Contrastive study of english voiceless stops by native and non-native speakers

FACULTAD DE FILOLOGÍA GRADO EN ESTUDIOS INGLESES TRABAJO DE FIN DE GRADO CURSO: 2024 / 2025 TÍTULO: CONTRASTIVE STUDY OF ENGLISH VOICELESS STOPS BY NATIVE AND NON-NATIVE SPEAKERS AUTORA: CLARA MARÍA ROSILLO GARCÍA FECHA: 28 / 05 / 2025 TUTOR: DARÍO BARRERA PARDO FIRMA ALUMNA: FIRMA TUTOR: 0. Abstract This study investigates the production of the voiceless stops consonants of English (/p/, /t/, /k/) by native speakers of Spanish who are learning English as a second language (L2), comparing them with native speakers of English. Although there is no doubt that these phonemes exist in both languages, the way in which they are articulated differs significantly between the two languages, specifically in terms of the time interval between the release of the sound and the onset of vocal fold vibration (VOT). The present experiment analyzed the VOT in 15 English words produced by two groups: five native English speakers and five native Spanish learners, who are aged between 20 and 26 years. Each participant recorded the word list a total of four times, resulting in approximately 600 samples. The results of the recordings revealed that native Spanish speakers produced the voiceless stops with shorter VOT values compared to native English speakers. These findings support previous research with conclusions similar to those of this study, as they suggest that second language learners tend to face difficulties in accurately imitating the phonetic patterns of the target language, due to the influence of their native language, even after some years of exposure to the target language. Keywords: voiceless stops, VOT, native English, nonnative English Index 1. Introduction .......................................................................................................... 1 2. Method .................................................................................................................. 4 2.1. Design ............................................................................................................... 4 2.2. Participants ....................................................................................................... 4 2.3. Materials ........................................................................................................... 5 2.4. Procedure .......................................................................................................... 6 2.5 Analysis ............................................................................................................ 6 3. Results .................................................................................................................. 9 3.1. Descriptive statistics ......................................................................................... 9 3.2. Inferential statistics .......................................................................................... 11 4. Discussion .......................................................................................................... 16 4.1. My experiment’s contribution to knowledge in this area of phonetics .......... 18 4.2. Faults and limitations ..................................................................................... 20 5. Conclusion .......................................................................................................... 21 6. Acknowledgements ............................................................................................ 23 7. Bibliography ....................................................................................................... 23 1 1. Introduction The globally dominant language today is English, which offers great advantages such as connecting people across important sectors such as business, science and entertainment. As it is considered a lingua franca, it enables trade and universal communication in a very effective way. It has approximately 375 million speakers as a mother tongue and over 750 million speakers as a second language, making English the most widely used language worldwide (Rao, 2019). Among the number of those learning this language as a second language are a large number of native Spanish speakers, who struggle to cope with a number of specific difficulties due to linguistic variations between the two languages. Pronunciation is considered one of the most exhausting factors, because of the phonological system of English, which compared to Spanish is considered to be more complex. The learner's mother tongue has a significant influence on their competence in acquiring English sounds and structures, which in some situations makes it difficult for him or her to dominate certain pronunciations. According to the Speech Learning Model (Schoonmaker-Gates, 2012), when an individual learns a new language, they are inclined to associate the new sounds with those from their native language. For example, if in Spanish and English there is a sound that is very similar between the two languages, the learner may consider it to be exactly the same, and therefore not appreciate the important small details that differentiate