Repeat after me:
Mississippi, Mississippi, Mississippi
A study into the use of teaching articulatory setting differences in secondary educationEliene Gritter S1532642
Acknowledgements
2 CONTENTS
INTRODUCTION 4
ARTICULATORY SETTING 6
2.1 An idea put into words 6
2.2 A renewed interest 8
2.2.1 Articulatory settings and voice quality 8 2.2.2 Articulatory settings investigated 11
2.3 Definitions and technical aspects 13
2.4 Articulatory setting in education 15 2.4.1 Impressionistic descriptions 15
2.4.2 Techniques to teach AS 19
2.4.3 Empirical evidence 21
2.5 Present study 22
2.5.1 Rationale 22
2.5.2 Pronunciation in Dutch secondary education 23 2.5.3 The articulatory settings of Standard Dutch and British English 24 2.5.4 Rating pronunciation 26
2.6 Statement of purpose 28
METHOD 29
3.1 Participants 29
3.2 Materials 30
3.2.1 Picture description task 30 3.2.2 Training programme AS 31 3.2.3 Training programme segments 33
3.3 Procedure 34
3.4 Design and analysis 36
RESULTS 39
3
4.2 Between group differences 39
4.3 Individual differences 41 DISCUSSION 43 CONCLUSION 49 BIBLIOGRAPHY 51 APPENDIX I i APPENDIX II ii
APPENDIX III iii
APPENDIX IV xii
APPENDIX V xxi
APPENDIX VI xxii
APPENDIX VII xxxiv
4 INTRODUCTION
When speakers of a certain L1 are verbally communicating in a language other than their own, listeners usually know that the language used is not their mother tongue. Even when learners have otherwise perfectly mastered their L2, they can often be distinguished as non-native users of the language because of the foreign accent with which they speak it. Many people do not think this is a problem; as long as L2 speakers can make themselves
understood, they do not have to sound like native speakers. However, there are a number of reasons why L2 learners might want to improve their L2 pronunciation. Firstly, because of an L1 accent, L2 speakers might not be accepted, or considered of lower social status, or even lower intelligence by native speakers of their L2. But even more importantly, a foreign accent can also lead to a reduction of a speaker’s intelligibility which may then lead to a breakdown in communication.
However, even after many hours of practising L2 speech sounds learners may still not sound like L1 speakers, because differences in pronunciation between languages are based on more than just the speech sounds that are used. Over the years, it has been argued that in order for learners to improve their L2 pronunciation they should not focus on segmental but on supra-segmental differences in pronunciation. A good start for improving L2 pronunciation is acquiring the right articulatory basis or articulatory setting, the overall posture of the speech organs which underlies all articulations of the individual speech sounds of a language.
Beatrice Honikman (1964) even argued that it is not possible for learners to fully master their L2 pronunciation while applying the articulatory setting that belongs to their L1. The idea of articulatory setting, especially related to the teaching and learning of a foreign language, forms the topic of the present study.
5 articulatory setting in pronunciation teaching have been conducted. Therefore, the present study tries to establish the effectiveness of teaching articulatory setting differences by
answering the research question ‘can the teaching of the English articulatory setting tot Dutch secondary school pupils improve their pronunciation of English’ and with that will add to the very little empirical evidence there is concerning the use of articulatory setting in education.
6 ARTICULATORY SETTING
2.1 An idea put into words
In 1964, Beatrice Honikman was the first to use the term ‘articulatory setting’ in her essay titled ‘Articulatory Settings’. Published among other articles on phonetics in the influential In Honour of Daniel Jones (1964), Honikman’s article was, and still is, well read. In the article, described by Laver to be ‘the best extended discussion of the concept of setting’ (1991: 380), Honikman provides her readers with a thorough description of what she understands to be articulatory setting (from now on referred to as AS) and how the concept of AS is an important aspect of the study of pronunciation in general and articulation specifically. Honikman also provides a detailed description of the settings of both British English and French, languages she was familiar with through her position as a language teacher, while adding some remarks about the settings of Arabic, German, Russian and a number of other languages here and there.
