Masaryk University Faculty of Arts

Department of English and American Studies

Teaching English and Literature for Secondary Schools

Bc. Kristýna Zemková

Segmental versus Suprasegmental Mistakes in English Pronunciation Master’s Diploma Thesis

Supervisor: PhDr. Kateřina Tomková, Ph.D. 2018

I declare that I have worked on this thesis independently, using only the primary and secondary sources listed in the bibliography.

…………………………………………….. Author’s signature

Acknowledgement I would like to thank my supervisor, PhDr. Kateřina Tomková, Ph.D., for her help with the recordings that I could not have done without, as well as for her guidance throughout the writing process.

Table of Contents

Table of Contents ...... 4 List of Figures ...... 6 Introduction ...... 7 1. The History of Teaching Pronunciation ...... 9 2. Contrasting Theories...... 10 3. Segmental Pronunciation Features ...... 15 3. 1 Vowels ...... 15 3. 2 ...... 21 4. Suprasegmentals ...... 25 4. 1 Stress ...... 25 4. 2 Assimilation ...... 27 4. 3 Elision ...... 29 4. 4 Linking ...... 29 4. 5 Rhythm ...... 30 4. 6 Sentence Stress ...... 31 4. 7 Intonation ...... 31 5. Greater Importance for Intelligibility: Segmentals or Suprasegmentals? ...... 32 6. The Current Status of Pronunciation in Schools ...... 34 7. The Theory of Teaching Pronunciation ...... 37 7. 1 Teaching Segmentals ...... 37 7. 2 Teaching Suprasegmentals ...... 39 7. 3 Error Correction ...... 40 7. 4 Testing Pronunciation ...... 41 8. Research methodology ...... 42 9. Questionnaires analysis ...... 50 9. 1 Characteristics of the Research Participants ...... 50 9. 2 Recording A ...... 50 9. 2. 1 Native speakers ...... 50 9. 2. 2 Speakers of Asian ...... 51 9. 2. 3 Speakers of Romance native languages ...... 52 9. 2. 4 Speakers of Slavic languages ...... 52 9. 3 Recording B ...... 53 9. 3. 1 Native speakers ...... 53 9. 3. 2 Speakers of Asian languages ...... 54

4

9. 3. 3 Speakers of Romance languages ...... 55 9. 3. 4 Speakers of Slavic languages ...... 55 9. 4 Recording ...... 56 9. 4. 1 Native speakers ...... 56 9. 4. 2 Speakers of Asian languages ...... 56 9. 4. 3 Speakers of Romance languages ...... 56 9. 4. 4 Speakers of Slavic languages ...... 57 9. 5 English as Lingua Franca ...... 57 10. Overall Results Analysis ...... 58 10. 1 Suprasegmental Mistakes ...... 59 10. 2 Segmental Mistakes ...... 60 10. 3 The Differences Between Language Groups ...... 61 10. 4 Implications on Teaching Pronunciation ...... 63 Conclusion ...... 65 References ...... 67 Other sources ...... 68 Summary ...... 70 Summary in Czech ...... 71 Appendices on CD ...... 73

5

List of Figures

Figure 1 - Types of interaction (Levis, 2005) ...... 12 Figure 2 - Primary cardinal vowels by Daniel Jones (Ladefoged, 1975) ...... 16 Figure 3 - Differences in English and Czech vowels (Ondráček, 2014) ...... 17 Figure 4 - Questionnaire ...... 47 Figure 5 - Intelligibility chart ...... 58 Figure 6 - Perceived level of English chart ...... 58 Figure 7 - The most native-like recording ...... 59 Figure 8 - The recording the most pleasant to listen to ...... 59 Figure 9 - The recording with the most mistakes ...... 59

6

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

Introduction

The methods of teaching pronunciation have been discussed in the community of

ESL teachers for the past hundred years; more so with the development of various English language teaching methods in the second half of the twentieth century. In some methods, pronunciation plays a crucial role; in others (mostly obsolete), it is entirely disregarded. Even after having established that pronunciation is an element of language acquisition that is not be overlooked, other questions remain pertinent, like the issue of the goals of teaching pronunciation, the extent to which pronunciation is teachable and the extent to which it should be taught. It is also important to determine what pronunciation features are crucial for the learners to acquire. The first question itself divides the theorists and language learners in two groups: those who believe that the goal is a native-like pronunciation, and those who say that we should strive for mutual intelligibility. The second question raises the issue of a critical age hypothesis (the age after which, according to some, it is not possible to gain native-like pronunciation), as well as the issue of the learner’s personal preferences and motivation. The third question, regarding the importance of various pronunciation features, is the main topic of this thesis. The research carried out for this thesis will try to determine whether it is segmental1 or suprasegmental2 pronunciation features that are more important for learners to acquire. To further delimit the research area of the thesis, the focus will be on

Czech-specific mistakes, i.e. the mistakes that are particular for Czech ESL learners. The aim is based on the intelligibility principle: the goal is to find out what kinds of mistakes Czech

ESL learners need to avoid to in order to be understood by their interlocutors. Because the presumption is that some elements of perception may be variable according to the native language of the interlocutor, interlocutors of many native languages will be asked to participate. For a general theoretical background, the research will first briefly introduce the

1 Individual phonemes. 2 Prosodic. 7

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION approaches towards pronunciation teaching in history and discuss some key contemporary questions regarding pronunciation teaching. A comparative analysis of Czech and English will follow. The analysis will be divided in two large sections: one focusing on segmental and one on suprasegmental features. In each section, English and Czech phonological features will be presented and compared, aiming to determine the areas of

English pronunciation which are difficult for Czech ESL learners. In order to find out how pronunciation is taught in Czech schools, a little online survey will be carried out concerning textbooks used in class. The most common textbooks will then be examined in terms of their approach towards teaching pronunciation. A chapter on the theory of pronunciation training will present traditional and modern methods of teaching segmental and suprasegmental pronunciation features. A research will follow, with the goal of determining what group of pronunciation features is more important for intelligibility: whether it is segmentals or suprasegmentals. It will also study whether this depends on the interlocutor’s native language.

To be able to examine this, three recordings will be made. The first one will be segmentally correct but contain suprasegmental mistakes. The second one will contain segmental mistakes, but no suprasegmental issues. Due to the accuracy required, these two recordings will be made by a phonetician. The third recording will be made by an authentic Czech ESL learner. These three short recordings will then be given to four groups of people. Native speakers of English, speakers of Asian languages, speakers of Romance languages and speakers of Slavic languages, with the assumption that their assessment of the recordings might differ in some respects. Questionnaires will be given out to the participants with the aim of finding out how intelligible each recording is. Other aspects focused on will be the impressions the recordings make on the interlocutors and the assessment they give to the speaker’s English based on their pronunciation. A more detailed analysis will examine which specific pronunciation mistakes are most perceptible by the interlocutors and which go unnoticed. The respondents’ personal written evaluations will also be studied to find out

8

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION their views of ESL learners’ pronunciations. The results will be followed by a suggestion of what implications they may have on ESL pronunciation teaching.

1. The History of Teaching Pronunciation

The particularities of teaching pronunciation are still debated in TESOL focused journals. The attitude towards teaching pronunciation has been changing with the overall teaching approaches, but there are still a few critical points connected to it on which the field specialists do not agree. It is the goals themselves of pronunciation instruction that come into question, as well as methods leading towards achieving those goals. Historically, pronunciation has become prominent with the audio-lingual method and situational language teaching, both of which considered it very important (Morley, 1991, p. 484). These approaches have used articulatory explanations, imitation and pattern drills, with emphasis on correction. As Morley (1991) points out, while these approaches still exist, they have abandoned the method of articulatory explanations and replaced it by a more communicative and functional method. In the following eras, namely the 1960s, the importance of teaching pronunciation started being questioned, including the issue of whether it is directly teachable at all. The result is that many curricula put less emphasis on pronunciation or drop it entirely

(Morley, 1991, p. 485). An example of this is the cognitive approach that ephasizes grammar and vocabulary rather than on pronunciation (Celce-Murcia, Brinton & Goodwin, 1996, p.

5). It becomes clear that the previous methods of teaching pronunciation have become obsolete, but there are not any reasonable alternatives yet. The issue of teaching pronunciation is discussed theoretically in the 1970s and, much more extensively, in the

1980s. The latter period can be characterized by a new wave of interest in pronunciation teaching, influenced by the notions explored theoretically in the previous decade. One of the methods that rediscover the importance of pronunciation training is the communicative approach. Because of its focus on effective communication, the method stresses the need to help the students achieve a certain level of pronunciation skills; a threshold above which the

9

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION communication is not threatened by misunderstandings caused by an incorrect pronunciation (Celce-Murcia, Brinton & Goodwin, 1996, p. 9).

2. Contrasting Theories

Nowadays, it becomes apparent that it is necessary to revisit the fundamental question of the goals of teaching pronunciation. Furthermore, the issue needs to be approached by research and not by intuition only. What emerges is the contrast between the nativeness3 and intelligibility principle. According to Levis (2005), “nativeness” was the dominant paradigm in the early days of language teaching, which was later disproved by research showing that native-like pronunciation is usually not attainable after the critical age period (p.370). This period concerning pronunciation is placed to puberty by Flege. He explains that “adults usually cannot learn to speak a foreign language without an accent because the central nervous system undergoes some permanent reorganization at about puberty, thereby distinguishing adults and adolescents from younger learners” (Flege, 1981, pp. 443 – 444). Despite a considerable amount of evidence supporting the existence of age after which gaining a native-like pronunciation is highly unlikely, the authenticity principle is still very much present in contemporary language teaching. Research among Canadian ESL teachers has shown that an “overwhelming majority considered speaking with perfectly native pronunciation to be the desired goal” (Derwing & Munro, 2005, p. 384). It also manifests in the accent reduction industry, which promotes the goal of “perfecting” one’s accent to the point of sounding like a native speaker of a chosen language variety (Levis,

2005, p. 370).

The intelligibility principle, in its term, is concerned with the communicative aim. It is based on a belief that an imperfect pronunciation is not a problem as long as the utterance is intelligible for the interlocutor without a significant effort on the listener’s part.4 The

3 A native-like pronunciation. 4 The term “comfortable intelligibility” is also used. 10

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION overall goal of pronunciation instruction in a modern communication-based class can therefore be set as an improvement in the learner’s intelligibility. Nevertheless, the learner’s motivation is also a factor that must be considered. Research has shown that highly motivated individuals with a particular aptitude can sometimes achieve native-like pronunciation patterns (Derwing & Munro, 2005, p. 384) and they should not be denied this opportunity if this is their goal in language learning. However, this kind of demand from the students can come from ignorance of the limits of pronunciation teaching, and it may easily cause frustration of both the teacher and the learner. It might therefore be advisable to discuss this issue openly with learners past the critical age who are determined to gain native- like pronunciation.

Deciding on the role model for pronunciation teaching is another matter. A clear majority of textbooks use solely the or in their listening exercises. In reality, most native speakers of English speak neither of the varieties. The learners can therefore face problems with understanding some other widely spoken pronunciation varieties, such as Australian English or some dialects from the south of the US. Fortunately, some textbook publishers are aware of this issue, and there are now books available with authentic recordings made by various kinds of English speakers (for example the Listening series by Collins English for Life). As for choice of the model pronunciation itself, it is important to keep in mind that it should serve solely as a model and not as a strict norm to follow. The RP seems like a reasonable choice for the reasons provided by Dalton and Seidlhofer (1994). Firstly, it is the most thoroughly described accent which makes it easier for the learners to access materials that will help them with their pronunciation. Secondly, being a prestige language variety, the RP tends to be “more widely accepted in a wider range of communicative situations, while “non-prestige accents are often regarded as “odd” or idiosyncratic” (Dalton & Seidlhofer, 1994, p.7).

11

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

As Morley (1991) points out, much like authenticity, intelligibility is also a “slippery concept” (p. 499). In any research, subjects who are asked to judge the intelligibility of a speaker are also influenced by other factors such as their individual ideas of what makes an accent attractive, or the personality of the speaker. The author even goes as far as warning about possible prejudices or xenophobia. Derwing and Munro (2005) share this view and add the issue of the interlocutors’ experience with accented speech as a possible influence on the research results (p. 381). These are factors that need to be considered in any research in this field.

Nevertheless, there is another aspect in favour of the intelligibility approach. It is the fact that English is no longer spoken only by native speakers, but it has become a lingua franca instead. That means that it is used as a language of communication between individuals who do not share a native language. Levis (2005) illustrates this in his “Speaker-Listener

Intelligibility Matrix” (p. 372). It demonstrates that there are four types of interactions in total, as demonstrated by the following table.

Figure 1 - Speaker-listener intelligibility matrix (Levis, 2005)

This thesis will only deal with the bottom part of the table, i.e. a non-native speaker speaking with two types of interlocutors: native and non-native. Quadrant C corresponds to the speaker being a regular ESL/ELF learner, whereas quadrant D describes an ELF5 learner.

5 English as a lingua franca. 12

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

English has, in fact, more second language speakers than it has native speakers. The implication this has on teaching pronunciation is that there are now groups of learners whose goal is not at all interaction with native speakers of English, but a general ability to communicate with people from other countries. The goal for those learners is not authenticity, but rather mutual intelligibility with speakers from various linguistic backgrounds. An avid promoter of this theory is Jennifer Jenkins who has written several books on this issue. She strongly opposes teaching RP as the pronunciation norm. She argues that not only very few English people speak RP, but the dialect is also going through many changes not reflected in the teaching materials so that the students end up learning obsolete pronunciation habits. An example of that can be the disappearing in the word poor

(Jenkins, 2000, p. 15). What she proposes instead is a system of Lingua Franca Core – pronunciation features which are necessary for learners to acquire to be able to communicate effeciently with other non-native speakers of English. She supports the claim by citing the examples of L2 norms that are emerging in countries like Sierra Leone or Kenya (Jenkins,

2000, p. 29). In practice, this approach can be illustrated on the following quote: “When learners do not perceive that their continued use of a particular phonological transfer causes their pronunciation to be unintelligible to their EIL6 interlocutors, they are unlikely to be motivated to make the effort of replacing it simply in order to approximate more closely to the accent of an absent native speaker. Nor should they feel any obligation to do so” (Jenkins,

2000, p. 120). It is important to note that Jenkins’ approach faces criticism for several reasons. Some are worried that English is negatively influenced by these “compromises”.