them. By unintentionally ignoring them, this makes it difficult to pronounce some English words correctly, such as words that contain aspiration or words that differ in VOT to a great extent. Speaking with a pronunciation identical to that of a native speaker is a great challenge for people learning a second language, as only 5% manage to achieve it. One of the reasons is that they retain some phonetic features of their native language, and as a result, very few individuals are capable of fully mastering English pronunciation (Thornburgh & Ryalls, 1998). A number of factors influence second language pronunciation, of which consonant production is valued as one of the most important. Consonants play a crucial role in speech accuracy, accent and intelligibility, and thus directly affect the way in which a non-native speaker is appreciated. This study focuses on consonants, specifically on voiceless stops. English and Spanish share the phonemes /p/, /t/ and /k/, however their 2 pronunciation is not similar between the two languages. In English, voiceless stops are aspirated when they appear at the beginning of a word, meaning there is a noticeable burst of air (Voice Onset Time or VOT) before the vowel begins. In contrast, Spanish stops are articulated differently, with regional variations. Generally, Spanish speakers produce /t/ and /d/ as denti-alveolar sounds, where the tongue makes contact closer to the teeth than the alveolar ridge. Unlike English, voiceless stops in Spanish do not have strong aspiration at the beginning of words (Gravelle, 2020). A little more than fifty years ago, Lisker and Abramson introduced a simple way to measure acoustic differences between stop consonants based on their voicing, known as VOT. Since then, this method has been used in hundreds of studies (Abramson & Whalen, 2017). The time interval between the release of a voiceless stop consonant (in this case, /p, t, k/) and the start of vocal fold vibration for the following vowel defines what is known as Voice Onset Time. In everyday terms, this gap is often perceived as a burst or slight aspiration when producing English /p, t, k/. This study adopts the VOT definition by Cho & Ladefoged (1999), which specifies that VOT begins at the moment the voiceless stop is released and concludes at the first zero crossing that marks the onset of complete vocal fold vibration. In Spanish, the voiceless stops /p, t, k/ are produced with a nearly simultaneous release and onset of vocal cord vibration, resulting in VOT values that range from zero to 10 milliseconds. In English, voiceless stop consonants /p, t, k/ have a noticeable delay between their release and the start of vocal cord vibration. This delay, known as VOT, lasts at least 30 milliseconds and corresponds to the period of aspiration (Benkí, 2005). Previous studies such as those by Bond & Bermejo (1980), found that Spanish speakers learning English tend to produce /p, t, k/ with VOT values that are notably shorter than those of native English speakers, especially if they develop "coexistent" phonetic systems. The inability of L2 learners to authentically produce English /p, t, k/ in earlier studies may be linked to the age at which the participants began learning English. Fokes & Steinberg (1985) found that Arabic-speaking children between the ages of 2 and 11 produced /p, t/ in English words with VOT values similar to those of native English speakers (82 ms). Although their speech was not directly compared with that of native English children of the same age, this suggests that these Arabic-speaking children may 3 have been closer to the English VOT pattern than adult L2 learners in earlier studies (Flege & Eefting, 1987). The purpose of the present study is to analyse and compare the VOT of voiceless stops (/p, t, k/) in English, as produced by Spanish-speaking learners and native speakers of English. To that end, five Spanish-speaking learners of English as a second language and five native speakers of English, aged between 20 and 26, were recorded pronouncing 15 English words, each beginning with one of the aforementioned phonemes. Once the data is collected, the results will be organized and presented in a descriptive statistical table for analysis. By examining VOT differences between the two groups, this study seeks to provide a detailed phonetic analysis of how Spanish learners of English articulate voiceless stops in comparison to native speakers, contributing to a better understanding of L2 phonetic acquisition. The current research is part of a project that describes the phonetic features of English obstruents produced by native speakers and Spanish learners of English (Barrera-Pardo, in preparation). 1. Research question: To what extent does the VOT of English voiceless stop differ between native English speakers and English learners? 2 Hypothesis: Based on previous research and findings from other studies, my hypothesis for this study is that Spanish learners of English will produce the voiceless stops /p, t, k/ with shorter VOT values than native English speakers (Flege & Eefting, 1987). Due to the influence of the phonetic features of their mother tongue, Spanish speakers are expected to produce VOT values that are intermediate with respect to the VOT patterns of both languages. In other words, since their production shows similarities to Spanish, their adaptation to English is not fully complete. 