Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
FACULTAD DE EDUCACIÓN
PEDAGOGÍA EN INGLÉS
CONCEPCIÓN, 2017
1
UNIVERSIDAD DE CONCEPCIÓN
FACULTAD DE EDUCACIÓN
PEDAGOGÍA EN INGLÉS
CONCEPCIÓN, 2017
2
“The power of imagination is the
ability to create your own future”
3
ACKNOWLEDGEMENTS
Since we started to work on this thesis, we have been blessed with the company of
many people we would like to acknowledge in this page. This thesis would not have
been possible without the kind support and wise guidance of Professor Yasna
Pereira. It is because of her that we were able not only to complete this process, but
also to see phonetics in a way we could not have imagined before. We cannot
express how grateful we are for the opportunity to work with her.
Also important was the unconditional love and support of the Figueroa-Herrera and
Cuevas-Diaz families. They trusted in our work and provided us with encouragement
from our first years in this world until now that we are finishing our years of study.
We would also like to recognize the support of our friends and relatives, who were
by our side and always found a way to rise our spirits. In a special way we are grateful
for having Claudia Arenas “Kuky” in our lives, whose support and friendship has
been with us for many years.
Finally, we are grateful with the foreign languages department of our University and
its teachers, who taught us much more than just the language we learnt. Thank you
very much to all of you.
4
ABSTRACT
* This research is part of the project “Language attrition in Teachers of English: The
Perception of English Vowels”. VRID UdeC: 216.065.021-1.OIN
5
RESUMEN
Aprender las vocales del inglés representa un reto para los hispanohablantes,
principalmente debido a las diferencias entre ambos sistemas de vocales (Iverson y
Evans, 2007). Hay una gran variedad de factores que afectan este proceso (Flege,
1995; Kuhl, 2000). El propósito de este estudio es encontrar los factores que afectan
la percepción de vocales inglesas en estudiantes avanzados. Para ello, se aplicó
una encuesta a un grupo de estudiantes de cuarto año de un programa de formación
de profesores de inglés acerca de su experiencia con el idioma y sus hábitos con
respecto a su exposición al idioma. Además, se aplicó una prueba de nivel, una
prueba de audición que midió su capacidad para discernir pares mínimos y una
prueba de identificación de vocales. Se evaluó la relación entre la capacidad del
participante para percibir las vocales inglesas y las variables bajo estudio. El análisis
mostró una relación moderadamente significativa entre la capacidad de escucha
demostrada en la prueba de pares mínimos y los resultados en la prueba de
percepción de vocales.
6
INDEX
Acknowledgements 4
Abstract 5
Resumen 6
Index 7
List of figures 9
CHAPTER I: INTRODUCTION 10
1.1 Presentation of the problem 10
1.2 Research questions 10
1.3 General objectives 10
1.4 Specific objectives 11
1.5 Importance of the study 11
CHAPTER II: LITERATURE REVIEW 13
Factors related to L2 Speech perception 13
2.1 Linguistic factors 13
2.1.1 Listening capacity as a predictor of perception of vowels 13
2.1.2 Language experience as a predictor of perception of vowels 15
Perceptual assimilation model 15
Speech learning model 16
2.2 Non-linguistic factors 16
2.2.1 Motivation as a predictor of perception of vowels 16
2.3 English vowel perception difficulties 18
2.3.1 Duration 18
2.3.2 Typical confusions for L1 Spanish speaker learners 19
2.4 L2 speech perception and production link 19
2.5 Implications for teaching and learning 23
2.6 Training to improve vowel perception 25
Summary 26
7
CHAPTER III: METHODOLOGY 28
3.1 Participants 28
3.2 Testing material 28
3.2.1 Language proficiency test 28
3.2.2 Listening Test 29
3.2.3 Vowel Identification Test 30
3.3 Survey 32
3.4 Statistical analysis 32
CHAPTER IV: RESULTS 34
4.1 General results 34
4.1.1 Vowel identification test 34
4.1.2 Listening test 37
4.1.3 Language proficiency test 38
4.1.4 Survey 39
4.2 Specific results 42
4.2.1 Confusions in the perception of English vowels 42
4.2.2 Correlations 43
CHAPTER V: DISCUSSION 47
VI. REFERENCES 50
VII. APPENDIX 57
7.1 List of lexical items used in the vowel identification test 57
7.2 List of audio files used in the vowel identification test 59
7.3 Results per participants 62
7.4 Individual results per stimuli in the vowel identification test 63
7.5 Survey 64
7.6 Listening test 68
7.7 List of participants 72
7.8 Poster presented at RICELT 2017 73
8
LIST OF FIGURES
Fig 3.1 Proficiency Test 29
Fig 3.2 Vowel identification test 30
Fig 3.3 Response buttons in the vowel identification test 31
Fig 3.4 End of the Vowel Identification test 32
Fig 4.1 Boxplots for the vowel identification test 34
Table 4.1 Mean score, tense-lax and position distinction per vowel 35
Table 4.2 T-test between tense-lax vowels 36
Fig 4.2 Variability in individual performance vowel identification test 37
Fig 4.3 Variability in individual performance Listening Test 38
Fig 4.4 Hours of English outside the classes 39
Fig 4.5 Use of English Skills 39
Fig 4.6 Participants’ self-assessment 40
Fig 4.7 Participant’s activities to improve pronunciation 40
Fig 4.8 Participant’s strategies to learn pronunciation 41
Table 4.3 Most common confusions in the vowel identification test 42
Fig 4.9 Scatter plot Correlation vowel identification test and listening test 43
Fig 4.10 Scatter plot Correlation vowel identification test and proficiency test 44
Fig 4.11 Scatter plot Correlation vowel identification test and listening input 45
Fig 4.12 Scatter plot Correlation vowel identification test and self-assessment 46
9
CHAPTER I: INTRODUCTION
Pronunciation is an important part of any language and English is not the exception.