After starting her article with a short comment on the analysis of languages so far, Honikman introduces the term ‘articulatory setting’ and explains what she means by the term:
By articulatory setting is meant the disposition of the parts of the speech mechanism and their composite actions, i.e. the just placing of the individual parts, severally and jointly, for articulation according to the phonetics substance of the language
concerned. To put this another way, it is the over-all arrangement and manoeuvring of the speech organs necessary for the facile accomplishment of natural utterance. Broadly, it is the fundamental groundwork which pervades and, to an extent,
determines the phonetic character and specific timbre of a language. It is immanent in all that the organs do. (1964: 73)
Thus an articulatory setting is the gross oral posture and mechanics, both external and internal, requisite as a framework for the comfortable, economic, and fluent merging and integrating of the isolated sounds into that harmonious, cognizable whole which constitutes the established pronunciation of a language. (1964: 73)
7 much, but focuses on the shared characteristics of those individual sounds that together form an underlying system. Secondly, AS is language specific, meaning that every language has its own unique setting which is shared by all its native speakers. Actually, as Collins and Mees argue, AS is even dialect specific; different varieties of the same language have their own unique AS (Collins & Mees 2003b). The ideas discussed in Honikman’s article would be used and explored by many phoneticians after her.
However, long before the term ‘articulatory setting’ was first used by Honikman, the concept was already known to and commented upon by those interested in languages, linguists and phoneticians. In his historical survey, Laver (1991) states that phonetic
comments on AS have been found in writings as early as the seventeenth century. In 1653, for example, John Wallis discussed the settings of a number of European languages. He observed that ‘the English as it were push forward the whole of their pronunciation into the front of the mouth, speaking with a wide mouth cavity, so that their sounds are more distinct,’ while the Germans speak in the back of their mouths and the French articulate near to the palate which makes their mouth cavity less wide (1653 in Laver, 1991). Not much later, in 1668, Wilkins also discussed the ASs and voice quality of several languages as Wallis had done earlier. In addition to simply observing setting differences, Wilkins noticed that the frequency and appearance of certain segments in different languages might cause the differences in ASs, a new step in the development of the discussion of language-specific AS.
From the second half of the nineteenth century onwards, the idea of AS started to receive widespread attention from researchers in the field of (articulatory) phonetics.
Especially in Germany many phoneticians showed an interest in the subject. Laver (1991) and Jenner (1997; 2001) list a number of German phoneticians who discussed the idea of AS, then referred to as artikulationsbasis, either as the main subject of their studies or as part of their more general discussions of phonetics. At the same time as the early German accounts of AS, phonetician Henry Sweet first wrote about AS, which he named ‘organic basis’, in Britain, in 1877 (in Laver, 1991). In his A Primer of Phonetics (1906), he states that ‘every language has certain general tendencies which control its organic movements and positions, constituting its organic basis or basis of articulation’ (p. 74).
8
the 1960s, when Honikman’s article was published and others followed. By this time, the earlier phoneticians who had written about the subject of AS were as good as forgotten and since Honikman did not make any references to earlier sources in her article, later work on AS often contains only references to Honikman’s article and not to Sweet or any of the German phoneticians (Jenner, 1997; 2001). 1
2.2 A renewed interest
Although by now it has become quite clear that Beatrice Honikman did not herself develop the idea behind the term ‘articulatory setting’, she certainly deserves the credit that is given to her. In addition to coining the term ‘articulatory setting’, she was, as Jenner argues,
responsible for ‘restoring [the idea] to respectability’ (1997: 31), and, according to Collins and Mees, for reviving academic interest in the idea of AS in Britain as well as in other
Western countries ‘in relatively recent times’ (1993: 2). After Sweet’s works on organic basis, the focus of articulatory phonetics in Britain had switched to a segmental approach, and Honikman’s essay was to be the beginning of a new period of British, Western-European and American interest in AS.
2.2.1 Articulatory setting and voice quality
The first British phoneticians to pick up on the idea of AS put forward by Honikman were David Abercrombie, one of the leading phoneticians of the twentieth century, and his student John Laver in their works on the broader subject of voice quality. In his 1967 publication Elements of General Phonetics, Abercrombie starts his chapter on voice quality and voice dynamics by stating that up until that time phonetics had been ‘occupied almost exclusively’ with the segmental aspects of speech (1967: 87), but that it was about time that the non-segmental components of speech were recognized. He then recognises voice quality as one of these non-segmental components and, in contrast to the much shorter-period segmental articulations, describes it as ‘those characteristics which are present more or less all the time that a person is talking: it is a quasi-permanent quality running trough all the sound that issues from his mouth’ (1967: 91). Abercrombie then refers to Honikman’s article which shows that he thinks AS to be a component of voice quality. Later, Laver (1980; 1991; 1994) explores
1
9 the subject of voice quality, ‘the characteristic auditory colouring of an individual speaker’s voice’ (1980: 1), and the idea of AS more extensively.