More realistic reproaches mention the issue of perception: the fact that the interlocutors make assumptions about the speaker based on the accent the speaker has. Jenkins is not concerned with this: she believes that the accent, albeit a foreign one, belongs to one’s identity and that as long as intelligibility is assured, it can be kept however strong it is. While

6 English as an international language. 13

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION this may be true for users of English as a lingua franca, it may not be the case of some ESL learners. Morley (1991) discusses the influence of a foreign accent on the social status of certain groups of people such as immigrants who have resided in an English-speaking country for 5 to15 years (p. 490). Those people can increase their chances of finding good jobs by improving their accents. An example of a job demanding good pronunciation skills would be jobs that demand public speaking, such as university and other lecturers. Another exception would be students pursuing a degree in teaching English as a second language who should have good pronunciation to be good role models for learners. Nevertheless, Morley

(1991) generally agrees with the intelligibility principle: he distinguishes six levels of “student communicability” (p. 502), with two communicative thresholds, one between levels 2 and 3 and the other between 3 and 4. According to the intelligibility principle, learners should aim at least towards level 4, where “sound and prosodic variances from NS norm are obvious

[but] listeners can understand if they concentrate on the message” (Morley, 1991, p. 502).

The interference is primarily at the distraction level.

Another interesting approach towards this issue is the question of identity mentioned above. That is the notion that an accent constitutes a part of one’s identity and therefore should not be changed past the point of intelligibility unless the learner wants to. However, the identity question can also be in favour of learning pronunciation according to the authenticity principle. If the learner wants to blend in with a particular social group, they may want to acquire the accent of this group. Dalton and Seidlhofer (1994) even believe that a learner who has adapted their accent to a certain native speaker community can become its

“honorary member” and therefore not to be seen as an outsider anymore, with all the advantages that come with it.7 (p.7) However, having an accent that corresponds to the learner’s level of competence can also have positive implications: it signals to the interlocutor

7 The authors also mention the other possible outcome, that is being criticised for approximating too closely to the pronunciation of a community and apparently claiming membership of a group without being properly qualified; that can result in being seen as an intruder. 14

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION that the person may require modified input. In case of learners with a near-native accent, an analogical problem can occur: if the interlocutor cannot recognize that the speaker is non- native, any mistakes the learner makes will be judged as if the person were native, which can lead to being misjudged as having gaps in education.

If a language learner and their teacher determine that the goal of teaching pronunciation is comfortable intelligibility, the next point of interest are the areas of pronunciation themselves that the student needs to focus on in order to meet their goal.

There are many aspects of pronunciation, and they can be divided into two major groups: segmental and suprasegmental pronunciation features. The following chapter will serve as an overview of these features, comparing English and Czech phonology in the process.

3. Segmental Pronunciation Features

The segmental pronunciation system of a language deals with phonemes which are distinctive segmental entities. They are one of the smallest units of speech that make one word different from another. Phonemes are distinguished by distinctive features, i.e. the presence or absence of a certain quality. There are two types of phonemes: vowels and consonants8. The following chapter will present the characteristics of Czech and English vowels, describe their differences, and pinpoint areas which are difficult for the intelligibility of Czech learners of English who have yet to master English pronunciation.

3. 1 Vowels

Vowels, according to Pennington’s Phonology in English Language Teaching (1996), “can be distinguished definitionally from consonants as having a relatively open aperture of articulation” (p. 89). Cruttenden (2008) also stresses the absence of any closure or narrowing that would “result in the noise component characteristic of many consonantal sounds” (pp.

32 – 33). Vowels are also characterised by a great sonority and an ability to be prolonged.

For this reason, it is vowels only that form the nucleus of a syllable in English. Czech is

8 Some sources also distinguish semi-vowels or semi-consonants. 15

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION different in this regard, as the consonants \r\9, \l\, and less frequently \m\ can also become the nucleus of a syllable: these are called syllabic consonants. Examples of words containing syllabic consonants can be vlk (wolf), krk (neck) or osm (eight). This difference between the languages affects native speakers of English who are learning Czech but does not have a real impact on Czechs learning English.

One distinction between individual vowels from an articulatory point of view is the shape and the position of the tongue, which determines whether the vowel will be front, central or back. According to the amount of opening of the jaw in the production of the vowel, we distinguish close, half-close, half-open or open10. All primary cardinal vowels can easily be placed on a chart, which has been devised by the British phonetician Daniel Jones.

Figure 2 - Primary cardinal vowels by Daniel Jones (Ladefoged, 1975)

(Ladefoged, 1975)

A third way of classifying vowels is according to the shape of the lips in the moment of articulation. Is this respect, vowels can be divided between spread or unrounded, like /i/, and rounded, like /u/. While Kučera (1961) classifies all vowels between these two categories

(which is valid for Czech), Pennington (1996) describes neutral English vowels such as the schwa. Generally, these three types of qualities are “unrelated to particular values in languages” (Cruttenden, 2008, p. 35).

9 The text between the backslashes is in Czech format – i.e. it is written using the Czech alphabet and therefore it has its sound values (a method adopted from Ondráček). 10 Alternatively called high, mid-high, mid-low and low. 16

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

As Roach explains in English and Phonology (1983), the primary cardinal vowels are “the vowels that are most familiar to the speakers of most European languages” (p. 13).

Thus, the chart above would be a basic overview of both English and Czech vowels.

However, as Cruttenden (2008) points out, it is wrong to claim that every native speaker’s pronunciation of a vowel will have the same quality due to many accents and individual differences. The following observations and statements will therefore take into consideration the RP as the model for English and the Standard Czech.

Thanks to modern sound mapping devices, phoneticians have been able to place the vowels of each of the European languages on such chart to demonstrate that, for example, the phoneme /a/11 can have different allophones12 in different languages. A chart demonstrating the differences in Czech and English vowels can be found in Ondráček’s

(2014) work (p. 33):

Figure 3 - Differences in English and Czech vowels (Ondráček, 2014)

11 The text between these marks is in the IPA format – i.e. it is written with the phonemic alphabet and therefore has its sound values (a method adopted from Ondráček). 12 Allophones are concrete realisation of phonemes. 17

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

The chart shows that while the Czech \ú\ and the English /u:/ are both back close vowels, their realisations are not the same: in fact, the English /u:/ is more close and less back than the Czech equivalent. Such distinction can be observed on every single vowel.

Consequently, the more significant the difference there is between the vowels, the more

“foreign” the speaker will sound if they employ the phoneme of their mother tongue. For some phonemes the difference is not big enough to cause any confusion: for example, the

Czech \o\ and the English /ɒ/ (pronounced in standard RP) will sound very similar.

However, the chart shows that English employs some vowels that do not have any close equivalent the Czech phonetic system. An example of this can be the English vowel /æ/ which is an open front vowel. If a Czech native speaker is not taught to pronounce this vowel correctly, they can resort to pronouncing it as the most similar Czech vowel they can find, the Czech \e\. However, given that this vowel is much closer than /æ/ and more similar to the English /e/, it can easily result in confusion. For instance, a Czech speaker may tend to pronounce the words bad and bed in the same way (only with a difference in length, or sometimes even without). Analogously, a Czech speaker can have problems with the recognition of words that contain a schwa, as this vowel does not occur as a phoneme in the

Czech language. It only exists as “hesitation sound” in a Czech regular speech and Czech native speakers might not even be aware of using it. Krčmová (2008) only mentions its usage for alphabet recitation, where it enables better identification of the .

Quantity, also known as the length of the vowel, is also a feature of both Czech and

English vowels. However, it does not have the same function in the two language systems.

Krčmová (2008) states that Czech uses a “relative quantity” (p.68), which is the ratio between a short and a long version of an otherwise similar vowel to distinguish semantic units. Czech vowels therefore exist in pairs. Each vowel has a short and a long equivalent, enabling the interlocutor to distinguish minimal pairs by the length only. Research has shown that although the length of the vowel can vary according to its surroundings (Krčmová, 2008, p.

18

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

68) (just like in English), such differences are usually not perceptible in regular speech.

English, on the contrary, has three lengths of vowels: long, semi-long and short. As explained in Ondráček’s (2014) work: “All English vowels change according to whether the succeeding consonant is fortis or lenis.” (p.35). For example, the vowel in bead is longer than the one in beat. Preceding vowels, however, do not affect the length. Quantity is not the primary cause of problems in intelligibility between Czech ESL learners and other speakers of English, but it can accompany other problems such as final consonant devoicing. For example, the pair beat - bead differs in the final consonant as well as in length in English, whereas a Czech native speak might wrongly pronounce both words as /biːt/.

Semivowels (also called gliding vowels or glides) are a particular group of phonemes that share certain characteristics with both vowels and consonants. English has two semivowels: /j/ and /w/, which Cruttenden (2008) describes as being “purely vocalic from the phonetic standpoint.” (p. 93). However, they “function very much as if they were consonants, marginally rather than centrally in a syllable, and […] tend to be voiceless in words such as queen or tune, which is phonetically characteristic of a consonant”

(Cruttenden, 2008, p. 94). The only has one semivowel, which is the phoneme /j/. According to Kučera (1961), “In casual speech, /i/ may occur where /j/ is distributionally expected. On the other hand, /j/ may substitute for /i/ in slow and solemn speech.” (p.29). The semivowel /w/ does not exist in Czech, which causes Czech speakers to approximate it as the phoneme /v/. Interesting is also the phenomenon of hypercorrection, which causes Czech speakers to pronounce /w/ where in fact the correct pronunciation is /v/. An example of this can be the pronunciation of very well as [ˈweriˈwel].

The Czech language, according to Krčmová (2008), only contains the so-called false , where the first half of it is fully vocalic and the second half becomes a semi-

19

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION vowel. She further divides diphthongs between rising and falling13. In a rising diphthong, it is the beginning of it that is weakened (these do not occur in Czech but are common for

Slovak, for example in the word mlieko). In a closing diphthong, the second part is weakened

(for example the ou in the word louka). Only falling diphthongs are natural for the Czech language: \au\, \ou\, and \eu\. The \aj\, \ej\, \ij\, \oj\ and \uj\ sounds are not diphthongs per-se, but Ondráček (2014) explains that they function similarly to the English diphthongs such as /ɔɪ/. However, he also states that the Czech \oj\ sounds much more strongly at the end.

English is generally richer in diphthongs: the RP has 8 in total, 3 of which are centring

(/ɪə/, /eə/ and / ʊə/) and five are closing (/aɪ/, /eɪ/, /ɔɪ/, /aʊ/ and /əʊ/). This overview demonstrates that diphthongs are a feature that Czech learners of English can have problems with, as most of them do not have an equivalent in Czech. As mentioned above, a Czech speaker has to soften their pronunciation of the diphthong /ɔɪ/ to avoid saying the Czech

\oj\. As Ondráček (2014) observes: “The English boy will seem unfinished in Czech ears, whereas \boj\ will seem to be overdone or exaggerated in English, even though at first hearing they sound almost identical” (p.37). Diphthongs containing a schwa are also troublesome in the RP. Czech ESL learners therefore usually resort to pronouncing /r/ in words that would end in a centring diphthong in the RP (in words like here or there). Similarly, they replace the diphthong /əʊ/ with the diphthong /ɔʊ/ (Zemková, 2016). Luckily, this does not have to become a problem, because these two alternatives belong to a standard

General American pronunciation, so these alterations in the speech of an otherwise British- sounding non-native speaker are not likely to cause confusions.

Unlike Slovak, Czech does not contain any triphthongs. English has five of them:

/aɪə/, /eɪə/, /ɔɪə/, /aʊə/ and /əʊə/, used in words such as fire, hour, etc. They are also

13 Krčmová uses the term “falling” instead of “closing” but the English-language sources prefer the term “closing”. It is the same phenomenon, only a different terminology. 20

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION tricky for Czechs to pronounce because of the schwa that they all contain. English native speakers often reduce these triphthongs into diphthongs using smoothing, which means the omission of the central element. However, since the schwa is still the phoneme that stays, the pronunciation problem is analogical to the one explained in the section on diphthongs.

3. 2 Consonants

Both Czech and English have 24 consonant phonemes (Kučera, 1961, p.30).

However, both languages contain phonemes that do not exist in the other language. They also have phonemes which do not have precisely the same characteristics in the two languages (for example, the phoneme /p/ is never aspirated in Czech). They can, therefore, present a challenge for Czech learners of English. In articulatory terms, Pennington defines consonants as “sounds made with closed or nearly closed articulations” (Kučera, 1961, p.37), although this definition is somewhat simplified. Regarding phonology, they are segments that occur at the edges of syllables (Cruttenden, 2008, p.25) (except for the Czech syllabic consonants). Consonants can be classified according to several characteristics, namely the place of articulation, the manner of articulation, and voicing.

The place of articulation “is defined by the location of an active articulator functioning alone or in concert with another articulator” (Pennington, 1996, p.38). Generally,

Czech and English share these characteristics. /p/, /b/ and /m/ are bilabial consonants; their primary articulators are the lips. /f/ and /v/ are labiodental, articulated by the lower lip and upper teeth. The first difference comes with dental consonants; Czech has none,

English has /θ/ and /ð/. When pronouncing them, “the tongue tip and rims articulate with the upper teeth” (Cruttenden, 2008, p.27), which is quite easy to demonstrate for English teachers. Nevertheless, Czech learners can have a problem using these phonemes in free speech. The most common examples of wrong pronunciation of /θ/ are /t/ and /f/

(therefore [fɪŋk] or [tɪŋk] instead of [θɪŋk]). Less common with Czechs, but also employed,

21

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION is /s/ [sɪŋk]. The voiced equivalent of the consonant, /ð/, is often replaced by /d/ ([dɪs] instead of [ðɪs]). Analogically, /ð/ is less frequently replaced by /z/ [zɪs]. As Derwing and

Munro (2005) point out, many pronunciation problems can be rooted in perception issues

(p.388). It may therefore be advisable to spend some time training students to distinguish between /θ/, /ð/, /t/ and /d/ to be able to pronounce them well.