This is why teaching and learning pronunciation is of utmost importance. The
perception of speech is a major component of pronunciation. In this matter, the
perception of speech, and more specifically to perceive the vowels of the English
language becomes the focus of this thesis. Considering this, in order to understand
this aspect of linguistics, the factors that influence the perception of the English
vowels will be looked into.
Find the relation between linguistic (listening capacity and language proficiency) and
non-linguistic (motivation) factors with the capacity to perceive English vowels in
advanced learners of English.
10
1.4 Specific objectives
Find the relation between the learners’ listening capacity and the vowel
perception capacity.
Find the relation between the learners’ language experience and the vowel
perception capacity.
Find the relation between the learners’ language proficiency and the vowel
perception capacity.
Explore the contribution of motivation in the learners’ general capacity to perceive
L2 sounds.
The focus of this thesis is on the perception of English vowels. Almost every time a
person produces sounds, they are producing vowels and thus the interlocutor
perceives vowels. In English, vowel contrast perception is slightly more important
than consonant perception. A misperception of a vowel may lead to a
misunderstanding in communication. Additionally, in accent distinctions, most of the
differences are focused on the use of vowels. (M. Nespor, 2002).
11
The most accepted theory is that the reduced or relatively small vowel system of the
Spanish language speakers may affect the way they learn other vowels. There is
evidence for this in studies that compare the performance of learners of English with
German as L1 with English learners with Spanish as L1 (Iverson and Evans, 2009).
Their study concluded that as the German language has a larger vowel system,
German speakers have an advantage when learning new vowels in comparison with
Spanish speakers.
Vowels were included in this thesis considering also the suggestion made by Flege
and Bohn that there is no critical period for vowel learning (Flege, Bohn & Jang,
1997). This means that there is a chance to teach vowels at every age and that
mistakes in production of vowels can be corrected at any time.
For the purpose of this thesis, several possible factors were considered when
searching for the ones that may be influential in the perception of English vowels.
Listening capacity, motivation and language proficiency were selected to be tested.
12
CHAPTER II: LITERATURE REVIEW
Different factors that may influence the perception of a second language (L2) have
been explored such as age, use of L2, motivation, among others (Cenoz,
Lecumberri, 1999). In this thesis, linguistic and non-linguistic factors were
considered such as learners’ listening capacity, learners’ language proficiency level,
and learners’ motivation. In this chapter, the relevance of this thesis in the context of
L2 speech perception and production and the literature that supports the basis for
this thesis will be presented.
Research has demonstrated that an average adult spends over 40% of their
communicative time listening (Gilman & Moody, 1984). It is logical then, to consider
listening capacity as a highly influential factor affecting the perception of speech in
general and particularly about vowels. Another reason to consider it as a factor is
that, in comparison with other variables that have proved to be significant in the
acquisition of a second language, the listening skill is special in the sense that it can
be improved with training, whereas non-linguistic factors like motivation cannot be
controlled (Feyten, 1991). Following this line of thought, by studying whether there
is a relation between the learners’ listening capacity and the perception of English
vowels, there will be a chance to improve it later by designing training focused on
listening. However, the literature regarding the relation between listening capacity
and the perception of L2 vowels is rather scarce. It is important to state that for the
purpose of this thesis the listening capacity will be understood as the capacity a
learner has to recognize and differentiate sounds, specifically vowels. Thus, it is
13
different from general listening comprehension capacity which refers to the ability to
obtain meaning from spoken language.
One study that deals with the relation between listening capacity and the perception
of L2 vowels was conducted by Iverson and Evans (2007). The aim of their study
was to find whether learners with a larger first-language vowel system differ on the
cues they use to learn the English vowel system. Even though the main focus of their
study was not on the relation between listening capacity and vowel perception, it
considered auditory traits like formant frequency and duration and the impact they
have on the learning of the English vowels by learners with different first-languages.
One of the main results of Iverson and Evans’ study was that the patterns of formant
movement in English vowels are important to L2 listeners when recognizing them.
This is also true for native speakers of English who seem to rely on acoustic
information more than on abstract representations of the sounds when categorizing
vowels. From this it can be assumed that the role played by vowels’ acoustic cues
in the recognition and categorization of sounds is of utmost importance and hence,
the role that listening capacity has on the general perception of vowels must be
equally relevant.
Another important finding for the purpose of the present thesis that can be
highlighted from Iverson and Evans’ study is that listeners learn the phonetic details
of a vowel all at once rather than only learning the cues that seem best for
distinguishing categories. This can be an explanation to the way L2 learners learn
vowels in terms of phonetic details, and how this information continues to be relevant
to them at the moment of perceiving vowels during the rest of their lives. It can be
assumed that this phonetic information from the vowels is useful for communication
at the moment of perceiving vowels, but just as long as the listening capacity is
present to be able to hear said information.
In a study conducted by Feyten (1991) the relation between listening skill and several
other second language aspects is demonstrated. The aim of her study was to find
whether more attention needs to be paid to the listening skills in the preparation of
foreign language students and, more relevant for this study, whether the listening
14
skill could be a good predictor of language achievement. The results of Feyten’s
study showed that there was a relation between listening ability and overall foreign
language proficiency; there was also a relation between listening ability when
correlated with foreign language comprehension and with oral proficiency skill
(Feyten, 1991). These findings support the importance of the listening skill
(perception) as a factor affecting the general improvement of English language
proficiency and more specifically, the way it affects the perception of English vowels,
which is the focus of this thesis project.
