality of the teaching, the type of exposure to the language (e.g., through international programs like Erasmus), and individual learning experiences, may also play significant roles in shaping vowel sound perception accuracy. While it seems likely that extended time in a language school could contribute to a better understanding of vowel contrasts, the evidence from the participants in this study indicates that it is not the sole determining factor. Therefore, while Hypothesis 4 is supported to some extent, it is important to consider other variables that could also influence the perception accuracy of English vowel sounds. For instance, Participant 2 and Participant 3 have the same number of years of study and the differences in their results may be due to factors such as motivation, exposure, or personal interest in the language. In this case, since the study is relatively small, we can gather more information about the participants. These two are twins, meaning they have had the same opportunities. What is interesting, however, is that despite this, their results differ significantly; Participant 2 scored 25 correct answers, while Participant 3 had 5 mistakes. From the information provided by the participant, we know that Participant 2 has a job that requires the use of English to communicate with foreign colleagues, even though he is a software engineer. Additionally, the music he listens to is mostly in English. On the other hand, Participant 2 has not had any consistent contact with English since he left the language school. This contrast in their results can likely be explained by these differences in motivation and exposure. I would say that this hypothesis hasn’t turned out as expected, as some participants who have never attended a language school have better English skills and results than other who have. Therefore, I associate these differences with each participant’s 30 Table 5: Minimal pairs and IPA 11 Luke /luːk/ 12 Look /lʊk/ 13 Buck /bʌk/ 14 Back /bæk/ 15 But /bʌt/ 16 Bat /bæt/ 17 Cap /kæp/ 18 Cup /kʌp/ 19 Caught /kɔːt/ 20 Cot /kɑːt/ 21 Port /pɔːrt/ 22 Pot /pɑːt/ 23 Vet /vɛt/ 24 Vat /væt/ 25 Heck /hɛk/ 26 Hack /hæk/ 27 Sat /sæt/ 28 Set /sɛt/ 29 Chick /tʃɪk/ 30 Check /tʃɛk/ 31 Table 5: Minimal pairs and IPA 31 Deck /dɛk/ 32 Dick /dɪk/ 33 Pick /pɪk/ 34 Peck /pɛk/ 35 Cork /kɔːrk/ 36 Cook /kʊk/