Alveolar consonants are created when the blade, or the tip and the blade, articulate with the alveolar ridge (these can be further divided between apico-dental and apico-alveolar

(Kučera, 1961, p.30)). Consonants that belong to this category in both Czech and English are: /t/, /d/, /l/, /n/, /s/, /z/. Czech has supplementary 3 alveolar consonants: \l\, \ř\ and \r\, which differs from the post-alveolar English /r/. The /r/ of the RP is pronounced by the tip of the tongue articulating with the rear part of the alveolar ridge. Theoretically,

Czech speakers should be prone making the mistake of pronouncing the Czech alveolar \r\ instead of the English post-alveolar one. However, the teaching practice of the author of this thesis shows that many Czech learners learn the correct pronunciation of the /r/ rather quickly, although they are often not consistent in its usage. Furthermore, there are various variants of the pronunciation of the /r/ among native speakers of English of different dialects, so the presumption is that an incorrect pronunciation of the /r/ is not likely to cause major problems in intelligibility. Other palato-alveolar (or post-alveolar14) consonants are

/ʃ/, /ʒ/, /tʃ/ and /dʒ/ for both languages.

When the front of the tongue articulates with the hard palate, palatal consonants are made. Czech and English only share the /j/, but Czech has three more:

\ť\ \ď\ and \ň\. Velar consonants are articulated by the back of the tongue and the soft palate. English and Czech share three velar consonants (/k/, /g/ and /ŋ/, Czech also has

/x/, which manifest in the text as “”. Even though the phoneme /ŋ/ exists in both Czech

14 Terminology used by Ondráček. 22

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION and English, Czech learners still can have problems with its pronunciation. The problem consists in the fact that /ŋ/ only exists in the middle of phonetic words in Czech (such as in the name Lenka), but never at the end. Czech learners of English therefore often pronounce the final /g/ in words ending with a g (like sing – [sɪŋg]). Even if they are aware that they should not pronounce the final g, they may know how to pronounce the nasal n correctly.

One remaining consonant is //: it is glottal, which means that it is pronounced by means of an obstruction or a narrowing that causes friction but not vibration between the vocal folds (Cruttenden, 2008, p.28).

The manner of articulation determines how much obstruction there is between the articulators, ranging from a total obstruction to friction only. The consonants shared by

Czech and English usually share the manner of articulation. (/p/, /b/, /t/, /d/,

/k/, /g/), also called stop consonants (Pennington, 1996, p.45), are characterised by a complete closure in the vocal tract, followed by a sudden release of the air pressure. Even though they share these characteristics in Czech and in English, they can differ by the presence of aspiration, which is not present in Czech phonology. Aspiration is a burst of noise followed by “a period during which air escapes through the vocal chords, making a sound like /h/” (Roach, 1983, p.30). Only voiceless consonants /p/, /t/ and /k/ are aspirated in English. The aspiration depends on the position of the vowel in a word.

Aspiration only occurs either at the beginning of a stressed syllable (apart pronounced with an aspirated p as [əˈphɑːt]), or at the beginning of a word, regardless of the stress. It is important to note that if the consonant is preceded by the phoneme /s/, it is not aspirated.

Czech speakers tend to forget to aspirate or do not know how to do it, which has implications on intelligibility. Roach (1983) explains that if a native speaker hears a word with unaspirated

/p/, /t/ or /k/ in a place where it should be aspirated, they will mistake it for /b/, /d/ or

/g/, because “it is aspiration, not voicing, which distinguished the initial /p/, /t/and /k/ from /b/, /d/ and /g/.

23

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

Fricatives are produced by air escaping through a small passage that creates a hissing sound (Roach, 1983, p.37). Just like vowels, they belong to the category of continuants, which means that you can make them without interruption as long as you have air in your lungs.

/f/, /v/, /s/, /z/, /ʃ/, /ʒ/, /θ/, /ð/ are all . are a mixture between plosives and fricatives in that they begin as plosives with a complete closure in the mouth, but end as fricatives. The affricates shared by Czech and English are /tʃ/ and /dʒ/. Nasal consonants are distinguished by a complete closure in the mouth in combination with an airstream escaping through the nose. /m/, /n/ and /ŋ/ are nasal consonants.

Finally, are characterised by a narrowing in the mouth which is not narrow enough to cause friction. Being continuant and friction-less, approximants share characteristics with vowels. An shared by English and Czech is /j/. The English

/w/ is also an approximant, and the English /r/ as well; the Czech /r/ is, however, a trill.

Unlike the standard RP /r/, a trilled r causes the airstream to vibrate. The Czech trilled r is similar to the Scottish allophone. The consonant /l/ is a an approximant called a lateral approximant. A partial closure with an airstream escaping on both sides of the closure allows its pronunciation.

Another distinction between consonants is their either voiced or voiceless quality.

Many of them, especially in the category of plosives, fricatives and affricates, exist in pairs.

Voiced and voiceless consonants are distinguished by the vibration of the vocal chords, or the lack thereof. Even though the Czech and English consonants share their voiceless or voiced quality, their usage can be an issue for Czech learners in one particular case. English does not employ final consonant devoicing but Czech does. Czech learners therefore often mistakenly devoice final consonants; in words like pig; [pɪɡ] becomes [pɪk]. Because these consonants often function as minimal pairs, such pronunciation can lead to a semantic confusion: a native English speaker can think that the word pronounced is pick. Such problems can arise with each minimal pair: leave x leaf or belief x believe for instance. It is also

24

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION worth noticing at this point that many misunderstandings between Czechs and the speakers of other languages are caused by a combination of wrongly pronounced consonants and vowels in one word. For example, if the lack of distinction between /æ/ and /e/ and final consonants devoicing are combined, the words bat, bet, bad and bed can sound exactly the same. The interlocutor then has to rely on the context of the utterance, which does not have to be so clear with lower-level learners of English, especially when they also have problems with suprasegmental features.

4. Suprasegmentals

Czech and English does not differ only in individual phonemes but also is various suprasegmental elements. Suprasegmental pronunciation features are what distinguishes human speech from artificial speech of machines. Even though there are robots who can produce very “real” speech on segmental level, it still does not sound natural to a human ear, especially in longer utterances. It is because suprasegmental features are also an important part of pronunciation. Phonologically, they are “sound contrasts over several segments”

(Roach, 1983, p. 37) These prosodic features form connected speech and are not any less important for intelligibility. Suprasegmental pronunciation features can be divided into two categories: at the word level, we distinguish word stress, reduction, assimilation, elision, and linking; features at the clause level comprise of rhythm, sentence stress, and intonation.

4. 1 Stress

Stress has distinct functions in Czech and English. English, being a stress-timed language, gives more importance to stress than Czech, which is syllable-timed. Generally,

English and Czech stress differ in “strength, placement and function” (Ondráček, 2014, p.

39). Although the concept of stress is easily explainable to English language learners, its phonological description is more difficult15. First, it can be described from either a

15 The term stress itself is not unequivocal: some authors, like Palková (1994) or Cruttenden (2008), distinguish between stress and accent. Cruttenden also distinguishes sonority. 25

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION production or a perception point of view. Production-wise, research has shown that the speaker uses more muscular energy for stressed than for unstressed syllables (Ondráček,

2014, p. 39). From the point of view of perception, there are more factors to consider. Firstly, the stressed syllables are perceived as louder than the unstressed ones. Secondly, the pitch of stressed syllables distinguishes them from the rest. Thirdly, vowel duration is a key factor in indicating stress. In English, vowels are stressed syllables are longer than those in unstressed syllables. This is, however, not true for Czech, where the syllables are long or short regardless of them being stressed or not. Lastly, the quality of vowels also changes. Collins and Mees

(2008) demonstrate this on the example of the words present (noun; [ˈprez(ə)nt]) and to present

(verb; [prɪˈzent]) (p.124). Although having the same written form, because of a different stress placement, the words differ in two vowels. The vowels that occur in unstressed syllables the most are /ɪ/, /ʊ/ and /ə/ (Roach, 1983, p. 73). This feature, called the vowel reduction, does not exist in Czech, as the unstressed syllable does not experience any change in vowel quality. Vowel reduction plays an important part in the rhythm of speech, which will be discussed later.

As far as stress placement is concerned, Czech is a language with a fixed stress. That means that it has a “constant relationship” with a structural element of a word, in case of

Czech with the boundaries between words. (Palková, 1994, p. 157). The Czech stress is fixed on the first syllable of a word, except for words preceded by a preposition, where the stress moves to the preposition which phonetically functions as a whole with the lexical words. In

English, the stress can be found on any syllable. Monosyllabic word can also be either stressed or unstressed (Ondráček, 2014, p. 40). The placement of English stress, however, is much more complicated. Roach (1983) states that “many writers have said that English stress is so difficult to predict that the best approach is to treat stress placement as a property of the individual word, to be learned when the word itself is learned” (pp. 75 – 76). However, that does not mean that English word stress in unpredictable: when a native speaker comes

26

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION across un unfamiliar word, they are usually able to place the stress correctly. The rules of stress placement are quite complex and have many exceptions. Generally, they depend on the number of syllables the word has, the presence or absence of suffixes and prefixes, the parts of speech, the character of vowels, etc.

As Ondráček (2014) points out, the English stress can help to differentiate nouns from verbs, it can help the interlocutor reconstruct unstressed syllables unconsciously in a flow of speech, and is an essential foundation for sentence stress (p. 40). In Czech, word stress mainly helps the interlocutor to determine individual words in an utterance.

All of this knowledge indicates that word stress and vowel reduction will be an important issue in the intelligibility of Czech learners of English, as they have to acquire an entirely different system of word stress than their native language has, and it is omnipresent in every utterance. It is not difficult to determine that the main problem consists in Czech speakers wrongly putting stress on first syllables of words. Thus, other ESL learners and mainly native speakers can have troubles making out the individual words in the speech.

4. 2 Assimilation

The changes that occur in a pronunciation of a word based on the interactions with its neighbouring words are called assimilation. It is a variable pronunciation feature: the amount of assimilation can depend not only on the speaker, but also on the speed of the speech (assimilation is more likely to occur in quick speech). The degree of assimilation can also vary. Assimilation exists in both Czech and English, but it has different characteristics in each of them. Assimilation can be divided into several types, for example according to the direction. A regressive (Roach, 1983, p. 105), or leading (Collins & Mees, 2008, p.116) assimilation is created when the first phoneme is influenced by a phoneme that comes after it, e.g. white pepper, pronounced [waɪt ˈpepə], becomes [waɪp ˈpepə], or the Czech word shoda

(agreement) pronounced as \zhoda\. Progressive (lagging) assimilation is analogical but in the opposite direction. An English example of progressive assimilation would be the

27

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION expression on the house pronounced as [ɒn nə ˈhaʊs] instead of [ɒn ðə ˈhaʊs]. According to

Krčmová (2008), progressive assimilation is rare in Czech.

The second type of distinction is according to the area of influence. The literature on both English and Czech phonology distinguishes place assimilation, energy assimilation and manner assimilation. Assimilations caused by the place of assimilation can cause for example alveolars to become bilabials, velars or palatals. In Czech, kotník (ankle), \kotník\, can become \koťník\; in English, wet blanket, [wet ˈblænkɪt], can become [wep ˈblænkɪt].

Replacing for example a by a nasal or a lateral is called assimilation of manner, which is usually non-standard in Czech. Collins & Mees (2008) give the example of the expression fail the test, the pronunciation of which changes from [feɪl ðə ˈtest] to [feɪl lə ˈtest]. Czechs can assimilate for example the word kratší (shorter): \kratʃi:\ to \kratʃI:\. The final type of influence is the energy assimilation (also called the assimilation of (Ondráček, 2014, p.42)). This type causes the fortis and lenis phonemes to change into the opposite category, for example I have to pronounced as [aɪ hæf tə]. According to Collins & Mees (2008), this type of assimilation is rare in English (p. 117). Czech, on the other hand, often assimilates voiced consonants to voiceless. The opposite direction, i.e. a voiceless consonant becoming voiced, is unacceptable in English, but frequent in Czech (for example pes byl, pronounced as \’pez ‘bil\ instead of \pes bil\. In conclusion, even though assimilation is present in both languages, it is by no means used similarly. Czech speakers can therefore make the mistake of assimilating phonemes which cannot be assimilated in English. However, not assimilating in cases a native speaker would is not considered a problem and it can only cause the non- native speaker to sound foreign, which the supporters of the concept of English as a lingua franca would consider perfectly acceptable. It is therefore possible to agree with Ondráček’s

(2014) conclusion, which suggests that learners should only be acquainted with the concept of assimilation at a stage beyond the comfortable intelligibility. First, they have to master the

28

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION accuracy and the clarity of a normal English speech before they can practice assimilation of a usually quicker speech.

4. 3 Elision

Elision is the phenomenon of a deletion of a phoneme. Collins & Mees (2008) distinguish historical and contemporary elision (p. 118). Historical elision is fixed in the present day. For example, the word know is now only pronounced without pronouncing the phoneme /k/, a pronunciation that was created by a gradual process of elision.

Contemporary elision is optional: for example, the word tasteless can be pronounced as

[teɪstləs] or [teɪsləs]. Elision in Czech is not so frequent, but can be found in declinations of names. For example, the surname Šimek becomes Šimka in the accusative. In this case, the elision was istorically followed by the reform of the written form. The implications of elision are the following: lower-level learners of English have to be taught the obligatory elided pronunciations, such as in the words know, comb, wrong, etc. It is probably not necessary to explain elision to them; they can learn it as mere exceptions. The optional elision can fall in the same category as assimilation; learners of English below a certain level do not need to be taught to elide.