Regarding aspects such as years of study, formal or informal study of the language,
exposure to the language among others, learners’ language experience was
considered as another linguistic factor, referring to characteristics of the learning
process that an individual has during his/her life. There are some models to explain
the influence of experience and the way a person learns a second language sound
system. In this section the two most influential approaches will be presented.
15
Speech Learning Model
Flege (1995) proposed a model based on his research that differs in several ideas
with Best’s model (PAM). The Speech Learning Model (SLM) states that if the target
phoneme in the L2 is similar to a phoneme in the L1 vowel system, the learner will
have problems learning it. The author suggested that the learner will probably
perceive that the L2 phoneme is a realization of the L1 category. This model predicts
that speech sounds are specified in long-term memory representations called
“phonetic categories”, which are used to process speech. These phonetic categories
and the mechanism used to learn the native categories last during the whole life,
and those same mechanisms are used when learning new categories. When new
sounds are learned, existing categories evolve to adapt themselves in order to cover
the L1 and L2 sounds. This is why bilingual learners struggle to define or differentiate
the sounds when learning them. In the context of Spanish speakers with Flege’s
model it could be predicted that learners will have problems with English /ɪ/ and /i:/
because they share some features of the native Spanish category /i/.
First of all, motivation has been defined as those factors that energize behaviour and
give it direction (Hilgard, Atkinson, 1979). The relation between motivational factors
and proficiency in second language has been a widely covered topic by authors all
around the world. Gardner, in his 2007 manuscript on motivation and second
language acquisition reviewed his model and compared it with later studies on the
topic. He begins by highlighting the importance of motivation on second language
16
acquisition, and argues that this factor is much more complex than just wanting to
learn the language (Gardner, 2007). Motivation is one of the three main aspects of
his model, alongside with Integrativeness (also called Openness to cultural
identification) and attitudes towards the learning situation. The motivation aspect of
his model refers to the behavioural, cognitive, and affective components that move
the learner to keep improving his/her proficiency in an L2. Integrativeness or lately
called Openness to cultural identification refers to the idea that a learner of a second
language will be influenced by the target culture of the said language. Finally, the
attitudes towards the learning situation refer to those aspects that are directly related
with the environment or the way in which the second language is learned.
Other authors have proposed that motivation is a more changeable and dynamic
variable in learning a second language (Dörnyei, 2005). A model with this idea is the
one proposed by Dörnyei and his L2 Motivational Self-System. This approach was
created considering that the concept of integrativeness proposed by Gardner and
Lambert in (1959), and supported by several authors in the following years, was not
updated. Dörnyei thought that this concept lacked a connection with the new
cognitive motivational concepts emerging from motivational psychological studies.
The main concept this approach provides is that learners have an image of
themselves that they compare with two other images: the Ideal self and the Ought-
to self. The Ideal self refers to the attributes and knowledge that the person would
like to possess. In the L2 Motivational Self-system, it refers to the level of proficiency
in the language that a person would like to have. The Ought-to self refers to the
attributes and knowledge the person considers he should possess. This approach
refers to the level of proficiency or knowledge of the language that a person
considers should have. These two concepts were based on the psychological theory
of the ideal self and the Ought-to self (Markus & Nurius, 1986). From this, it can be
assumed that the learners’ Ideal-self will be one with a better pronunciation (native-
like), thus, providing a hypothetical link between motivation and pronunciation.
Both approaches mentioned previously are similar in the sense that they consider
motivation as a key factor. These approaches and their implications played an
17
important role while creating the tool used to measure motivation and at the moment
of analysing the results searching for the possible effects that motivation has on the
perception of English vowels.
Considering the difficulties that Spanish speakers may have in perceiving English
vowels, two factors related to vowel perception were considered. The first is duration
of a sound which influences the perception and the categorization of L2 vowels. The
second factor included was the typical confusions in the perception of vowels by
Spanish speakers, due to the smaller vowel system this language has.
2.3.1 Duration
One of the factors that affects the perception of vowels for non-native speakers of
English is duration. Cebrian (2006) tested a group of adult speakers of Catalan with
varying levels of English and their experience with the language. The main objective
of his research was to study the effect of L2 experience on the categorization of
vowels, and the ability to distinguish the tense-lax vowel contrast. The results
obtained showed that native speakers of English focus more on the vowel quality
whereas non-native speakers tend to extend the duration of vowels, in order to
differentiate each of them with the vowels of their mother tongue. It was also
demonstrated that the ability to categorize vowels in the target language seems to
be downgraded depending on the age a person starts learning another language.
Their results suggested that the vowel categorization capacity may be improved by
being exposed to the target language, and being able to use the L2. Therefore,
exposure and use of the L2 play an important role in the process of learning the
sounds of another language.
18
2.3.2 Typical confusions for L1 Spanish speakers
Iverson & Evans (2009) proved that individuals with smaller and larger L1 vowel
inventories may learn L2 vowel systems likewise. This was demonstrated by testing
the perception of English vowels to a group of Spanish speakers (with a small L1
vowel system) and to a group of German speakers (with a larger L1 vowel system).
The results showed that the Germans were much better than the Spanish after a
short period of training. However, the Spanish speakers improved as much as the
Germans after additional sessions of training. The findings of their research suggest
that a larger vowel category may help in the process of learning an L2, and in case
of Spanish speakers, training improves identification of L2 vowels.