4 2. Method 2.1. Design In terms of the formal design of my study, I focused on two types of variables: independent and dependent. A variable is something that can take on different values. It is essential to distinguish between these two types of variables in order to obtain an appropriate interpretation of data. An independent variable is characterised by the fact that it is not affected by other variables in the study: it is autonomous with respect to the other variables and can influence the dependent variable. In contrast, a dependent variable represents the outcome that researchers observe, and is modified depending on the independent variables. In the context of my research, the dependent variable corresponds to the VOT, given that it is altered between speakers and samples, and is measured in milliseconds. The independent variables are stop (with three levels: /p/, /t/, and /k/, which are voiceless stops) and L1 (with two levels: English and Spanish). These two factors are expected to affect the way VOT is produced in English. The aim is to examine whether the place of articulation of the voiceless stop, as well as the speaker’s native language, influence the duration of the VOT. 2.2. Participants Regarding the participant details, five native English speakers took part, consisting of two women and three men (some of whom were from Washington, California or United Kingdom), and five native Spanish speakers, all of whom were women (coming from Seville, Spain). As mentioned earlier, all participants were between the ages of 20 and 26 years. The participant characteristics are presented in Table 1: 5 Table 1. Speakers’ characteristics. Speaker Gender Age L1 1 Female 21 English 2 Female 23 Spanish 3 Female 24 Spanish 4 Female 25 Spanish 5 Female 26 Spanish 6 Female 23 Spanish 7 Male 21 English 11 Female 20 English 12 Male 20 English 13 Male 21 English 2.3. Materials The stimuli for this experiment consisted of 15 English words, each beginning with one of the target voiceless stops: /p/, /t/, and /k/. The selected words were: part, cod, tap, pit, kept, king, top, put, tip, ten, pot, pet, cat, tooth, and cook. All of them were chosen to ensure that the target consonants appeared in word-initial position followed by a stressed vowel, which is ideal for measuring VOT. Each participant read the full word list four times, with the aim of obtaining complete statistics and to avoid small mistakes by the participants, giving them more opportunities. The stimuli were presented in written form on a printed list during the recording session. The same set of words was used for both the Spanish learners of English and the native English speakers, ensuring consistent conditions for comparison. As a result, a total of 600 measurements were analyzed among all participants. 6 2.4. Procedure This experiment was carried out in several stages. First, we enlisted the 10 participants (including myself) to record the corresponding English words in the recording studio. Finding the native Spanish speakers was relatively easy, but it was more challenging to locate the native English speakers. It took us longer to find them and to get them to agree to participate in the recordings. Once we had them, the recordings took place in the phonetics laboratory at the University of Seville, where each speaker filled out a form answering questions such as their gender, age, place of birth, and language proficiency. The speakers' names were kept completely anonymous. The next step was to enter the recording room, where we put on headphones and began reading the 15 English words containing the sounds /p, t, k/. The participants recorded the words four times in the carrier sentence ________ is the word I say in the phonetics lab, and this produced 600 tokens. Each recording session lasted no more than 20 minutes. Once the recordings were completed, we collected the VOT data for each speaker using the Microsoft Excel tool (Microsoft Corporation, 2018), with the assistance of PRAAT, which allowed us to analyse and visualize the speech sounds from each of their repetitions. 2.5 Analysis The following section presents analyses of the VOT of two native English speakers and two Spanish learners, using the software PRAAT. Figure 1 shows an example of how VOT was extracted from a speech sample, based on the methodology described in Cho & Ladefoged (1999). This figure shows a waveform and spectrogram of a native English speaker producing the voiceless stop /k/ in wordinitial position, obtained using the software PRAAT. This acoustic representation illustrates how VOT is measured — specifically, the interval between the release burst of the stop and the onset of voicing, visible as the beginning of regular vocal fold vibration. In this case, the speaker is saying the word cod. The VOT can be seen in the red shaded area, which is the moment when the release of air occurs during the articulation of the /k/ sound (visible as a sharp peak in the waveform), and ends at the onset of vocal fold vibration that produces the continuous vowel, which 13 provides evidence on whether the observed patterns in VOT can be extended to the broader populations of L1 English and L1 Spanish speakers. The results of the ANOVA are as follows: Table 3. ANOVA with VOT as the dependent variable and stop and L1 as independent variables ANOVA – VOT p STOP < 0.001 L1 < 0.001 STOP ✻ L1 < 0.001 The p in the table stands for probability, that is, how likely it is that the results obtained are representative of the population (instead of the sample). In statistics, a probability smaller than 5% is accepted as indicating a reasonable amount of certainty that the results are representative of the population. The results of the ANOVA for stop (the first independent variable) is p < 0.001, which is clearly below 0.05 (the 5% mentioned earlier). This means that the three stops have statistically different VOT values. The results of the ANOVA for L1 (the second independent variable) is p < 0.001, which is also clearly below 0.05. This indicates that the two L1s have statistically different VOT values. Finally, the interaction between the independent variables is examined, which is the most relevant part of the analysis. The two independent variables show a p value of < 0.001, which again suggests that the stops across both L1s exhibit a statistically significant difference in VOT values. The program then compares the interaction of the two independent variables, with the results plotted in Table 4. 14 Table 4. Post Hoc Comparisons – STOP ✻ L The post hoc comparisons further revealed specific differences between stop-L1 combinations. - The difference in VOT between L1 English /p/ and L1 Spanish /p/ is 29.41 ms, and this difference is statistically significant, since p < 0.001 (which is below 0.05). - The difference in VOT between L1 English /p/ and L1 Spanish /t/ is -11.40 ms, and this difference is statistically significant, since p = 0.014 (which is below 0.05). Comparison STOP L1 STOP L1 Mean Difference p p English p Spanish 29.41 < 0.001 t English -11.40 0.014 t Spanish -1.47 0.998 Spanish t English -40.82 < 0.001 t Spanish -30.88 < 0.001 t English t Spanish 9.94 0.047 k English k Spanish 11.49 0.011 p English 8.12 0.183 p Spanish 37.54 < 0.001 t English -3.28 0.932 t Spanish 6.66 0.388 Spanish p English -3.37 0.928 p Spanish 26.04 < 0.001 t English -14.77 < 0.001 English t Spanish -4.84 0.728 15 - The difference in VOT between L1 English /p/ and L1 Spanish /t/ is -1.47 ms, and this difference is not statistically significant, since p = 0.998 (which is much higher than 0.05). - The difference in VOT between L1 English /p/ and L1 Spanish /t/ is -40.82 ms, and this difference is statistically significant, since p < 0.001 (which is below 0.05). - The difference in VOT between L1 English /p/ and L1 Spanish /t/ is -30.88 ms, and this difference is statistically significant, since p < 0.001 (which is below 0.05). - The difference in VOT between L1 English /t/ and L1 Spanish /t/ is 9.94 ms, and this difference is statistically significant, since p = 0.047 (which is below 0.05). - The difference in VOT between L1 English /k/ and L1 Spanish /k/ is 11.49 ms, and this difference is statistically significant, since p = 0.011 (which is below 0.05). - The difference in VOT between L1 English /k/ and L1 Spanish /p/ is 8.12 ms, and this difference is not statistically significant, since p = 0.183 (which is greater than 0.05). - The difference in VOT between L1 English /k/ and L1 Spanish /p/ is 37.54 ms, and this difference is statistically significant, since p < 0.001 (which is below 0.05). - The difference in VOT between L1 English /k/ and L1 Spanish /t/ is -3.28 ms, and this difference is not statistically significant, since p = 0.932 (which is much higher than 0.05). - The difference in VOT between L1 English /k/ and L1 Spanish /t/ is 6.66 ms, and this difference is not statistically significant, since p = 0.388 (which is greater than 0.05). - The difference in VOT between L1 English /k/ and L1 Spanish /p/ is -3.37 ms, and this difference is not statistically significant, since p = 0.928 (which is much higher than 0.05). - The difference in VOT between L1 English /k/ and L1 Spanish /p/ is 26.04 ms, and this difference is statistically significant, since p < 0.001 (which is below 0.05). 16 - The difference in VOT between L1 English /k/ and L1 Spanish /t/ is -14.77 ms, and this difference is statistically significant, since p < 0.001 (which is below 0.05). - The difference in VOT between L1 English /k/ and L1 Spanish /t/ is -4.84 ms, and this difference is not statistically significant, since p = 0.728 (which is greater than 0.05). As for the differences between languages, the data collected show that native English speakers tend to produce longer VOTs than native Spanish speakers with respect to the same sounds, indicating that aspiration is an aspect more commonly used in English. This is reflected in the following examples: the difference between /p/ in English and /p/ in Spanish (29.42 ms), as well as between /k/ in English and /k/ in Spanish (11.49 ms), both showing a significant distinction between the two sounds (p < 0.001 and p = 0.011). These results reaffirm that English voiceless stops are pronounced with greater aspiration. In contrast, focusing on similarities, in certain cases the VOT values tend to be similar between the two languages. The reason for this can be found in sounds that are articulated in similar parts of the mouth, which tend to be very similar acoustically. For example, no significant distinction was found between the VOT values of /k/ in English and /t/ in Spanish (p = 0.388), suggesting that these sounds may be perceived as similar. 