4. 4 Linking

Linking, also called liaison, is another point of difference between Czech and English.

Czech uses junctures to separate the speech in smaller units (Krčmová, 2008, p. 179).

Semantic words are usually separated by a . English does so as well, but much less: a juncture enables the interlocutor to distinguish phrases with same phonemes, such as a name and an aim. As Cruttenden (2008) remarks, “the glottal stop before a vowel beginning an accented syllable in the example [above] is optional and generally not used unless emphasis is required” (p. 307). Generally, English is a language that joins phonological words together; a smooth speech is the norm. There are three types of linking: consonant-vowel linking, a linking and an intrusive /r/, and a linking /w/ and /j/ (Ondráček, 2014, p. 43).

29

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

In an English sentence such as It is an interesting story, the first four words require to be linked, because the bordering phonemes (the first and the last one in the words) are always one vowel and one consonant. In Czech, there is no such linking; a Czech speaker of English can make the mistake of inserting glottal stops in between individual words, making the speech sound very unnatural.

The linking /r/ is a feature that causes the final /r/ to be pronounced in a word where a non-rhotic speaker would not pronounced it, should the word stand on itself. For example, the word there is pronounced [ðeə] in Received Pronunciation, but the expression there is is pronounced [ðeərɪz]. This could potentially be difficult for Czech learners of

English, but as data suggests (Zemková, 2016, p. 50), they avoid this issue by being more or less consistently rhotic, which is a perfectly acceptable thing to do. While a linking /r/ is required, an intrusive /r/ is optional. It is used for connecting two words, the first of which finishes with a vowel and the second of which begins with a vowel. For example, even if there is no reason for it from the spelling, an English speaker may include an intrusive /r/ in the expression vodka or two and pronounce it as [ˈvɒdkərɔːˈtuː]. Similarly, a slight linking

/j / can appear in vocalic junctures in expressions like my ears. The /j/, however, is not as strong as /j/ as to be confused with a phonetically similar expression my years. Finally, linking

/w/ can appear in expressions like two-eyed, again still clearly distinguishable from too wide.

While there is no reason not to explain the phenomenon of intrusive consonants to reasonably advanced learners, it is by no means required for intelligibility. The benefit of the knowledge lies in better perception skills for the learners.

4. 5 Rhythm

Rhythm is “a regular succession of weak and strong stresses, accents, sounds or movements” (Ondráček, 2014, p. 45). Regarding rhythm, Ondráček (2014) presents English as a stress-timed language. That means that the repetition of stressed syllables is regular, which makes it “somewhat analogous to the beat of music” (Ondráček, 2014, p.45).

30

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

However, he doesn’t deny that this is not completely true, as English also contains some elements of syllable-timed rhythm. This is further explained by Roach (1983). He claims that research suggests that we tend to hear speech as more regular than it really is and that English has not manifested the expected regularity. He points out that rhythmicity can vary according to the situation the speaker is in: a public speaker is likely to prepare a very rhythmical speech, whereas a person in a casual conversation will be more arhythmical. As far as Czech rhythmicity is concerned, the language traditionally belongs to syllable-timed languages

(Palková, 1991, p. 283). Roach (1983) mentions the general dependence of the English rhythm on strong and weak syllables. This leads us to sentence stress, which is closely tied with rhythm.

4. 6 Sentence Stress

While word stress is fixed and therefore always put on the same syllable, sentence stress depends somewhat on the speaker. They are allowed to put stress on, and therefore to emphasize, the part of speech that is the most important in the utterance. Czech and English have this in common, although Czech has also other means of emphasis; it allows for certain changes in the word order, which is quite flexible. English, on the contrary, relies more on sentence stress.

4. 7 Intonation

Finally, intonation is a crucial element of correct English pronunciation. Ondráček

(2014) even claims that errors in intonation are perceived as worse than errors in individual segments, because errors in intonation are not expected by the native speakers (p. 46). The term intonation is defined by Ladd (2008) in Intonational Phonology as follows: intonation is used “to convey “post-lexical” or sentence-level pragmatic meanings in a linguistically structured way” (p. 4). In a simpler manner, intonation is a variation of pitch, melody or even speed that helps to convey a meaning. For example, it is only intonation that phonetically distinguishes phrases “You did.” And “You did?”. This also illustrates the dependence of

31

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION intonation on punctuality, even though is not its only aspect. According to Gordon (2016),

“the typological study of intonation is still in its infancy”. Nevertheless, the contemporary research already works with acoustic analysis rather than the previous “impressionistic judgements”. (Gordon, 2016, pp. 243 – 244). English intonation can be divided in several categories according to the types of nuclear tones16. Cruttenden (2008) distinguishes these types: falling nuclear tones, rising nuclear tones, falling-rising nuclear tones, rising-falling nuclear tones and level nuclear tones (p. 270 – 275). Falling nuclear tones start on a high pitch and end on a low pitch. These are used for statements, wh-questions, imperative sentences and exclamatory sentences (Ondráček, 2014, p. 46). Rising nuclear tones are used for yes/no questions, in requests and for unfinished utterances. The fall-rise pattern is reserved for limited agreement that includes some form of hesitation. The rise-fall pattern indicates a strong emotion: approval, surprise, disapproval or indignation. Finally, the level nuclear pitch expresses the exact opposite: something boring or routine. Although English and Czech share basic intonation patterns (a rising intonation for some questions and requests), there are still a lot of differences, which can lower the intelligibility of Czech ESL learners. A good motivation for learners could be to expose them to English spoken by lower-level native speakers of Asian languages. Given that Asian languages often have a very different intonation, it could make it clear for them as to why it is important to focus on the issue of intonation.

5. Greater Importance for Intelligibility: Segmentals or Suprasegmentals?

Whether it is segmentals or suprasegmentals which are more important for English language learners is still debated in the TESOL community. Levis (2005) claims in his essay that “during the past 25 years, pronunciation teachers have emphasized suprasegmentals rather than segmentals in promoting intelligibility, despite a paucity of research evidence for

16 A nuclear tone is the “pitch pattern beginning at the primary accent and ending at the end of the intonational phrase” (Cruttenden, 2008). 32

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION this belief” (p. 369). The author expresses and incertitude regarding the importance of suprasegmentals in the context of English as an international language and raises the issue of learnability of suprasegmentals in comparison with segmentals. Derwing and Munro

(2008) share this opinion with Levis. They cite Levis’ (2005) research proving that native speakers only distinguish meanings in three of five intonation contours (p. 385). Morley

(1991), on the contrary, emphasises the suprasegmental approach, basing it on the works on intonation by Bolinger and the treatment of paralinguistic features by Brown (p. 493). As far as Czech speakers are concerned, Ondráček (2014) identifies these pronunciation problems: wrong word stress and a general lack of clarity in diction (suprasegmental features) and specific segmental features problems such as /ð/ replaced by /d/ and /w/ pronounced as

/v/. The last problem mentioned is “pronouncing words as if they were Czech” (Ondráček,

2014, p. 194), which can be a combination of segmental and suprasegmental errors. This debate illustrates the interest of the following research questions of this thesis: Is it segmental or suprasegmental features that are the most important aspect of teaching pronunciation to

Czech ESL learners? Is there any difference between the perception of native speakers of

English and of ESL learners? Is there any difference in the perception of ESL learners of native languages of different languages families?

It is difficult to predict the outcome for the first research question, as previous research in this area has shown both possible results. The assumption for the second research question is that there will be a difference and that native speakers will judge the speech differently than the second group of non-native speakers. The presumption for the third question is that interlocutors whose native language belongs to the Slavic language group just like Czech will face the least problems with understanding. As for subjective perception, their leniency towards pronunciation mistakes cannot be predicted with such certainty. On the contrary, according to Ondráček’s (2014) findings, it is the group of native speakers who might be the most accepting of pronunciation errors.

33

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

6. The Current Status of Pronunciation in Schools

To further analyse the issue of teaching pronunciation, it is important not to look only on theoretical issues, but also on common current teaching practices. Derwing and

Munro’s (2008) informal survey of journals devoted to teaching English shows that very little papers, are published on the issue of pronunciation teaching, compared to those published on other issues. This suggests that pronunciation is not a major focus in the community of today’s teachers, which is proved by another statistic. 67% of teachers surveyed in Canada have reported having no training in teaching pronunciation and several studies conducted in

Australia have reported that local teachers avoid teaching pronunciation because they feel that they are lacking the confidence and skills to do that. (Derwing & Munro, 2005, p. 389).

Derwing and Munro’s (2005) research as well as personal observations of the author of this thesis suggest that learners are able to identify only segmental features when asked about their own pronunciation issues, disregarding suprasegmentals completely. Another point of view on this issue is an analysis of textbooks used in Czech schools. A questionnaire posted in a Facebook group containing young people from the whole Czech Republic has revealed what textbooks are currently most used in Czech schools. 86 people participated in the survey. Firstly, they were asked to name a textbook they had most recently been working with in their English classes at school and to specify it as much as they can (for example which edition and what level). Secondly, they were asked to evaluate how good the textbook is in teaching pronunciation, taking into consideration segmentals and suprasegmentals17.

They had to place it on the following scale: very good – rather good – average – rather bad

– very bad – not able to assess. To be able to count an average evaluation, “very good” was given the value 1, “rather good” value 2 etc. Finally, they were able to put any other remarks in a box provided.

17 This was explained in a way that even general public with no knowledge of phonology could understand it. 34

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

The 86 participants have revealed that textbooks that are used the most are the following: Maturita Solutions, English File, New Headway, New Opportunities and Insight. All of these textbooks have then been studied to determine how they deal with pronunciation, focusing on either pre-intermediate or intermediate level, which were the levels most commonly indicated by the students.

Maturita Solutions, the most popular textbook according to the questionnaire, exists in multiple editions; most participants are working with the second edition. The textbook has been given the best evaluation of them all: 2.3. This is rather surprising, as the book does not seem to focus much on pronunciation. The pronunciation exercises are not put in regularly and are not indicated in the overview of skills at the beginning of the book, but there is generally at least one per unit. They are focused on segmentals, suprasegmentals and abbreviations. Some tasks manage to engage the learners and to let them figure out something for themselves, most of them simply ask them to “listen and repeat”.

Pronunciation of new vocabulary is not indicated and the teacher’s book does not remind the teacher to cover it.

The second most used textbook is English File which has received an average grade of 2.5. This textbook has a very elaborated system of pronunciation instruction; the overview of pronunciation covered in the book is even put next to vocabulary and grammar in the contents. It focuses evenly on segmentals and suprasegmentals and it teaches the learner to understand phonetic symbols. It is very visual in the explanations and contains follow-up activities. It lets the learners discover a lot of rules by themselves. Furthermore, the teacher’s book diligently reminds the teacher to drill and model pronunciation of new vocabulary. Out of the five books studied, English File Third Edition is the only one that visibly puts a lot of emphasis on pronunciation. Its only possible drawback might be a strict emphasis on RP as the only role model for pronunciation (this would be different in the American version of this book).

35

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

New Headway has received the worst grade average: 3.1, despite the fact that it does focus systematically on pronunciation. Interestingly, only suprasegmental pronunciation features are covered in the pre-intermediate textbook. The book tries to draw the students’ attention to suprasegmentals by calling them “music of English” and by explicative drawings.

Apart from this, pronunciation is not found in the table of contents and its instruction is not encouraged in the teacher’s book. It does, however, contain a vocabulary overview where the correct pronunciation of the words is indicated.

The textbook Insight, which has received an average of 2.5, has an interesting approach to pronunciation. There are no pronunciation exercises to be found in the student’s book, but it has a separate section the workbook. There, a full 5-page overview of pronunciation can be found. It is focused on both segmentals and suprasegmentals, contains a lot of good explanations and practical exercises. Vocabulary in the wordlist is also accompanied by pronunciation. The teacher’s book regularly reminds the teachers to model and drill pronunciation. Overall, provided that the teachers find the time to include pronunciation exercises from the workbook, Insight is a very good textbook for teaching pronunciation.

Finally, New Opportunities has received the second worst grade average of 3. Although pronunciation is listed in the contents and has a fixed place at the end of units, there is very little of it. There are precisely eight pronunciation exercises in the whole textbook, located at the end of each second unit. Moreover, they are, with one exception, focused only on segmental features. The teacher’s book does not incite the teachers to focus on pronunciation and the pronunciation exercises seem very out of context.

This little piece of research shows that there are very good and very bad textbooks on the market and that the textbooks currently used un Czech schools fall in both categories.

A good coverage of pronunciation in a textbook does not necessarily mean that it is taught well by the teacher or that it is not skipped, but at least it is always there for the students to

36

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION read on their own. Nevertheless, it seems to be very difficult to teach pronunciation well using some of the worse textbooks described above and it would require a teacher aware of the importance of pronunciation, and possibly some supplementary materials, to teach it well.

7. The Theory of Teaching Pronunciation

Some Czech teachers might not have a possibility to decide what textbook they will work with in class: they might be prescribed one by the school without being able to change it. Even if the textbook belongs to the lower-quality ones, it does not mean that the teacher needs to give up teaching pronunciation, as there are many reference books available instructing teachers in the field of pronunciation training.

7. 1 Teaching Segmentals

The best starting point for teaching segmentals can be the phonemic chart, as it can help the learners to identify and understand segmental pronunciation features. It can help especially visual learners, who may passively be able to perceive that there are differences in certain similar phonemes but not be able to actively produce them. With the help of a phonemic chart, the students have a visual guideline for a correct pronunciation. However, work with the phonemic chart must be thoughtfully prepared, as a table with a lot of unknown symbols could be daunting for the learners, should they be presented all its features all at once. Motivation is also the key. Celce-Murcia, Brinton and Goodwin (1996) give examples of very short activities to try with the learners to provoke their interest in the phonetic alphabet, such as explaining to them George Bernard Shaw’s famous word ghoti, pronounced [fɪʃ]. A simpler exercise can be asking the students to pronounce the word banana, pointing out the two pronunciations of the letter a, or asking the question “How do we pronounce th in English?”, followed by a demonstration of the voiceless and the voiced variant. (Celce-Murcia, Brinton & Goodwin, 1996, pp. 40-41). These exercises should raise

37

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION the awareness of the issue of pronunciation in the students and hopefully make them understand the need for a phonemic chart.