One of the confusions made by the Spanish speakers was that they tended to
assimilate L2 vowels into the same L1 categories e.g. the English vowels / ɪ / and /
i: / and the Spanish / i / (Escudero & Boersma, 2004; Flege et al., 1997; Iverson &
Evans, 2007). In addition to this, individuals with a smaller vowel system, such as
Spanish, tend to use less dimensions to distinguish L1 vowels, considering other
aspects in order to differentiate L2 vowels such as diphthongs or nasal sounds
(Iverson & Evans, 2007). In this thesis, these results will be used as parameters of
comparison for confusions in the perception of English vowels.
When talking about the perception of vowels, it is also necessary to discuss the
relation between perception and speech production, being both components of
human communication. Historically, language research has split into these two
different fields; psychologists and psycholinguistics worked on problems of phoneme
perception while phoneticians covered articulation and speech acoustics (Casserly
& Pisoni, 2010). The main reason for the separation of both fields is that their
research procedures and study methodologies differ greatly from one another. By
doing so, researchers have been allowed to comprehend this highly complex and
unique behaviour that is human speech by dividing it into manageable chunks.
19
However, Casserly and Pisoni predicted that in order to gain a full understanding of
human speech it will be necessary to reunite both concepts of perception and
production in a joint approach to common problems. It is both logical and necessary
to consider in this thesis the link between perception and production given that they
are the main components of the speech chain as proposed by Denes and Pinson
(1963).
Even if this experiment is rather recent, research on this kind of phenomena goes
back as far as 1911, with Etienne Lombard and his “Lombard Speech phenomenon”
20
(Lane & Tranel, 1971). His rather early experiments showed that there is a relation
between what a person hears and produces by modifying the context with
background noise. The speakers raise or lower their volume of speech according to
the level of background noise. They even modify the way of articulating speech to
make it more understandable under these conditions in real time. This modification
of articulation was predicted in 1990 by Björn Lindblom and his “Hyper- and Hypo-
articulation theory”. This theory postulated that individuals modify their speech in
accordance to two conflicting forces: economy of effort, which leads the speaker to
hypo-articulate; and communicative contrast, which leads the speaker to hyper-
articulate in order to be more understandable under certain conditions. This theory
also adds evidence to the existence of a direct real-time relation between perception
and production.
Considering the existence of this link, there is a study that takes this a step forward.
In 2004, a group of researchers suggested that the link between perception and
production could provide future therapeutic treatment to stuttering (Stuart,
Kalinowski, Rastatter, Saltuklaroglu & Dayalu, 2004). They found evidence that if
21
people who stutter perceive voluntary stuttering (syllabic repetition) before talking,
they tend to reduce significantly the stuttering in their speech. With these findings it
is clear that the active relation between perception and production not only exists,
but also is of great importance for the future of the field and for the understanding
and improvement of human communication. This could also be supported by
evidence of the effect of training on perception and the way it impacts production in
learners.
Okuno and Hardison (2016) conducted a study in which they demonstrated that
training individuals to perceive duration (in this case Japanese vowel duration) has
an impact on their production. There are, however, several studies that attempted to
prove the relevance of the link here discussed and failed to find strong empirical
evidence. For instance, Hattori and Iverson (2010) conducted a study with Japanese
learners of English on their perception and production of the English /r/- /l/ contrasts.
They were not successful in finding a strong relationship between these two
processes, suggesting that they seem to be more autonomous than they are thought
to be. As a counterargument for the existence of a strong link between perception
and production, the evidence provided by Smith and Hayes-Harb in their study of
German learners of English could be mentioned (Smith & Hayes-Harb, 2008). They
found moderate correlation between some acoustic measures of the German
learners’ production and their accuracy in perception of the English language. Even
if they failed in finding evidence to support the link, they agree that there is a
connection, and that it is important to define it.
Furthermore, the relation between perception and production is relevant for the
purpose of language teaching and learning. When talking about learning a new
vowel system, it is logical to consider a relation between what can be heard or
listened to and what will be produced. As an example, Flege’s Speech Learning
Model (1995) can be considered. In his study Flege states that Spanish speakers
with no experience in English tend to identify the English vowel /æ/ as a realization
of the Spanish /a/. If they never establish a new category for the new vowel, they will
not produce the vowel accurately. However, if they establish a new category for the
22
new vowel, separating it from the Spanish /a/, the learner will produce it accurately,
making clear differences from the /a/ category in their vowel system. From his model
it can be assumed that if a learner perceives a vowel sound and categorizes it
correctly, the vowel that the learner will produce will be more accurate, thus
suggesting a link between perception and production in the learning of vowels.
Consequently, there are several studies that demonstrate a relation between
perception and production, and the way the two capacities may work which are worth
considering for the future of language research.
There are two main ways in which pronunciation of English as a second language is
taught nowadays. One way is the analytic-imitative approach which depends on the
learner’s ability to imitate rhythm and sounds and relies on the availability of media
resources with pronunciation models. A second way is the analytic-linguistic
approach, which uses information and tools such as phonetic alphabet, articulatory
descriptions, among others (Celce-Murcia, Brinton & Goodwin, 1996). These two
approaches are meant to complement each other in the teaching of pronunciation.
In the following paragraphs some of the most relevant approaches to teach
pronunciation will be revised, excluding those approaches that did not consider
pronunciation as relevant as grammatical accuracy or listening comprehension such
as the grammar translation method and the reading-based method.