4. Discussion The results of the study revealed that Spanish-speaking learners of English produced shorter voice onset times (VOT) for the voiceless stops /p/, /t/ and /k/, compared to native English speakers. This difference was notably observed for the stops /p/ and /k/, where native speakers showed significantly longer VOTs. According to Flege & Eefting (1987), the pattern of English as a second language learners evidences the impact of the mother tongue (L1) influence, which leads Spanish-speaking learners to approach English voiceless stops with shorter VOTs, similar to their native language, due to the characteristics of the Spanish phonological system. 17 The data also indicated that both the type of stop and the speaker’s native language had a clear influence on VOT production. In general, English speakers produced more strongly aspirated stops, while Spanish speakers’ productions reflected patterns typical of their first language, with less aspiration. The most notable difference was found between the English /p/ and the Spanish /p/, with a mean difference of approximately 30 ms (p < 0.001). This clearly reveals the influence of the mother tongue on the acquisition of a second language. My results align with the study by Benkí (2005), who found that Spanish speakers recognize a consonant as voiceless with a shorter VOT than English speakers, who need a longer VOT to differentiate them. This shows that mother tongue influences the way they perceive and produce sounds in English. Even though they have been exposed to English for a long time, Spanish-speaking participants still maintained native-like patterns, which is why they make less use of aspiration in voiceless consonants. These results are consistent with developmental studies showing that younger speakers initially produce stops with short lag VOTs and gradually develop the articulatory control needed to produce longer lags or pre-voicing, as observed by Bond & Bermejo (1980), who found that the ability to produce adult-like VOT values emerges progressively with age and motor development. These findings directly answer the research question by showing that there is indeed a significant difference in the VOT of English voiceless stops between native English speakers and Spanish learners. The statistical analysis confirmed that both stop type and L1 background play a crucial role in shaping how these sounds are produced. Moreover, the original hypothesis is fully supported by the results: Spanish speakers learning English are inclined to produce shorter VOTs than native speakers. This idea proposes that even higher-level learners may retain phonetic features of their L1, especially in terms of subtle acoustic aspects such as aspiration. These results can contribute to a better understanding of how pronunciation is learned and can be improved. As it has been proven, one of the aspects that still persists in native Spanish speakers is the lack of typical English aspiration, for example, they still use shorter VOT values in English words compared to native English speakers. One possible improvement would be to teach pronunciation focusing more on the details that learners find more difficult, such as VOT differences. Teachers could make use of exercises aimed directly at getting learners to listen to the typical English VOT duration, and produce 18 those sounds correctly. This would lead to more direct and efficient learning with respect to pronunciation. In addition, they would be more aware of how their native language influences them when learning a second language, and thus correct certain common errors. Likewise, showing them the sounds graphically—for example by means of spectrograms—they could deepen their understanding and make a more controlled use of their pronunciation, especially in the aspects discussed. 