After a slow exposure to the chart, the teacher can use it for multiple purposes.

Underhill gives examples of activities that can be done with the chart. For instance, the learners can train their receptive skills by pointing out the phonemes in words said by the teacher. It can also be used as a means of error correction: the teacher can point to the phonemes that are pronounced by a learner and to those that should be pronounced and give the learner a chance to correct themselves: the advantage is that the learner instantly knows what part of the word they should focus on. Finally, the students can begin to point out what other learners have said, giving them auditory and pronunciation practice at the same time. (Underhill, 2005, pp. 101 – 105).

Apart from working with the phonemic chart, there are other practical exercises that can be done in class to help the students pronounce segmentals correctly. Dalton and

Seidlhofer (1994) argue that segmentals are the easiest pronunciation feature to teach and prove this claim by providing a lot of practical exercises (p.126). The most basic ones are imitation exercises and minimal pair drills. A more advanced type of drill is a “contextualised minimal pair drill”, in which the teacher sets the context and presents the key vocabulary, and the students respond with a meaningful sentence while practising minimal pairs18 (Celce-

Murcia, Brinton & Goodwin, 1996, p. 9). Tongue twisters and games make pronunciation training more varied. Such games can be, for example, the Odd One Out (students in groups have to race other groups in identifying a word that has a different vowel then the rest of the words), Homophone Go Fish (students look for words that sound the same), Pronunciation maze

(students connect words that contain the same sound), IPA Bingo, Minimal pair slap (students slap the card with the word they hear), or Rhyming Pair Memory Game (the students look for two words that rhyme by turning cards over, being allowed to turn only two at a time). (Pesce,

18 For example: “The blacksmith (a. hits / b. heats) the horseshoe.” A. with a hammer, b. in the fire. 38

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION n.d.). Anagrams, crosswords and hangman can also be played using IPA symbols. (Underhill,

1996, p. 204). Technology can also help: a camera can be used to record face and mouth movements during a speaking activity. The video can be replayed to the class to demonstrate the shape of mouth in the pronunciation of various phonemes. Then, it can be replayed without the sound for the students to try to lip read. (Underhill, 1994, p. 203). Celce-Murcia,

Brinton and Goodwin (1996) suggest using some less-known techniques: In “kinesthetic reinforcement” (suitable for kinesthetic learners), the students and teachers use hand signals and body movements; for example, placing a finger on the nose to indicate a nasal sound or indicating the length of a vowel by showing a space of appropriate length between the thumb and the index finger.

7. 2 Teaching Suprasegmentals

Suprasegmental pronunciation features can also be presented to the learners in a manner that will raise their interest. For example, the teacher can start the lesson by a couple of sentences with no stress at all and ask the students for their observations. Then, they can read the segment correctly and ask the students to point out the difference. In the third reading, the stress is placed randomly and the students are again asked for their observations. This simple listening exercise should demonstrate the importance of learning suprasegmental pronunciation features. (Underhill, 1994, p. 177). Even though suprasegmentals are arguably more difficult to teach, a variety of techniques still exists for their instruction. Some are easy to put in practice, for example having the students to listen to a recording and repeat, or to have them read aloud texts focused on specific features.

Kinesthetic activities can also be employed in teaching suprasegmentals like pitch, volume, the rate of speech or stress. The students can, for example, demonstrate stresses by clapping their hands. Owing to technology, teachers can also give pronunciation exercises as homework. For example, the students can be asked to record themselves talking about a topic. Naturally, most of them will re-do the recording several times in order to choose the

39

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION one with the best pronunciation. It can then be uploaded and the teacher can comment with feedback. Interactive activities can also be done in class. Underhill (1994) suggests the activity of “parallel speaking”, in which the learners speak at the same time as a recording with the aim of imitating the stress, rhythm and intonation, ideally using headphones in a language laboratory. (p. 203). Visual learners can benefit from established written sign clues that signalise certain features, for example dots for words stresses, capital letters for sentence stresses, arrows for intonation or arcs for linking. The teacher can also develop a system of hand gestures that, if used regularly and coherently, can signal to the learners the specific pronunciation mistakes they have made and help them correct them immediately. The teacher should combine pronunciation instruction with grammar, which can be beneficial for example for learning the correct stress placement in compound nouns, the correct intonation depending on the sentence type, or the reduction of grammar words.

Pronunciation training can also be creatively combined with teaching life and institutions of

English speaking countries. Ingels and Huensch (2007) suggest using Martin Luther King’s

“I have a dream” speech to teach suprasegmentals. The students are first asked to make predictions about certain suprasegmentals they are about to hear and to correct them after listening. (Huensch and Ingels, 2007). Other authentic materials can also be useful: for instance, the students can study and learn to recite poems, which can help them with rhythm and other pronunciation features. (Celce-Murcia, Brinton & Goodwin, 1996, p. 301). Poems can have the advantage in being very rhythmical, which can be a good starting point before moving on to less metrical material.

7. 3 Error Correction

There are various approaches towards error correction in pronunciation training. It is possible to distinguish self-correction, peer correction or teacher correction; it can be immediate or delayed. The overlapping rule is that the correction should not be overdone: too much error correction in fluency exercises could be contra-productive and destructive

40

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION for the learner’s self-esteem. It does, however, have its place in accuracy speaking exercises.

Gestures with set meanings can help the students to self-correct minor errors without much interruption. Delayed correction by the teacher can enable the correction of larger mistakes or ones that need an explanation. Focusing on one pronunciation feature in one class can help the students stay aware of their pronunciation of the feature and even remember it when speaking English outside the class setting. The advantage of peer feedback is that it engages the students who are not speaking at the moment, who can practice their perception skills.

Celce-Murcia, Brinton and Goodwin (1996) recommend practicing peer-correction in groups of three or four, as in pairs it is difficult to assess if the word in question has been wrongly pronounced or wrongly heard. (p. 351). This kind of exercise can be focused on minimal pairs distinction: one learner pronounces one of the words on a list and the rest of the learners must identify which one it was. Peer correction can also be employed during free practice: the students can be asked to watch out for the pronunciation of the feature that is studied in the class. It is necessary to make sure that the learners feel comfortable with peer correction, as it could potentially be an unpleasant experience for learners in some classes where the atmosphere among the learners is not ideal.

7. 4 Testing Pronunciation

If the curriculum contains pronunciation training, the teachers should be able to evaluate it as well. Nevertheless, evaluating pronunciation is more difficult than evaluating some of the more straightforward and objective systems, such as grammar. It may be advisable to first evaluate perception, as there is a number of easily prepared exercises to test that for both segmentals and suprasegmentals. Most of them require a recording and a worksheet for the students to pick the correct phonemes, intonation patterns or reduced words; some features can be evaluated even without recordings (choosing the correct words stress). The actual pronunciation is more difficult to assess. It can take the form of a diagnostic passage in which all kinds of potentially difficult pronunciation features are

41

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION included. The students should be allowed to rehearse reading it first, as this allows to obtain the most representative sample of their pronunciation; even native speakers can stumble when reading an unknown text. Despite its advantage in containing all problematic elements, the drawback of this assessment is its artificiality. A free speech sample can be a suitable alternative, but the students should be allowed to speak about a topic they are familiar with.

Standardised tests are also available, the most widely used one being the Test of Spoken English

(TSE). The tasks include “reading aloud, describing a picture, telling a story from a series of pictures, sentence completion, presenting a schedule or syllabus, and either expression one’s opinion about a controversial subject or describing a familiar object” (Celce-Murcia, Brinton

& Goodwin, 1996, p. 347). The main assessment criterion is the intelligibility of the speaker.

All of these tests can serve as diagnostic tests. If pronunciation is tested in class and evaluated, the process should correspond to general principles of testing: i.e. to test only what has been covered in the course.

After having analysed what kinds of textbooks are suitable for teaching pronunciation and having explored the theory of pronunciation teaching, a research will follow with the aim of finding out what kind of mistakes lowers the intelligibility of Czech ESL learners.

8. Research methodology

The research for this thesis is constructed to find out whether it is segmentals or suprasegmentals in general that lower the intelligibility of Czech learners of English. Three recordings have been made: one containing only suprasegmental mistakes, one comprising only segmental mistakes and one containing both (see Appendix 1 on a CD). The first two recordings require a certain set of skills, and previous research by Zuzana Kanioková (2014) suggests that even proficient speakers may not be able to record them well without proper training (p. 56). The recordings have therefore been made artificially by a trained phonetician,

PhDr. Kateřina Tomková, PhD. The last recording, which contains both segmental and suprasegmental mistakes, has been made by a regular English learner on an A2+ level. She

42

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION was given time to read the text beforehand and also to hear the correct pronunciation once.

Unlike the previous two recordings, this one is not artificial: the speaker has not been not asked to make any mistakes but to read the text according to her best ability. The reason for including this recording is its character of a “control” recording. It enables to pinpoint what kinds of pronunciation mistakes people notice when both segmental and suprasegmental mistakes are present. It can also be compared to the previous ones to demonstrate that the mistakes chosen artificially for the first two recordings correspond to what a general learner sounds like.

Regarding the recordings themselves, three short jokes have been selected for the recordings for several reasons. First, they should make it fun for participants who will have to listen to the recordings. Moreover, the nature of jokes makes it possible to easily assess the intelligibility of the speaker. If the participants do not understand the joke, it is most probably caused by a wrong pronunciation, as the vocabulary the jokes contain should be understandable for a learner at a B1 level or above19. In the following segment, a transcription of the three recordings is provided, and pronunciation mistakes are highlighted in yellow.

Recording A

An Irish priest is driving down a highway and is pulled over for speeding. The policeman smells alcohol on the priest’s breath and then sees an empty wine bottle beside him. He asks the priest, “Sir, have you been drinking?” The priest responds, “No officer, just water,” The policeman asks, “Then why do I smell wine?” The priest looks at the bottle and says, “The Good Lord! He’s done it again!”

[ən (.)20ˈaɪərɪʃ (.) priːst iz ˈdraɪvɪŋ daʊn (.) ə ˈhaɪweɪ ən iz pʊld (.) ˈəʊvə fə ˈspiːdɪŋ.

ðə ˈpoliːsmən smɛlz (.) ˈælkəhɒl ɒn ðə priːsts brɛθ ənd ðɛn siːs (.) ən ˈɛmpti waɪn (.) ˈbɒtl

ˈbɪsaɪd hɪm. hi ɑːsks (.) ðə priːst sɜː ˈhæf jʊ biːn (.) ˈdrɪŋkɪŋ ðə priːst rɪsˈpɒndz ˈnəʊ ˈɒfɪsə

19 The most difficult words found in the jokes are the words “genie” and “transplant” but they should be understandable from the context. 20 “(.)” marks a pause in speech. 43

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

ʤəst ˈwɔːtə ðə ˈpoliːsmən ɑːsks ðɛn waɪ dʊ aɪ (.) smɛl (.) waɪn ðə priːst lʊks ət ðə ˈbɒtl ənd sɛz ðə gʊd lɔːd hiːz dʌn (.) ɪt (.) əˈgɛn]

Recording A has been made artificially with the intention of making a lot of suprasegmental mistakes but with segmental elements pronounced correctly. The first incorrect pronunciation element to be noticed in the absence of linking (an Irish) accompanied by unnecessary pauses. Word stress is placed incorrectly in words policeman and beside. It is placed on the first syllable of the word, where Czech learners often incorrectly place it. Sentence stress in also placed on words where it should not be (bottle) and lacks where it should be (no). The intonation pattern in the question is also not natural to English.

Due to frequent pauses, the rhythm is equally not correct. A lack of reduction can be observed on indefinite articles and also on the word have which should be reduced in this context. There is one unintentional segmental mistake in the word have – final consonant devoicing.

Recording B

Pat and Murphy are out fishing and the boat motor dies. After two days and drifting miles from the coast, they find a bottle in the water. Pat rubs the bottle and a genie poofs out. “I will grant you one wish,” says the genie. Without a thought, Pat says, “I wish to turn the sea into Guinness.” The genie says, “Your wish is my command,” and the sea turns into Guinness. Murphy yells at Pat, “You fool! Now, we’ll have to pee in the boat!”

[pet ən ˈma:rfi rər aʊt ˈfɪʃɪnk ənd də boʊt ˈmoʊtər daɪs ˈɑːftər tuː deɪz ən ˈdrɪftɪn maɪls frəm dəkəʊst ðeɪ faɪnd ə ˈbɒtl21 ɪn ðə ˈvɔːtr pet rʌps də ˈbɒtl ənd ə ˈʤiːni poofs aʊt aɪ vɪl grɑːnt jʊ vʌn vɪʃ sɛz də ˈʤiːni wɪˈdaʊt ə tɔːt pet sɛz aɪ vɪʃ tə ta:rn də siː ˈɪntə ˈgɪnɪs də ˈʤiːni sɛz jə vɪʃ ɪz maɪ kəˈmeːnd ənd də siː ta:rnz ˈɪntə ˈgɪnɪs ˈma:rfi jɛlz ət pet jʊ fuːl naʊ viːl həf tə piː ɪn də bəʊt]

21 Clear l 44

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

Recording B has also been made artificially with the goal of making all kinds of segmental mistakes Czech learners usually make. On suprasegmental level, the recording corresponds to what a native speaker would sound like. The speaker replaces (although not consistently) voiced dental fricatives by [d] (the) and a voiceless dental fricative by [t] (thought).