During the 1970s two methods considered pronunciation in different ways, one of
them was the Silent way. This approach relied on the student's ability to develop
their pronunciation skill from the instructions given by the teacher. The instructions
were not oral, as the name of the approach suggests, but relied on articulatory
demonstration and several charts and tools to teach pronunciation with little oral
input from the teacher. The other approach from this period was the Community
language learning which relied on tape recorders and a teacher who acts as model
of the correct pronunciation. This method uses mainly repetition and drill based on
the requirements of the learner.
Considering the evidence provided by all these studies, to be on the right track to
teach pronunciation, classes need to include input from various speakers of different
24
gender, perceptual and production training and feedback. Ultimately, perceptual
training has proved to be beneficial for L2 learners.
Cenoz and Garcia-Lecumberri (1999) also tested the improvement made by Spanish
and Basque L1 speakers, all of them university students, after training in terms of
English vowel perception. The mean obtained after a few hours of training was 6.6
points higher than the mean obtained before training. Their study also showed that
the difficult sounds were the ones that presented the most important improvements,
being this finding of great pedagogical relevance because it provides evidence of
the positive effects of training on students.
Nishi and Kewley-Port (2007) suggested that training on larger sets is more effective
overall than concentrating on only the most difficult vowels. According to Iverson and
Evans (2009), with training, a subject can apply their vowel category knowledge to
natural variable speech without changing their knowledge, making the categorization
process more efficient and long lasting. In the same paper, they suggested that an
efficient training should have highly variable stimulus sets, because this way the
process of applying categories to real speech is trained. In the same line, the use of
High Variability Phonetic Training (HVPT) has been proved to be highly efficient on
25
training perception of English sounds (Thompson, 2012). This kind of training is
relevant because it provides the learner a variety of stimuli including a wide scope
of natural speech produced by speakers differing in gender, dialect, voice quality,
etc. In his study, Thompson tested learners of English with Mandarin as their L1. The
use of a computer based training combined with HVPT allowed the participants to
improve their perceptual ability and also maintain it through time.
With this evidence in mind it can be concluded that there are great benefits in training
speech perception. It can be also assumed that these benefits can be increased with
the use of computer based training (Thompson, 2012) considering that the new
generations are born surrounded by technology. This kind of training offers the
teacher the possibility to focus on specific problems, help students with different
capacities and provide accurate and effective feedback. These are two essential
learner’s needs that teachers can not normally cater for in the classroom.
Summary
Several factors have been presented in this chapter that affect a learner’s perception
of speech and more specifically of the English vowels. These factors are linked with
a variety of aspects of learning experience and to different skills such as learner’s
ability to discern minimal pairs or their English proficiency level. As a whole, the
umbrella term that covers the perception of vowels and the factors affecting it in the
classroom is pronunciation.
26
Furthermore, the relation between perception and production is a matter of
discussion and research as suggested previously in this chapter. It is for this reason
that problems regarding perception and production need to be addressed in Chilean
classrooms. By doing so, students could improve their pronunciation skill as a whole.
Then again, the solution comes in the form of training. Specifically, listening training
with a variety of input that could also allow the student to practice their perceptual
ability and, as a next stage, to practice their production.
Finally, future research on this matter could focus on clarifying the relation between
perception and production, and to link the different challenges in perception of L2
sounds with non-linguistic factors such as the number of students per classroom, the
material available for training and teaching in general, and the use of technology in
in the EFL classroom.
27
CHAPTER III: METHODOLOGY
The purpose of the present thesis was to find some influential factors that may affect
the perception of English vowels in advanced learners of English as a second
language. In this section the methodology used in this thesis is presented, including
the participants tested and the testing material applied.
3.1 Participants
The participants of this thesis were 17 fourth-year students of the English Teacher
Training Program at Universidad de Concepción (7 male and 10 female). All the
participants were tested in a computer’s lab in this institution. No hearing problems
were reported.
28
Fig. 3.1 Proficiency Test. Retrieved from Englishtag.com
To determine the ability to discern minimal pairs, the Oxford Placement Test 2 (Alan,
1992) was applied to the participants, in order to measure their listening capacity.
This test consisted of 100 sentences. All the participants were given this test after
the Vowel Identification Test. A percentage was calculated based on the number of
correct answers per participant.
29
3.2.3 Vowel Identification Test
Stimulus
In this test, a set of 84 audios of 11 vowels of English (/æ/, /ʌ/, /ɑ:/, /ɪ/, /i:/, /e/, /ɜ:/,
/ɒ/, /ɔ:/, /ʊ/, /u:/) in the context of /bVt/ or /hVd/ words were used. Non-words were
not included in the list (e.g. a word with the pronunciation /bʊt/ was not found). These
audios were recorded at University College London by two female and two male
speakers of Standard Southern British English (Pereira Reyes, 2013). Another list of
words containing the same 11 English vowels was recorded by a native female
speaker of English with the same accent to be used as examples in the practice
phase. For this, 11 words from Well’s (1982) lexical sets were used. The audio files
were obtained using a digital recorder (Philips voice tracer DVT2700) and then
presented in the Vowel Identification Test using the “TP - Perception Tests /
Perceptions Training Tasks” software. (Rauber, 2013). Results are stored in an
Excel document which can only be accessed by the researcher.