4.1. My experiment’s contribution to knowledge in this area of phonetics The findings of this study are consistent with those of previous research on VOT production in Spanish speakers learning English. For example, Maldonado-Valentín (2007) found that Puerto Rican Spanish speakers with more advanced levels of English are able to produce English voiceless stops with VOT values very close to those of native speakers, while learners with lower proficiency showed a noticeable influence from their L1, resulting in shorter VOTs. My results support this trend: although all Spanish speakers in my study were advanced learners, their production of English /p/, /t/, and /k/ still showed considerably shorter VOT values than those of native English speakers, particularly for /p/ and /k/. This confirms that L1 influence can persist even at higher levels of proficiency, and supports the idea that L2 phonetic acquisition is progressive and often incomplete. Thus, this study supports previous findings and provides additional evidence that Spanish learners tend to retain features of their L1 when producing voiceless stops in English. Furthermore, the present study aligns with the findings of Flege & Eefting (1988), who found that Spanish speakers tended to produce stops with shorter VOTs, usually without reaching the long VOT values similar to those of native English speakers. Their results indicate that, although learners may improve in producing new forms in the second language, their pronunciation is still likely to be influenced. My data reflect this as well: Although the Spanish speakers were able to approximate the aspirated sounds in English, they did not exactly replicate the typical phonetic patterns of native speakers. This suggests that their English pronunciation still retains Spanish phonetic features, which 19 may be because they do not yet have a complete understanding of how these sounds should be pronounced in the target language. In addition, these findings reinforce the data reported by Flege & Schmidt (1995), who analyzed how native English speakers and native Spanish speakers perceived the pronunciation of /p/ in English when produced with varying types of aspiration, differing in VOT. They found that Spanish learners of English gave higher ratings to stops with VOT values similar to those used by native speakers (around 50 ms), and also rated sounds with shorter aspiration times more positively than native speakers did. This suggests that, although advanced second language learners have an easier time perceiving sounds similarly to native speakers, their evaluation of those sounds is still influenced by their native language. My study shows similar results with respect to that conclusion: although Spanish-speaking participants had a higher level of proficiency, the VOTs they produced were still shorter than those of native speakers. This is evidence that although learners acquire new phonetic patterns specific to the second language, phonetic features of their native language remain present in their pronunciation. Finally, these results are in line with the analysis by Shimizu (2011, August), who carried out a comparative study in order to analyze how Korean, Thai and Mandarin Chinese speakers make use of VOT to differentiate stop consonants in their native languages and adapt those native patterns when learning English. Through the study, it was shown that these learners are inclined to relate their native stop patterns to those of the target language based on VOT values that are similar. For example, Korean speakers related their tense stops to English voiceless stops —sounds produced without vocal fold vibration—even though Korean tense stops are also characterized by greater articulatory tension and a lack of aspiration. At the same time, Mandarin and Thai speakers associated their aspirated stops with English voiceless stops, probably because both languages use aspiration. Although my study is focused on Spanish speakers, the results show a similar pattern: the VOT values for English voiceless stops produced by the participants were shorter, which may have been influenced by the shorter VOT durations typical of their L1. These data reinforce the idea that learners employ the sound patterns of their native language that resemble the sounds of the new language. 20 4.2. Faults and limitations As previously mentioned, this study offers an analysis of VOT values for English voiceless stops by native speakers and Spanish learners. While the findings support the influence of the mother tongue, several limitations must be acknowledged. One of the main limitations has to do with the study sample, which can be considered small. This could reduce the validity of the findings when extended to a larger population. The dataset includes a similar number of examples for each voiceless consonant. However, there is no consistent evidence to confirm that these data represent a sufficiently diverse sample of speakers, with respect to aspects such as gender, dialect, and language proficiency level. This could lead to doubts about the reliability of the results obtained in the VOT production. Another limitation is the lack of attention to certain aspects of speech, such as sentence stress and speaking speed, which may influence VOT. However, considerable attention was given to the position of the target words, which appeared at the beginning of the sentence (e.g., ‘Cat is the word I say’). Even so, the uncontrolled variables may have introduced inconsistencies that hide more subtle patterns in the data. In addition, the present study only elicited target words in utterance-initial position, while prior research (e.g., Flege, 1991) has shown that VOT patterns may vary depending on phonetic context. Including the target words in other positions (e.g., in the middle of the utterance) might facilitate a better understanding of how VOT