The open-mid central vowel [ɜ] which does not exist in Czech is replaced by [a:], which is very common for Czechs (turn, Murphy). Similarly, the vowel [æ] which is difficult to pronounce for Czech native speakers is approximated to [e] (Pat). Final consonant devoicing can be observed as well as a wrong pronunciation of the [ŋ], which does not occur in final positions in Czech. The speaker accurately mispronounces it as [nk] in the word fishing. Another mistake the pronunciation of [w] as [v] in words such as water, one or wish, which is caused by the absence of the phoneme [w] in Czech. Concrete realizations of phonemes are also altered to resemble a Czech learner. The [l] in bottle is too clear and the [r]

(water, turn) is alveolar and trilled instead of post-alveolar; both of these mistakes give the recording an appropriately “Czech” sound. Finally, one unclassifiable mispronunciation is made on purpose: the word command is pronounced as [kəˈmeːnd], based on a realistic assumption of what a lower-level Czech learner would do.

Recording C

An Englishman, an Irishman and a Scotsman were reading a newspaper article about which nationalities’ brains were for sale for transplant purposes. An Irishman’s or a Scotsman’s brain could be bought for €500 but an Englishman’s brain cost €10,000. “That proves, said the Englishman, “that

Englishmen are much cleverer than Irishmen or Scotsmen.” “No, it doesn’t,” said the Irishman, “it just means that an Englishman’s brain has never been used.”

[en ˈɪŋglɪʃmen en ˈaɪərɪʃmen ənd _ ˈskɒtsmen we:r ˈriːdɪn ə ˈnjuːzˌpeɪpər ˈɑːrtɪkl

'əbaʊt vɪʧ ˈneʃənelɪtis breɪns wer for seɪl fə ˈtrensplɑːnt ˈpərpousɪs ən ˈaɪərɪʃmenz ɔːr _

ˈskɒtsmenz breɪn kʊd bi bɔːgt fɔːr faɪf ˈhʌndrd ˈjʊro bət _ ˈɪŋglɪʃmenz breɪn kɒʊst tɛn

ˈtaʊzənd ˈjʊro dæt pruːfs sɛd də ˈɪŋglɪʃmen dæt ˈɪŋglɪʃmen ɑr mʌʧ ˈklɛvərər dən ˈaɪrɪʃmen

45

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

ɔːr ˈskɒtsmen noʊ ɪt dʌznt sɛd də ˈaɪərɪʃmen ɪt ʤʌst miːns dət (də) ˈanglɪʃmenz breɪn hes

ˈnɛvər biːn juːzd]

The third recording differs from the previous two in that it is authentic: the speaker is a real Czech-native English learner and the pronunciation mistakes she makes are genuine.

The only aspect that influences the authenticity is that she reads the joke rather than narrates it. The recording corresponds to the presumption that a genuine Czech ESL learner makes both segmental and suprasegmental mistakes. Much like in recording B, a schwa is often replaced by [e] in indefinite articles and other words; in were, [ɜ] is replaced by [e:]. Voiced dental fricatives are consistently replaced by [d] (the) and a voiceless dental fricative is replaced by [t] in thousand. The vowel [æ] is substituted by [e] in nationalities. Final consonant devoicing occurs multiples times in the recording. The consonant [w] is replaced by [v] in which. The phoneme [r] has several allophones in the recording: in some places, it is post- alveolar, is others it is alveolar. Some segmental mispronunciations are caused by the unfamiliarity of the pronunciation of some words such as transplant purposes, others probably by the stress of being recorded (cost, bought, the last occurrence of the word Englishman). As for suprasegmental mistakes, recording C is also similar to recording A. Misplaced word stress can be observed on words such as nationalities and about. The quality of the speaker’s intonation fluctuates, but it is wrong at the very beginning of the recording. The rhythm does not sound natural. The speaker has probably never been taught reduction: she pronounces nearly every word in its full form (for, were). Finally, there is insufficient linking between nouns and their determiners. There are two mistakes that are neither segmental nor suprasegmental: the speaker omits two indefinite articles and replaces one indefinite article by a definite one.

This recording not only serves as an authentic “control” recording, but also demonstrates that the two artificial recordings are rightly executed and that they reflect common pronunciation mistakes of Czech learners.

46

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

The three recording have been given to research subjects to assess them (see

Appendix 2 for the filled questionnaires). This is the original questionnaire that has been distributed to them:

Figure 4 - Questionnaire Questionnaire

You are going to hear 3 recordings. After hearing each recording, answer questions about it. At the end, there are 3 final questions that ask you to compare all 3 recordings. When assessing the impressions the recordings make on you, please try to focus on the speaker’s pronunciation rather than the quality of the joke  . First name: Native language: Have you ever lived in the Czech Republic? YES / NO If yes, indicate how long: How you ever been exposed to a Czech accent? YES / NO If yes, indicate how long: Recording A How would you rate the intelligibility (=how understandable it is) of recording n. 1? (10- perfectly intelligible, 0 – not intelligible at all)

☹ 0 1 2 3 4 5 6 7 8 9 10  How would you rate the speaker’s English? (1- very good, 2 – rather good, 3- average, 4- rather poor, 5- poor)

 1 2 3 4 5 ☹ How would you, in your own words, describe the speaker’s English? ______What are the biggest mistakes in pronunciation that you noticed? ______Recording B How would you rate the intelligibility (=how understandable it is) of recording n. 1? (10- perfectly intelligible, 0 – not intelligible at all)

☹ 0 1 2 3 4 5 6 7 8 9 10  How would you rate the speaker’s English? (1- very good, 2 – rather good, 3- average, 4- rather poor, 5- poor)

47

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

 1 2 3 4 5 ☹ How would you, in your own words, describe the speaker’s English? ______What are the biggest mistakes in pronunciation that you noticed? ______Recording C How would you rate the intelligibility (=how understandable it is) of recording n. 1? (10- perfectly intelligible, 0 – not intelligible at all)

☹ 0 1 2 3 4 5 6 7 8 9 10  How would you rate the speaker’s English? (1- very good, 2 – rather good, 3- average, 4- rather poor, 5- poor)

 1 2 3 4 5 ☹ How would you, in your own words, describe the speaker’s English? ______What are the biggest mistakes in pronunciation that you noticed? ______Final questions Which recording sounds the most like a native speaker? A B C Which recording is the most is the most pleasant to listen to? A B C Which recording contains the most mistakes? A B C Any additional comments? ______

First of all, “native language” has been an important factor in the choice of participants. While the methodology of the research resembles Iveta Žákovská’s (2017) thesis, this research includes participants of various native languages in order to examine how their perceptions differ. Four groups of non-Czechs have been asked. Firstly, native speakers of English, for whom the assumption is that their perception of pronunciation mistakes will differ from non-native English speakers. For a greater variety, speakers from different

English speaking countries have been asked to participate. The second group comprises of people whose mother tongue are other Slavic languages. This group should reveal if the

48

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION perception of segmental and suprasegmental mistakes differs if the interlocutor’s native language is from the same language family where Czech belongs to. Thirdly, speakers of

Romance languages have been asked to fill in the questionnaire. Neither English nor Czech is a Romance language, but these languages still belong to the Indo-European language family, representing a generic type of foreigners a Czech English learner might need to communicate in English with during their life. Finally, speakers of various Asian native languages form the last group of research participants. These have been chosen because

Asian languages often differ significantly from Indo-European languages on both segmental and suprasegmental level. They represent a group which might be one of the more difficult ones for Czech people to communicate with.

The following two questions of the questionnaire allow to recognize the possible influence a long exposure to the Czech language could have on the results, with the assumption that, for example, a native English speaker who has lived in the Czech for years will have accustomed themselves to the particularities of the accent. The following question on intelligibility of each recording is the core of the research. The term is explained for possible lower-level research subjects and scales are provided for the accuracy of results. The next question asks to rate the participant’s English as a whole, as good intelligibility is not everything that makes the speaker pleasant to listen to. This question is further elaborated by the participants’ evaluation in their own words, with the aim of finding out what impression each set of mistakes makes on the interlocutors. The forth question asks the participants to identify specific mistakes they have heard to be able to examine what stands out to them the most. A set of final questions asks the participants to compare the three recordings: their authenticity, their aesthetic impression of them and the mistakes they contain.

49

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

9. Questionnaires analysis

9. 1 Characteristics of the Research Participants

The respondents form a diverse group of people. The group of native speakers consists of five people: one American, one Irish, one Australian and two English people.

Their length of exposure varies from a week to three years. Only two of them have lived in the Czech Republic before. The Asian group is very varied, as it consists of participants from

Kazakhstan, Japan, South Korea and China. This diversity is not a drawback, as the main criterion for this group is a native language that is not of Indo-European origin, and therefore bears little resemblance to Czech. The participants, however, are not completely unfamiliar with the Czech accent as they have all lived in the country before, with the length of stay between six months and three years. The group consisting of speakers of Romance native languages comprises of Italian, Spanish and French people. Most of them are currently based in the Czech Republic with the length of stay ranging between a year and two years. One participant has only a limited exposure to some Czech speakers over the period of two months. The six participants who speak Slavic languages as their mother tongue come from four countries: Slovakia, Poland (both languages being similar to Czech), Bosnia and

Herzegovina and Bulgaria. The Bosnian language is a variety of Serbo-Croatian, which belongs to South Slavic languages and therefore is less close to Czech than the two previous; so is Bulgarian. Only two respondents have lived in the Czech Republic before; the rest of the participants has experienced only limited exposure to the Czech language.

9. 2 Recording A

9. 2. 1 Native speakers

As far as intelligibility on concerned, native speakers have indicated that recording A containing suprasegmental mistakes is the most understandable. Although they share a lot of opinions with the other groups of respondents, native speakers have been able to identify the specific pronunciation problems with a greater accuracy than the other groups. As a

50

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION group, they have been able to point out the majority of suprasegmental errors. Despite that, they are very positive in the assessment of speaker A’s pronunciation. Recording A has received the grade of 1.4 with the near perfect intelligibility of 9.6. They have praised the speaker’s very clear speech which is not difficult to follow. Naturally, none of the native speakers has mistaken speaker A for a native speaker, unlike some non-natives. Nonetheless, they all have marked speaker A as being the closest to one. Regarding the specific pronunciation issues, three out of five subjects have identified the wrong word stress placement, with the word policeman being the most noticeable. They have also been able to identify other problems, such as a “strange pitch”, “unnatural flow” and a “robotic tempo”.

For some, the slow tempo has helped the intelligibility, for another, it has made it more difficult to understand. The native speakers have also indicated their personal impressions of the recording, calling the speaker “cautious” but also “very well spoken” and having an accent that is pleasant to listen to.

9. 2. 2 Speakers of Asian languages

The analysis of the results partially confirms the assumption that Asian people will not be so perceptive towards the suprasegmental mistakes, as the suprasegmental elements tend to be very different in their languages. Half of the participants could not detect any specific mistakes in recording A. One marked the pronunciation of pulled over as wrong, possibly meaning the lack of linking. One participant correctly observed: “Punctuations are not natural. The first letter tends to be pronounced too strong22.” Had she had some linguistic background, she would likely have named this a wrong rhythm and a misplacement of stress on the first syllable. It is worth mentioning that this participant has resided in the

Czech Republic for the longest time and works in a multi-national environment with both

Czech people and native English speakers. Despite this assessment, she gave the speaker 10 points for intelligibility, with the group average being 9. Even the written remarks say

22 The quotes are kept in their original wording. 51

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION unanimously that the recording is very clear and understandable. The overall grade assessing the speaker’s English as a whole is 1.25 and all participants have marked recording A to be the most resembling of a native speaker’s English, as well as the most aesthetically appealing one. Another conclusion can be drawn from the participants’ personal observations. Half of the participants have mentioned that the speaker has a British accent. This further suggests that for some respondents correct segmental features automatically mean sounding like a native speaker, disregarding the suprasegmentals.

9. 2. 3 Speakers of Romance native languages

This group’s results correspond to the other groups. All participants marked recording A as being the most resembling of a native speaker’s speech and also the most pleasant one to listen to. The recording has received an average score of 9 for intelligibility and a 1.75 grade average. The respondents have not really been able or willing to name individual pronunciation problems, despite the fact that they have perceived them: one of them calls them “the accentuation” another “not a very natural [speech]” One participant has remarked: “I would not call them mistakes. It is just a different accent”, which is an interesting demonstration of how suprasegmental mistakes may be perceived as less serious than mistakes in segmentals by the interlocutors. The group has been generally positive in the subjective evaluation of the recordings. They describe the speaker as “fluent”, “sounding like a native speaker”, “clear and understandable” and “very good”, although the “Czech accent is still present”.

9. 2. 4 Speakers of Slavic languages

Much like the group of native speakers, speakers of Slavic languages have given recording A a very high intelligibility score of 9.7. Equally, they have all rated the speaker’s

English as very good (the grade average is therefore 1) and they have all marked recording A as the most intelligible and the most pleasant to listen to. This has not prevented them to identify what is wrong with the recording. While they have praised the clarity of the

52

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION recording, they have also noticed the frequent pauses, a “disturbed flow” and in one case also the misplaced stresses. One participant has remarked that while the speaker has “good pronunciation, [she puts] rather a lot of emphasis on the proper ways of pronouncing. It takes away from the flow of speaking”. Much like in the Asian group, one participant has mentioned that “the woman can imitate British accent very well” and another one has stated that “the speaker’s English sounds as native English”. This further shows that correct

Received Pronunciation segmental features make the non-native interlocutors associate the speaker with being native even if the suprasegmental pronunciation features are not correct.

9. 3 Recording B

9. 3. 1 Native speakers

The group of native speakers has not been so indulgent towards the segmental mistakes in recording B. The respondents have given this recording a mark one grade lower than recording A, resulting in the average of 2.4. The intelligibility of this recording is also ranked much worse, with an average of 6.2. Even the remarks the participants have made suggest a significant impact of segmental mistakes on the overall intelligibility, which can be noted in the following examples: “I understood the overall gist of her joke, however, there were some words that I struggled to understand”, “I had to concentrate and pay more attention than in Recording A”, or “[…] mostly understandable, but I couldn't figure out some words until listening a second time”. The native speakers have also been able to identify a lot of specific pronunciation mistakes. The most prominent one was the substitution of

/w/ by [v]. The absence of dental fricatives has also been noted by several participants.