30
Procedure
For this test the participants were presented with 95 audio files of 5 speakers saying
words that contained the 11 English vowels chosen. The words were randomized by
the software. Once the audio was presented, 11 response buttons were shown on a
computer screen with one-syllable words. These response buttons used words
containing the English vowels being tested (cat, cup, card, sit, pet, seat, word, pot,
caught, put, food). A practice phase was given to the participants before the main
test. The participants took from 10 to 30 minutes to complete the test. The
participants did not receive any feedback on their answers.
31
Fig. 3.4 End of the Vowel Identification Test.
3.3 Survey
Apart from the tests given to the participants, an online survey (appendix 7.5) was
applied to this group, covering aspects of the participants’ experience with the
language, some items on their motivation for learning English and the habits they
have regarding English input. The participants received a link to this survey through
email and filled it in before taking the tests.
For the statistical analysis and graphs, SPSS software (23rd version) was used.
Pearson-moment correlations were run between the vowel identification test and the
listening test, the vowel identification test and the proficiency test, the vowel
identification test and listening input, and finally the vowel identification test and the
32
self-assessment as measured in the survey. Additionally, a Repeated-measures
ANOVA was run for the results of the Vowel Identification Test. A T-test was run in
order to compare the results of the different vowels. Finally, the resulting data from
the test was tabulated in a confusion matrix for their analysis.
33
CHAPTER IV: RESULTS
In this chapter, the results of the statistical analysis run for the data are presented.
IBM SPSS statistics analysis software (v23) was used (see 3.3). In the first section,
the general results from the different tests are shown. After that, there are specific
analysis which are relevant for the present thesis.
Fig. 4.1 shows boxplots for the perception of 11 English vowels used in the Vowel Test.
34
In a more specific analysis, table 4.1 shows the mean results per vowel in the vowel
identification test, and also the highest and lowest scores per vowel. The highest
score was 74.88 for the /iː/ vowel, and the lowest score was 31.73 for the /ʊ/ vowel.
FRONT
CENTRAL
BACK
Table 4.1 shows the mean score per vowel and the highest and lowest scores, including tense-lax
distinction and their articulatory position.
35
Additionally, in order to find whether there was any significant difference between
the tense-lax pairs, a one-sample T-test was used for all the pairs of vowels (Table
4.2). Only the vowels /ʊ/ and /u:/ showed a significant difference (p < 0.05)
Table 4.2 shows a T-test analysis between pairs of vowels, and their differences regarding tense-
lax and/or lax-lax distinctions.
36
4.1.2 Listening Test
To measure the participants’ listening capacity, a listening test was applied to them.
The test had 100 items, the scores were transformed into percentages (1 percentage
point per item). The group overall mean in the listening test was 78.18 % (SD 8.73).
11 participants were above the mean. The highest score reported was 91% and the
lowest score was 59%.
Fig.4.2 Graph shows variability and individual performance in the listening test.
37
4.1.3 Language proficiency test
To know the participants’ level of English, a language proficiency test was applied.
The test had 50 items, each item was given a point and scores were transformed
into percentages. The group overall mean was 90.12 (SD 8.73), with the majority of
participants being above the mean. The highest score was 98% and the lowest score
was 76%.
Fig. 4.3 graph shows variability and individual performance in the English proficiency test.
38
4.1.4 Survey
To gather information about habits regarding use of English and listening, a survey
was given to participants. From a general point of view, most of the participants use
their English more than an hour outside classes on an average day.
Fig. 4.4 shows the amount of hours participants use English outside classes on an average day.
Considering participants’ listening skills, all the participants use them for fun. In terms
of improving their level of English and also teaching, most of the participants use
their listening skills at least once a week.
39
As for participant’s self-assessment on the basic skills: speaking, reading, writing
and listening, results show a different evaluation depending on the skill. Specifically,
in terms of listening, participants expressed their self-assessment with the highest
scores.
Fig. 4.6 shows the participants’ self-assessment considering speaking, reading, writing and
listening.
Listen to media
21% 17%
Look up phonemic
transcription
21% Drilling
41%
Speaking with
others
40
As for participants’ strategies to learn the pronunciation of words, the preferred
strategies were listen and repeat and use of dictionaries to check pronunciation.
There was also a considerable amount of participants reporting they do not have any
strategy to learn the pronunciation of new words.
No strategy
32% 27%
Look up phonemic
transcription
Speaking (saying
out loud)
14% 27% Listen and repeat
41
4.2 Specific results
The most common confusions in the perception of English vowels were between the
tense and lax vowels. Table 4.3 shows a selection of the first and second most
common confusions (in columns) observed in this test.
Table 4.3 shows the most common confusions in the Vowel Identification Test.
42
4.2.2 Correlations
To find whether there was any relation between the participants’ capacity to perceive
English Vowels (M 58.3, SD 9.8) and their listening capacity as tested in the listening
test (M 78.2, SD 8.7) a Pearson-moment correlation test was run. Results showed
there was a moderate correlation between English vowel perception capacity and
listening for minimal pairs capacity: r: .565*, N: 17, p<0.05.
Fig. 4.9 shows a scatter plot for Pearson-moment correlation test between English Vowel
identification test and the Listening for minimal pairs test.
43
Vowel identification test and language proficiency test
To find whether there was a relation between the vowel identification test (M 58.3,
SD 9.8) and the language proficiency test (M 90.12, SD 6.9), a Pearson moment
correlation test was run. Results showed there was no significant correlation
between participant’s vowel identification capacity and the participant’s level of
English: r: .324, N: 17, p > 0.05.
Fig. 4.10 shows a scatter plot for Pearson-moment correlation test between English Vowel
identification test and the language proficiency test.