production varies in bilinguals depending on the word’s position in the sentence. Furthermore, participants from both the United States and the United Kingdom were included in the native English speaker group. Placing them in the same group might have introduced some variation into the results, since there are known differences in the production of voiceless stops across these language varieties of English. Similar limitations have been noted in previous research. Schoonmaker-Gates (2012) for instance, found that VOT explained only a small part of foreign accent perception, and emphasized the importance of other linguistic cues and listener-related factors. As in the aforementioned study, this work does not consider participants' possible exposure to the different native varieties of English, nor does it consider whether they have ever been 21 familiar with VOT as a phonetic cue. Analyzing these factors would be quite useful in future research, as participants' pronunciation could be influenced by them. To try to remedy these limitations, future studies should work with a larger sample size to ensure an efficient and proportional representation of speaker variables, such as regional dialects and assessing speakers’ proficiency levels assessed through official language proficiency tests. Another option would be to distinguish between dialectal varieties such as American and British English. This could reduce variability that can interfere with the results and facilitate a better understanding of how L1 influences VOT production. Another interesting research approach could be to examine phonetic elements in more depth beyond VOT, such as the aspiration noise of the VOT or the characteristics of the voice when the following vowel is initiated. These aspects could help to obtain a broader analysis on the production and perception of voiceless stops, as well as provide further insight into why accent perception varies, as this is something that cannot be explained by VOT duration alone. Finally, it would be useful for future research to conduct long-term studies to gain a better and more in-depth understanding of how VOT production develops as learners reach higher levels of English. Researchers could observe the same group of learners over a long period of time, perhaps months or even years, to see whether their pronunciation becomes more like that of native speakers as they use the language regularly. This would make it possible to see whether the patterns observed in this study change as the learners' level improves with respect to their continued exposure to the target language, or whether those patterns are only temporary. 5. Conclusion This study focused on the analysis of the different values of VOT duration for the English voiceless stops /p/, /t/ and /k/ produced by native speakers of English and Spanishspeaking learners of English with a high level of proficiency. The findings reflected that 22 the VOTs produced by the Spanish learners were clearly shorter in duration especially for /p/ and /k/, further suggesting a strong influence of their mother tongue. Through statistical analysis (ANOVA), it was corroborated that the type of consonant has a relevant influence on the production of VOT, and also that features of the mother tongue persistently appear in the VOT patterns. These data support the idea that the phonetic system of the mother tongue has a strong influence on the acquisition of a second language, even in highly proficient learners. Doing this research personally has been a very consistent and enriching work, full of new learning and detailed practice in relation to phonetics. Learning so much about the differences in VOT production between native speakers and myself (as a Spanishspeaking learner) has been very interesting, especially since I had no idea that so much type of data was collected and, above all, that there was such variation in aspiration at the onset of voiceless stop. I really enjoyed discovering how much our mother tongue influences us, because without this study, I wouldn't be aware of all the details involved in second language production. I was particularly curious when analyzing each recording and observing the VOT of each voiceless stop, even how slightly it varied in the different recordings of the same individual. Although there were quite a few recordings to analyze, being able to thoroughly examine the pronunciations of native speakers was incredible, one of the best experiences I have had in my academic journey. As for the difficulties I have had, being able to find native English speakers was a somewhat challenging task. I spent days preparing posters and distributing them in many places, as well as walking around the streets and the university, asking people in person if by any chance they were native English speakers. Some were a little surprised at first, although they were very kind afterward. Another challenge I faced during the study was that it was sometimes difficult to collect the spectrogram data, as the VOT region where the voiceless stop occurred was not always clearly visible at the endpoint, so I hesitated with some data, especially from the learners.