“Trilling the r’s” has been mentioned twice. The most commonly noticed mistake in vowels has been the mispronunciation of the word command. Apart from that, only one participant has noticed the remaining problematic vowels (the absence of schwa, the replacement of

/æ/ by [e]. Neither final consonant devoicing nor a clear /l/ have been mentioned in this group, allowing for a conclusion that these are not the most important segmental

53

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION pronunciation mistakes. The native speaker’s comments along with the lower grades they have given the recording suggest that they automatically associate the presence of frequent segmental pronunciation mistakes with lower level learners (“beginner or intermediate”), although all the suprasegmental elements are native-like. Interestingly, the intelligibility of a speaker or their level of English does not have to correlate positively with the aesthetic impression the recording gives. Two out of five native speakers of English marked recording

B to be the one most pleasant to listen to. One participant has made the following comment:

“Recording B was the most pleasant to listen to as sentences were well structured whilst also speaking in a conversational tone. Recording A felt too forced, which disengaged the listener.” This confirms the theory that some native speakers of English differ from other groups by preferring a suprasegmentally correct speech.

9. 3. 2 Speakers of Asian languages

The participants’ impressions of recording B are rather diverse. The average score for intelligibility is 7 points and the grade average is 2, but the individual comments and assessments vary significantly. For one participant, the recording’s intelligibility is “quite good”, another one has barely been able to understand it and had thought at first that she had been listening to German. The third participant has been able to understand the recording despite its lack of clarity, the forth participant called it “very good English”. Unlike in recording A, the participants have been able to identify specific pronunciation mistakes, namely the substitution of dental fricatives, replacement of /w/ with [v], or the trilled [r].

One of the participants raises an interesting question by stating: “The sound “r” and “w” are not pronounced according to the phonetic rules, though this is not a mistake; it is rather an accent.” While some phoneticians would join this participant in claiming that the usage of a trilled r is indeed not a mistake, the latter statement is more controversial. The author of this thesis believes that the /w/ versus [v] substitution is indeed a mistake. That is because the

54

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION different variants of the pronunciation of the letter r ale mere allophones, whereas /w/ and

/v/ are distinctive phonemes that form minimal pairs.

9. 3. 3 Speakers of Romance languages

This group has been consistent in a negative evaluation of recording B. In fact, they have given it the lowest average number of points for intelligibility (5 on average) and the worst grade average of 3.75. The most prominent mistake noted by this group is the substitution of /w/ with [v]. Other mistakes mentioned are the wrong vowels in words turns and command, mispronunciations of dental fricatives and trilled r’s. The individual comments reveal that mistakes in segmentals are perceived as rather serious by this group, as they have made comments such as “a very foreign accent”, “could be a lot better” or “it sounds broken”. One participant has written: “The accent is very noticeable. It sounds more like

German or Russian”, which corresponds to the mistakes the have pointed out.

9. 3. 4 Speakers of Slavic languages

Much like the speakers of Romance languages, Slavic respondents have found the intelligibility of recording B very low, giving it an average score of 5.2 points. The speaker’s level of English has been awarded grade 3, meaning “average”. The participants have noticed specific pronunciation mistakes such as [v] instead of /w/, a trilled r, the replacement of voiceless dental fricatives by [t] and a mispronunciation (command). One participant has stated being unsure whether the name is Beth or Pat, which nicely illustrates the impact the /æ/ versus [e] distinction has. One participant has also noticed the clear /l/. Despite being from the same language group, most respondents have found the recording “a little bit more challenging to understand” with certain words being barely understandable. Only one respondent has written: “The accent of the speaker was quite strong and was troubling the understanding in the first listening. After my ears were adapted to the foreign accent, I didn’t have problems with understanding”. The speaker’s accent is judged as “bad”, “rough”,

“strong” and “resembling Russian”.

55

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

9. 4 Recording C

9. 4. 1 Native speakers

Unbeknownst to the native speakers, recording C presents the only recoding of an authentic Czech ESL learner in the set. In the assessment of recording C, it is interesting not to focus on specific mistakes the participants point out, but to compare whether they attribute the strong accent to segmental or suprasegmental mistakes. In case of the group of native speakers, the most commonly criticised aspect of pronunciation are actual mispronunciations, followed by both suprasegmental (“Intonation was placed at weird parts”) and segmental (“she is trying too hard to pronounce the R sound”) issues.

9. 4. 2 Speakers of Asian languages

The group of Asian native speakers has given recording C 6.5 point for intelligibility, which is only 0.5 point lower than they attributed to recording B. The grade average is 2.5, which is also only slightly worse than recording B. The hypothesis that Asian people will be more perceptive towards segmental than suprasegmental mistakes has not been confirmed here, as the majority of participants has noted suprasegmental mistakes over segmentals. The participants have noticed the following mistakes: a trilled r (a segmental mistake), wrong rhythm and a lack of intonation “that makes her English unnatural” (suprasegmental mistakes).

9. 4. 3 Speakers of Romance languages

This group has considered recording C, despite all its issues, to be better than recording B, in that recording B has been judged as less intelligible, having a worse level of

English and containing more mistakes. The comments the participants have made address mostly the pace and rhythm (for two of them, it is too quick; for one, it is too slow). As for the segmentals vs. suprasegmentals comparison, which is the reason for the inclusion of this recording, the results suggest that the respondents value both. They have pointed out

56

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION segmental mispronunciations and mistakes in vowels (transplant, purposes), as well as suprasegmental issues “it sounds artificial” and a wrong intonation.

9. 4. 4 Speakers of Slavic languages

Recording C has received a slightly more positive assessment by this group than recording B, with 5.6 intelligibility points and a grade average of 3.2. The respondents have noticed suprasegmental issues (“the rhythm of the speech wasn’t smooth”), but focused more on segmental mistakes, listing the problems with vowels, dental fricatives, trilled r’s and unclassifiable mispronunciations. Despite the more positive assessment, four out of five respondents marked recording C as containing the most mistakes. It is interesting to note that one participant suggests that although the mistakes are similar in character to those in recording B, the intelligibility of recording C is lower. This could be due to the combination of suprasegmental and segmental mistakes.

9. 5 English as Lingua Franca

Another interesting observation can be made regarding the concept of English as a lingua franca. Without being asked to do it, some respondents have touched upon this subject. The Kazakh participant has stated in the “additional comments” box: “None of the records has critical mistakes; it’s rather about the speaker’s way of speaking and some particular accents which for the learners might be challenging at the beginning, however, as the learner gets more practice, all the records will be understandable enough”. This comment can be interpreted in favour of the intelligibility principle: as soon as the participant is able to adjust to an accent and understand the speaker, they do not assign any importance to making the effort to sound like a native speaker. The aforementioned comment from an

Italian participant also supports this theory by saying: “I would not call them mistakes. It is just a different accent”. So does the comment of a Bulgarian participant: “The foreign accent was the only factor making impact of the intelligibility of the recording, except that the level

57

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION of English is good.[…]. After [the first listening] my ears were adapted to the foreign accent,

I didn’t have problems with understanding.”, which further supports the theory.

10. Overall Results Analysis

The following charts demonstrate the overall results, which will later be commented on. The first chart demonstrates the intelligibility score: the higher, the better.

Figure 5 - Intelligibility chart

Intelligibility 10

8

6

4

2

0 Recording A Recording B Recording C

The second chart shows the perceived level of English: the smaller the bar, the better.

Figure 6 - Perceived level of English chart

Perceived level of English

Recording A Recording B Recording C 1 1,5 2 2,5 3 3,5 4 4,5 5 1 - very good, 5 - very bad

The last three graphs demonstrate which attributes the respondents have given to which recording.

58

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

Figure 7 - The most native-like recording Figure 8 - The recording the most pleasant to listen to

The most native-like The recording the most sounding recording pleasant to listen to

Recording A Recording B Recording C Recording A Recording B Recording C

Figure 9 - The recording with the most mistakes

The recording with the most mistakes

Recording A Recording B Recording C

10. 1 Suprasegmental Mistakes

The results have proved to be quite homogeneous across the groups of respondents.

They demonstrate that suprasegmental pronunciation mistakes, when not accompanied by segmental issues, are not perceived very negatively. With an average of 9.3 points out of 10, they do not lower the intelligibility of the speaker. To a non-native ear, a recording with suprasegmental mistakes only can even resemble a native speaker’s speech. The perception of the speaker’s level of English correlates with intelligibility. The segmentally correct recording has also been assessed positively in terms of the level of English the speaker has; the average grade is 1.35, making the verdict “very good English”. Isolated suprasegmental mistakes are also often not perceived as mistakes as such, but rather as an individual person’s pronunciation. The lack of segmental pronunciation errors is associated with a good clarity of a speech and an advanced level of English. It does not lower the aesthetic value of the

59

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION recording for most participants, as a great majority have chosen recording A to be the most pleasant to listen to. The negative connotations include sounding unnatural or “robotic”.

The most noticeable suprasegmental pronunciation error seems to be the misplacement of word stress, and (to a lesser extent) sentence stress. Another tricky issue are irregularities in intonation and rhythm, the latter being easy to notice even for lower level ESL learners. The combined analysis of recording A and C reveals that a lower tempo of speech helps the intelligibility of a speaker, but a quicker tempo is associated with a greater fluency23. It is therefore possible to conclude that suprasegmental mistakes in isolation do not lower neither the intelligibility of the speech, nor the evaluation of the speaker’s English. However, due to the artificial character of recording A, it is necessary to take into consideration the fact that in real ESL learners, making only one kind of pronunciation mistakes is rarely the case.

Recording C of an authentic ESL learner demonstrates that in combination with segmental mistakes, suprasegmental issues are not overlooked. With the interlocutors having to concentrate on making out the words with segmental mistakes, suprasegmental errors do lower the intelligibility. Nevertheless, the overall conclusion of this piece of research is that segmental mistakes still are more serious than suprasegmental ones.

10. 2 Segmental Mistakes

Regarding segmental pronunciation mistakes, the respondents have also shown a general agreement: frequent segmental pronunciation mistakes do lower the intelligibility significantly, as the intelligibility score average is 5.6. They are also associated with a lower level of English, as the average grade given by the participant to recording B is 2.8, pointing to an “average” English. This is demonstrated by the fact that both recording A and B are grammatically and lexically correct, but it is the pronunciation only that has lowered the participants’ impression of the speaker’s English. This is with an exception of some native

23 For some learners, recording C has been associated with a better level of English also due to the quicker pace of speech. 60

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION speakers, who have acknowledged the quality of the speaker’s English apart from the segmental pronunciation mistakes. The recording containing segmental mistakes has been judged more strictly. It gives an impression of a strong Czech accent, sometimes also compared to a Russian or a German one. The speaker’s English sounds “rough”, “foreign” or even “broken”, despite the fact that it has good suprasegmental pronunciation elements.

In case of recording C, even though suprasegmental issues are also reproached, segmentals seem to be the most problematic. With segmental mistakes, even ESL learners are able to pinpoint the specific difficult areas. The most noticeable element is the substitution of /w/ by [v]. Trilling the r’s is also very perceptible even by non-natives. The incorrect pronunciation of voiced and voiceless dental fricatives (as [d] and [t]) has also not gone unnoticed. It is important to note that the fact the respondents have noticed some specific issues more than others does not automatically mean that these are the only mistakes that lower the intelligibility. On the contrary, even though the /æ/versus [e] opposition is not mentioned so frequently in the questionnaires, a remark by a Polish respondent shows that a misunderstanding has been caused by just that, as she is not sure whether the name that appears in the recoding is Beth or Pat. However, the frequency of appearances of specific pronunciation mistakes in the questionnaire shows which kind of problems are the most conspicuous to a native and a non-Czech ear. Other mistakes mentioned are various mispronunciations in vowels. It is worth mentioning that these results correspond in many regards to those of Ondráček’s research from 2014 (presented on page 33 of this thesis).

10. 3 The Differences Between Language Groups

The questionnaire analysis has demonstrated that the differences in perception between different language groups is not so significant. The average intelligibility score and level of English assessments are generally similar regardless of the group. However, some trends can still be identified. The assumption that some native speakers will prefer suprasegmentally correct speech with segmental mistakes only due to its natural flow has

61

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION been confirmed, although only partially: two out of five native speaker respondents have marked recording B as the most aesthetically pleasing one. Despite that fact that some respondents have are not of a high English level themselves, they have managed to identify a lot of mistakes they have heard. Even if they do not have the linguistic ability to name them, they are still able to perceive them and suggest what is wrong with the recording. Not surprisingly, native speakers have demonstrated the highest ability to correctly identify and name both segmental and suprasegmental issues, probably due a wider range of vocabulary they dispose of. The assumption that the speakers of Asian languages will not be so capable to identify suprasegmental pronunciation mistakes due to them often being problematic in their own English accents has not proved to be correct. In fact, the Asian respondents have shown average aptitude in the recognition of suprasegmental errors, not differing them from the other groups. The similarity of Slavic languages has also not influenced the results significantly. Despite the proximity of the native languages to Czech, the participants from

Slavic countries are not any less able to identify pronunciation mistakes. Their personal assessments of the speakers’ abilities in terms of pronunciation and a general English level have also not differed significantly from other respondents’ answers. It is therefore possible to conclude that apart from the group of native speakers, which is quite particular, different native languages of the interlocutors generally do not influence the intelligibility of Czech

ESL learners.

Another point for analysis that has emerged from the respondents’ comments is the inclination of some of them towards the ideas of English as a lingua franca theorists. Some of the participants have expressed a level or reluctance to judge the speakers’ mistakes, stating that as long that the recording is intelligible, the mistakes themselves do not matter that much. Some respondents have not called the segmental and suprasegmental issues mistakes but rather pronunciation varieties. This illustrates that while some ESL learners are aware of non-native speakers’ pronunciation mistakes, they do not assign them a lot of importance as

62

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION long as they understand the utterance without making a lot of effort. The conclusion to the issue of this sub-chapter is that the interlocutor’s nationality does not play a key role in the accent perception. It would probably be more pertinent to focus on the length of exposure to the Czech accent. The results suggest that some participants who have been exposed to a

Czech accent for a longer period of time face less problems understanding it, although it has not had any influence on the positivity or negativity in the assessment of the speaker’s accent.