44
Vowel identification test and listening input
A Pearson moment correlation test was run between the results from the Vowel
identification test (M 58.3, SD 9.8) and the results from the survey regarding the
participant’s listening input (M 80, SD 10) (4.3). The Listening input was calculated
by transforming the scores for the frequency of use of their English for listening into
percentages (5 = 100%, 4 = 80%, 3 = 60%, 2 = 40%, and 1= 20%). Results showed
there was no significant correlation between participant’s vowel identification
capacity and the participant’s listening input: r: .126, N: 17, p > 0.05.
Fig. 4.11 shows a scatter plot for Pearson-moment correlation test between English Vowel
identification test and the participant’s listening input.
45
Vowel identification test and participants’ self-assessment
A Pearson moment correlation test was run between the results from the Vowel
identification test (M 58.3, SD 9.8) and the results from the survey regarding the
participant’s self-assessment (PSA) of their level of English (M 82.6, SD 9.7). PSA
was calculated by transforming the scores for the participants’ self-assessment of
their skills into percentages (5 = 100%, 4 = 80%, 3 = 60%, 2 = 40%, and 1= 20%).
Results showed there was no significant correlation between participant’s vowel
identification capacity and the participant’s self-assessment: r: .204, N: 17, p > 0.05.
Fig. 4.12 shows a scatter plot for Pearson-moment correlation test between English Vowel
identification test and the participant’s self-assessment.
46
CHAPTER V: DISCUSSION
From a general point of view, the resulting data from the vowel identification test
(VIT) showed that the majority of participants obtained an overall score above the
group mean. Considering that the participants are advanced learners of English, this
finding is in line with Cebrian’s study (2006) which suggested that the vowel
categorization capacity improves with exposure to the target language.
From a more specific point of view, the analysis of the data from VIT regarding the
difference between the identification of tense and lax vowels showed that tense
vowels were apparently easier to perceive. Participants obtained a higher score for
these vowels. It may be argued that duration seems to have an impact on the
perception of English vowels, showing that lax vowels with shorter duration seem to
be harder to perceive. These results are similar to findings in previous studies
showing that duration is an important cue when perceiving English vowels (Cebrian
1991; Iverson and Evans 2007).
Additionally, the analysis of the VIT’s results showed the most common confusions
between vowels. Minimal pairs of vowels (tense-lax pairs) were the most difficult to
discern. This finding is in line with Flege’s Speech Learning Model (1995), which
suggested that L2 vowels that are similar to the L1 sounds will be troublesome to
recognize. Some other studies have also predicted this outcome (Escudero &
Boersma, 2004; Flege et al., 1997; Iverson & Evans, 2007)
The main objective of this thesis was to find the relation between different factors
(linguistic and non-linguistic) with the capacity to perceive English vowels. For this
purpose the results from the VIT and from the Listening test were analysed. The
analysis showed that there is a relation between the participants’ listening capacity
and their vowel identification capacity. This result is in agreement with previous
studies regarding the importance of listening capacity as a predictor for language
proficiency level (Iverson & Evans, 2007; Feyten, 1991). More important, these
results suggest the influence that the listening skill may have on the perception of
English vowels.
47
Regarding the relation between language proficiency and vowel identification
capacity measured by VIT, no relation was found, unlike in Feyten’s study (1991)
which suggested that the general listening skill was related with the general
proficiency level of English as an L2.
Regarding more qualitative data, the results from the survey provided two variables
to test as factors that may have an impact on the perception of English vowels. The
first one is the amount of time the participants spend listening to English (Listening
input). The data showed no relation between this factor and the capacity to identify
English vowels. It might be the case that someone who spends more time listening
to English does so because he/she thinks needs more input. Additionally, this person
may have low scores in the VIT. Thus, the correlation analysis would not show a
direct relation. It could be worth adding an extra question to find out the motivation
behind the amount of hours spent on listening then.
A similar case was found for the analysis of the self-assessment of the participants’
English skills (Listening, Speaking, Reading and Writing) and the possible relation
with the vowel identification capacity. The lack of relation between these two
capacities may suggest that Dörnyei’s (2005) Motivational Self-system has no
implication regarding motivation and vowel perception capacity for this group of
participants. However, from this item of the survey it can be concluded that their
listening capacity plays an important role on their self-evaluation. The majority of
participants declared to feel confident about their listening skill and that they
understand spoken English fluently.
Finally, when participants were asked about the strategies they use to learn
pronunciation and what activities they do for this purpose, most of participants
mentioned they use some sort of drilling or repeating exercise. This evidence is in
line with the “auditory sensorimotor adaptation” phenomena as proposed by
Villacorta et al (2007). This phenomena could explain the preference of the
participants to repeat and drill new words in order to improve their pronunciation.
In the future, the importance of the perception of English vowels and listening
capacity for minimal pairs could be considered in the creation of future English
48
curriculum for Chilean classrooms. The results of this study could be used for the
creation and design of training material using a computer based platform with high
variability input (stimulus) and could be beneficial for teachers and students when
aiming at improving English pronunciation. More research on this topic is needed.
49
VI. REFERENCES
Casserly, E. D., & Pisoni, D. B. (2010, 07). Speech perception and production. Wiley
Cebrian, J. (2006, 07). Experience and the use of non-native duration in L2 vowel
doi:10.1016/j.wocn.2005.08.003
University Press.
50
Cenoz, J., & Lecumberri, L. G. (1999). The Effect of Training on the Discrimination
Denes, P., & Pinson, E. (1963). The Speech Chain. Garden City, NY: Anchor
Press/Doubleday.