10. 4 Implications on Teaching Pronunciation

The extent of the research is not sufficient for making big claims regarding its implications on teaching pronunciation, as the recordings are not long enough to contain examples of all pronunciation mistakes an ESL learner can make. Nevertheless, it is possible to suggest some discussion points. First, the research suggests that both segmental and suprasegmental mistakes lower the intelligibility of a Czech ESL learner, however, segmental mistakes do so more. Secondly, the research confirms that the approach towards teaching pronunciation depends on the learner’s goals. As explained in the previous sub-chapter, a learner whose goal is to make themselves understood by other speakers of English needs to focus primarily on elements that can lower the intelligibility and cause misunderstandings

(for example complete mispronunciations). A learner whose objective is to get as close to a native speaker’s pronunciation as possible, however, must focus on many more elements.

For example, while trilling r’s might not have any serious implications on one’s intelligibility, the research has shown that it is a segmental feature that is very noticeable to the interlocutors and that is worsens the perception of the level of English the interlocutors have of the speaker. Other such elements are the mispronunciations of dental fricatives and the consonant /w/ (the latter one of which can also cause some misunderstandings). As for suprasegmental issues, lowering one’s pace of speech can increase their intelligibility, but also cause the interlocutors’ judgement of the speaker’s English to lower. To sum up, the

63

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION approach towards teaching pronunciation should emerge from the goals the ESL learner wants to achieve with their English skills.

64

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

Conclusion

The theoretical section of this thesis has dealt with the approaches towards teaching pronunciation throughout history, from the point where pronunciation was not taken into consideration to the point where journal articles about it are published by ESL teachers. It has also demonstrated the main contrasting principles in pronunciation instruction that have emerged during that time. These principles aim either towards a native-like fluency, or towards comfortable intelligibility. The thesis explores the intelligibility principle. It is based on the notion of English as a lingua franca; the fact that Czech ESL learners are more likely to encounter other non-native speakers of English and a native-like pronunciation is therefore not required, but mutual intelligibility is. The research has suggested that a considerable number of respondents identify with this principle. The comparison of Czech and shows that the majority of segmental pronunciation issues is caused by the absence of the particular phoneme in the Czech language, or the usage of different allophones for certain phonemes. In case of suprasegmentals, it is usually the issue of distinctive characteristics of each pronunciation feature in the two languages.

The study of journal articles has also shown that while there are many theorists dealing with the issue of teaching pronunciation, the teachers themselves are sometimes not so willing to focus on pronunciation with their students. With the aim of examining the situation in the Czech Republic, a survey has been carried out regarding the textbooks used in English classes in Czech schools.

The results show that while some of the most commonly used textbooks comprise pronunciation sections of high quality, the majority of them is only average or below average in terms of teaching pronunciation. A theoretical section on teaching pronunciation has presented techniques that teachers can employ in class even if the textbook they use is not well equipped for pronunciation training.

The primary research of this thesis consists of three recordings distributed to four groups of either native speakers of English or various ESL learners with the aim of finding out what kind of pronunciation mistakes lowers the intelligibility of Czech ESL learners the most. The results

65

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION have shown that a speech that contains only suprasegmental mistakes and no segmental ones is judged by all groups as a very intelligible one, one that is pleasant to listen to and also the one that resembles a native speaker the most. It is therefore possible to conclude that isolated segmental mistakes are more serious than isolated segmental ones. The control recoding suggests that both kinds of mistakes in combination influence intelligibility and the other aspects studied; nevertheless, even this recording proves the more serious nature of segmental mistakes. The results further demonstrate that for a general interlocutor, pronunciation mistakes worsen the impression the they get of the speaker’s English. However, certain interlocutors do not consider pronunciation errors to be mistakes as such, but rather a person’s individual accent. They believe that it is acceptable to have one as long as the utterance is understandable. The research has identified segmental pronunciation mistakes that are the most noticeable by the interlocutors. For learners who aim to improve the impression their pronunciation makes on their interlocutors, it might be advisable to focus on rather easily learned features such as the /r/, /w/, /ð/ and /θ/. Contrary to the assumptions, the research has not revealed any substantial differences in the assessments of speakers of various native language groups, apart from the more obvious ones, such as native speakers’ more advanced ability to identify and name pronunciation mistakes. The research leaves room for a possible further continuation. It would be of interest to research intelligibility even more objectively. In this piece of research, the assessment of how intelligible the recoding is has been on the interlocutors themselves; they could have been influenced by other factors such as the aesthetic impression of the recording. Any further research in this topic could therefore devise a more objective system of assessing the intelligibility of the recordings.

66

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

References

Celce-Murcia M., Goodwin, J. & Brinton, D. (1996). Teaching pronunciation: a reference for teachers of

English to speakers of other languages. Cambridge: Cambridge University Press.

Collins, B. & Mees, I. M. (2013). Practical phonetics and phonology. Abingdon: Routledge.

Cruttenden, A. (2008). Gimson's pronunciation of English. London: Hodder Education.

Dalton, C. & Seidlhofer, B. (1994). Pronunciation. Oxford: Oxford University Press.

Derwing, T. M. & Munro, M. J. (2005). Second language accent and pronunciation teaching: a

research- based approach. TESOL Quarterly, 39(3), 379-397. doi:10.2307/3588486

Flege, J. (1981). The phonological basis of foreign accent: a hypothesis. TESOL Quarterly, (4), 443

455. doi:10.2307/3586485

Gordon, M. (2016). Phonological typology. Oxford: Oxford University Press.

Huensch, A. & Ingels, S. (2007). Teaching perception of suprasegmentals using popular media.

TESOL International Association. Retrieved from

https://www.tesol.org/docs/pdf/8266.PDF.

Jenkins, J. (2000). The phonology of English as an international language. Oxford: Oxford University Press.

Kanioková, Z. (2014). Segmental and suprasegmental properties of the English of Slovak learners. Brno:

Masarykova univerzita.

Krcmov̌ á, M. (2008). Úvod do fonetiky a fonologie pro bohemisty. Ostrava: Ostravská univerzita v Ostrave.̌

Kučera, H. (1961). The phonology of Czech. ’S-Gravenhage: Mouton.

Ladd, R. (2008). Intonational phonology. Cambridge: Cambridge University Press.

Ladefoged, P. (2005). A course in phonetics 5th ed. Boston, MA: Wadsworth Publishers. Retrieved from

http://www.phonetics.ucla.edu/course/chapter9/cardinal/cardinal.html.

Levis, J. (2005). Changing contexts and shifting paradigms in pronunciation teaching. TESOL

Quarterly, 39(3), 369-377. doi:10.2307/3588485

Morley, J. (1991). The pronunciation component in teaching English to speakers of other

languages. TESOL Quarterly, 25(3), 481-520. doi:10.2307/3586981

67

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

Ondráček, J. (2014). Comfortable intelligibility in spoken English discourse. Brno: Masarykova univerzita.

Palková, Z. (1994). Fonetika a fonologie češtiny. Praha: Univerzita Karlova, vydavatelství Karolinum.

Pennington, M. C. (1996). Phonology in English language teaching. New York, NY: Addison Wesley

Longman Limited.

Pesce, C. (n.d). 7 Fun ESL games to practice pronunciation. Busy Teacher. Retrieved from

https://busyteacher.org/14855-7-fun-esl-games-to-practice-pronunciation.html.

Roach, P. (1983). English phonetics and phonology: a practical course. Cambridge: Cambridge University

Press.

Underhill, A. (1994). Sound foundations. Oxford: Heinemann.

Underhill, A. (2005). Sound foundations: learning and teaching pronunciation. Oxford: Macmillan

Education.

Zemková, K. (2016). The impact of TV: formation of learners' English. Brno: Masarykova univerzita.

Žákovská, I. (2017). Segmental and suprasegmental mistakes in Czech speakers’ English pronunciation. Brno:

Masarykova univerzita.

Other sources

Begg, A. & Krantz, C. (2012). Maturita solutions second edition intermediate teacher’s book. Oxford:

Oxford University Press.

Castle, M., Krantz, C., Soars, J. & Soars, L. New Headway pre-intermediate teacher’s resource book third

edition. Oxford: Oxford University Press.

Davies, P. A. & Falla, T. (2017). Solutions third edition intermediate student’s book. Oxford: Oxford

University Press.

Davies, P. A. & Falla, T. (2012). Maturita solutions second edition intermediate student’s book. Oxford:

Oxford University Press.

Hancock, P. (2013). Insight intermediate workbook. Oxford: Oxford University Press.

Harris, M., Mower, D. & Sikorzsnka, A. (2006). New Opportunities pre-intermediate student’s book.

Harlow: Pearson Education Limited.

68

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION de la Mare, Christina. (2013). Insight intermediate teacher’s book. Oxford: Oxford University Press.

Mugglestone, P. (2006). New Opportunities pre-intermediate teacher’s book. Harlow: Pearson Education

Limited.

Myers, C., Thacker, C. & Wildman, J. (2013). Insight intermediate student’s book. Oxford: Oxford

University Press.

Soars, J. & Soars, L. (2007). New Headway pre-intermediate student’s book third edition. Oxford: Oxford

University Press.

69

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

Summary

The thesis deals with the pronunciation mistakes of Czech ESL learners. Its aim is to find out whether it is segmental or suprasegmental pronunciation mistakes that lower the intelligibility of ESL learners the most. The theoretical background for the thesis is a comparison of Czech and

English phonology which illustrates the problem areas for Czech ESL learners. The thesis also raises the theoretical questions surrounding pronunciation training, namely its goals. It presents the opposing approaches of either striving for a native-like pronunciation or following the goal of comfortable intelligibility. With the presumption that the interlocutor’s native language could play a part in intelligibility, speakers of languages from four language groups have been asked to participate: native speakers of English, speakers of Slavic languages, speakers of Romance languages and speakers od Asian languages. They have been asked to assess three recordings. The first one contains mistakes in suprasegmentals but no segmental mistakes. The second one contains segmental mistakes but has no segmental issues. Due to the level of accuracy required, these two recordings have been made artificially by a phonetician. The third recording has been made by an authentic Czech ESL learner and contains both segmental and suprasegmental mistakes in natural proportions. The respondents have been given questionnaires asking about the intelligibility of the recordings, about the general impression they make on them and also to identify specific pronunciation problems. The analysis of the results has revealed that segmental mistakes are perceived as more serious, they lower the intelligibility more than suprasegmental mistakes and they also lower the overall impression of the speaker’s English for the majority of participants.

These findings are valid for segmental or suprasegmental mistakes in isolation: when combined, both are judged as serious, as demonstrated by the third recording. The research has not proved any substantial differences in the assessments among the various groups of participants, apart from slight differences between the native speakers and the rest of the groups. The research has demonstrated specific pronunciation mistakes that are perceived the most by the interlocutors, namely the substitution of dental fricatives by [t] and [d], the replacement of /w/ by [v] and trilling

70

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION the r’s among segmental mistakes and wrong word and sentence stress among suprasegmental mistakes.

Summary in Czech

Diplomová práce se zabývá výslovnostními chybami Čechů učících se anglicky. Jejím cílem je zjistit, který typ chyb má větší vliv na srozumitelnost mluveného projevu; zda jsou to chyby v jednotlivých hláskách či prozodické chyby. Teoretická část práce obsahuje porovnání české a anglické fonologie s důrazem na oblasti anglické fonologie, které jsou pro Čechy problematické.

Práce se také zabývá teoretickými otázkami výuky správné výslovnosti, například jejími cíli.

Představuje dva její protikladné směry: snahu o výslovnost rodilého mluvčího či snahu o vzájemnou srozumitelnost s mluvčími ostatních jazyků. Vzhledem k předpokladu, že by tato srozumitelnost mohla být ovlivněna mateřským jazykem účastníků dialogu, zástupci čtyř jazykových skupin byli požádáni, aby zhodnotili výslovnostní chyby českých mluvčích: mluvčí, jejichž mateřským jazykem je angličtina, či jeden ze slovanských, románských nebo asijských jazyků. Byli jim dány k poslechu tři nahrávky. První obsahovala prozodické chyby, ale žádné chyby v jednotlivých hláskách. Druhá obsahovala chyby v hláskách, ale nikoliv v prozodii. Vzhledem ke specifickému charakteru těchto nahrávek byly tyto dvě vytvořeny uměle a nahrány fonetičkou. Třetí nahrávka je autentická; osoba, která ji nahrála, se sama učí anglicky. Tato nahrávka tedy přirozeně obsahuje jak prozodické, tak hláskové chyby. Respondenti odpovídali v dotazníku na otázky týkající se srozumitelnosti nahrávek a obecného dojmu z angličtiny mluvčího. Dále byli požádáni, aby popsali konkrétní výslovnostní chyby, které zachytí. Analýza výsledků odhalila, že chyby v jednotlivých hláskách jsou považovány za závažnější, více přispívají k nedorozuměním a také zhoršují dojem z angličtiny mluvčího. Tento závěr platí pouze pro izolované chyby; třetí nahrávka ukázala, že v případě kombinace obou typů chyb jsou oba považovány za závažné. Výzkum neukázal podstatné rozdíly v hodnocení mluvčích různých jazyků, kromě menších odlišností skupiny rodilých mluvčích. Z výzkumu vzešly konkrétní výslovnostní chyby, kterých si cizinci u Čechů nejvíce všímají. Je to výslovnost dentálních frikativ

71

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

/θ/ a /ð/ jako [t] a [d], náhrada fonému /w/ hláskou [v] a použití české alveolární vibranty /r/ co se týče chyb v hláskách, a chyby ve slovním a větném přízvuku co se prozodie týče.

72

SEGMENTAL VERSUS SUPRASEGMENTAL MISTAKES IN ENGLISH PRONUNCIATION

Appendices on CD

Appendix 1 - Recordings Appendix 2 - Questionnaires as filled out by the research participants

73