Derwing, T. M., & Munro, M. J. (2005, 09). Second Language Accent and
Escudero, P., & Boersma, P. (2004, 12). Bridging the Gap between L2 Speech
10.2307/328825
51
Figuereido, Rauber, Rato & Santos (2013).(version 3.1) [Software], : Worken.
Flege, J., (1995). "Second language speech learning: Theory, findings and
Flege, J. E., Bohn, O., & Jang, S. (1997, 10). Effects of experience on non-native
437-470. doi:10.1006/jpho.1997.0052
Gilman, R. A., & Moody, R. L. (1984, 09). What Practitioners Say About Listening:
Research Implications for the Classroom. Foreign Language Annals, 17(4), 331-334.
doi:10.1111/j.1944-9720.1984.tb03236.x
from http://www.gavo.t.utokyo.ac.jp/L2WS2010/papers/L2WS2010_P2-04.pdf
Iverson, P., & Evans, B. G. (2009, 08). Learning English vowels with different first
language vowel systems II: Auditory training for native Spanish and German
doi:10.1121/1.3148196
Iverson, P., & Evans, B. G. (2007). Learning English vowels with different first-
doi:10.1121/1.2783198
Lambacher, S. G., Martens, W. L., Kakehi, K., Marasinghe, C. A., & Molholt, G.
(2005, 04). The effects of identification training on the identification and production
53
Lane, H., & Tranel, B. (1971, 12). The Lombard Sign and the Role of Hearing in
doi:10.1044/jshr.1404.677
Spoken English, TESOL and applied linguistics (pp. 245-270). Palgrave Macmillan
UK.
Levis, J. M., Sonsaat, S., Link, S., & Barriuso, T. A. (2016). Native and nonnative
894-931.
2037-8_16
Markus, H., & Nurius, P. (1986). Possible selves. American Psychologist, 41(9), 954-
969. doi:10.1037//0003-066x.41.9.954
Morley, J. (1994). Pronunciation pedagogy and theory: New views, new directions.
54
N., & Guasti Maria, T. (n.d.). Focus-Stress Alignment and its Consequences for
https://www.rivisteweb.it/doi/10.1418/7558
Nespor, M., & Guasti, M. T. (2002). Focus-stress alignment and its consequences
Nishi, K., & Kewley-Port, D. (2007, 12). Training Japanese Listeners to Perceive
doi:10.1121/1.2178720
Pereira, Y., & Hazan, V. (2013). Impact of different training modes on the perception
55
Smith, B. L., & Hayes-Harb R. (2008). Speech Production and Speech Perception
Stuart, A., Kalinowski, J., Rastatter, M. P., Saltuklaroglu, T., & Dayalu, V. (2004, 01).
speech of people who stutter: Initial fitting and 4‐month follow‐up. International
10.1080/13682820310001616976
doi:10.1121/1.2773966
56
VII. APPENDIX
Practice Stage
Dress
Fleece
Foot
Force
Goose
Kit
Lot
Nurse
Start
Strut
Trap
Test
Bart
Bat
Beet
Bert
Bet
Bit
Boot
Bot
Bought
Butt
Had
57
Hard
Head
Heard
Heed
Hid
Hoard
Hod
Hood
Hudd
Who’d
58
7.2 List of audio files used in the vowel identification test
Practice stage
Dress_SSO_1.wav
Fleece_SSO_1.wav
Foot_SSO_1.wav
Force_SSO_1.wav
Goose_SSO_1.wav
Kit_SSO_1.wav
Lot_SSO_1.wav
Nurse_SSO_1.wav
Start_SSO_1.wav
Strut_SSO_1.wav
Trap_SSO_1.wav
Test
Bart2f2.wav
Bart3_m1.wav
Bart4f1.wav
Bart4m2.wav
bat1f1.wav
bat2f2.wav
bat2_m1.wav
bat3m2.wav
beet1m2.wav
beet4f1.wav
beet4f2.wav
beet4_m1.wav
Bert1f1.wav
59
Bert1f2.wav
Bert2m2.wav
Bert2_m1.wav
bet1f2.wav
bet1_m1.wav
bet2m2.wav
bet3f1.wav
bit1f1.wav
bit1m2.wav
bit4f2.wav
bit_m1.wav
boot2f2.wav
boot3f1.wav
boot3m2.wav
boot3_m1.wav
bot1_m1.wav
bot3f1.wav
bot4f2.wav
bot4m2.wav
bought1f1.wav
bought1_m1.wav
bought2m2.wav
bought3f2.wav
butt1f1.wav
butt1m2.wav
butt1_m1.wav
butt4f2.wav
had1f1.wav
had3f2.wav
had3_m1.wav
60
had4m2.wav
hard1f1.wav
hard1f2.wav
hard1m2.wav
hard1_m1.wav
head1f2.wav
head3f1.wav
head3m2.wav
head3_m1.wav
heard2f1.wav
heard2_m1.wav
heard3m2.wav
heard4f2.wav
heed2f2.wav
heed3f1.wav
heed3_m1.wav
heed5m2.wav
hid1_m1.wav
61
7.3 Results per participant
62
7.4 Individual results per stimuli in the vowel identification test
63
7.5 Survey
64
65
66
67
7.6 Listening Test
68
69
70
71
7.7 List of participants
N° Participant Code
1 F1
2 F2
3 F4
4 F5
5 F6
6 F7
7 F8
8 F9
9 F10
10 F11
11 M1
12 M2
13 M4
14 M5
15 M7
16 M8
17 M9
72
7.8 Poster Presented at RICELT 2017
73