Laver argues that a speaker’s voice quality is built up of two factors, the intrinsic (organic) and extrinsic (phonetic) features. Intrinsic features are those features that derive from the individual speaker’s anatomical build up. They are the ‘invariant, absolutely
uncontrollable physical foundation of the speaker’s voice apparatus’ (Laver, 1991: 163). They include, for example, the length of the vocal tract and the size of the mouth and articulators. According to Laver, extrinsic features of voice quality are ‘to do not with the nature of the vocal apparatus at a speaker’s disposal, but of the use to which he puts it’ (1980: 10). These are the features of voice that are sometimes consciously but mostly unconsciously under control by the speaker and that were ‘once acquired idiosyncratically, or by social imitation’ (Laver, 1991: 148).
These extrinsic features of voice quality are what Laver in his early works equates with AS when he states that ‘[e]xtrinsic features derive their auditory quality from long-term muscular adjustments of the intrinsic vocal apparatus’ (1991: 372) and ‘[i]t was Beatrice Honikman (1964) who first gave the name ‘articulatory settings’ to these long-term, extrinsic adjustments’ (1991: 372). In his later works, Laver comes to see AS, also referred to as settings of the supralaryngeal vocal tract, not as the extrinsic feature of voice quality but as one of the extrinsic features of voice quality, together with phonatory settings, articulatory range settings and overall muscle tension settings.
To provide a scientific means of describing the extrinsic features of voice quality of any speaker in such a way that every researcher would be able to understand and use it, Laver developed a model of reference which he called the neutral setting. This neutral setting, he argues, ‘is more properly to be thought of as a constellation of co-occurring settings at different locations in the vocal apparatus’ (1980: 14). And, Laver goes on to argue, it ‘should in no way be confused with any idea of some putative “normal” setting, nor with any concept of the “rest” position of the vocal organs’ (1980: 14). In other words, the neutral setting is by no means expected to be used by a ‘normal’ person when speaking; it is merely an invented situation which serves as a point of reference when describing settings used in a specific language. Modifications to this putative neutral setting can be made at different places in the human vocal tract; Laver distinguishes modifications made to the settings of the
settings. The current study will only be concerned with modifications to the settings of the supralaryngeal tract, the AS.
Laver distinguishes three groups of settings that can be modified in the supralaryngeal vocal tract. Longitudinal settings are modified by shortening or lengthening the vocal tract by raising or lowering the larynx or by protruding the lips. Latitudinal settings are modified by changes in the cross-sectional area of the vocal tract at five different places along its length. Laver distinguishes alterations to the neutral settings of the lips, the tongue (see figure 1 for possible tongue settings.), the faucal pillars, the pharynx and the jaw. Finally, the
velopharyngeal settings can be modified by altering the neutral setting of the velum, causing nasalisation for example. Using Laver’s model as a point of reference, (parts of) the AS of any language can now be described by its deviation from the neutral setting. I.e. French speech is palatalised and speech from certain parts of Lancashire is velarised (Laver, 1980).
Figure 1. A possible range of lingual tongue setting (after Laver 1980: 45)
Source: Collins & Mees (1993)
11
2.2.2 Articulatory settings investigated
After Laver, many researchers have shown an interest in the idea of AS. Initially, most studies on AS were impressionistic accounts, based on perceptual analyses, analytic listening and careful observation. Some of these studies will be dealt with in more detail in section 2.4 below. In addition to these studies focussing specifically on AS, the idea of AS has been applied in several sociolinguistic studies as one means of describing and distinguishing between speech varieties related to social class or dialectal areas. In his study of the social differentiation of Norwich English, Trudgill (1974) believes that many small segmental differences between working and middle class speech can be accounted for by the fact that the two social groups use different ASs. He states, ‘[in] those cases where there are slight
differences in the pronunciation of individual segments, (…), these often seem to be due to the same overall difference in the mode of articulation,’ or the AS, resulting in a distinctive Norwich voice (1974: 185). Other studies in which the idea of AS is used to account for the characteristic colouring of a variety of English are Knowles’ study of Scouse (1978 in Mesthrie, Swann, Deumert & Leap, 2000:72), and Collins and Mees’ study of the Cardiff accent (1988).
In order to prove the existence of language-specific AS and to establish how exactly settings for different languages are different, studies based on more than impressionistic analyses have been conducted as well. Laver (1980) already started to use acoustic measurements in his study of voice quality and phonetic settings to measure the acoustic effects of the use of different settings. More recently, Bultena (2007) established the
effectiveness of vowel formant frequency analyses to measure differences in the ASs used by Dutch-English bilinguals when speaking the two languages. Other acoustic techniques that have been used to investigate phonetic settings are pitch range measures and long-term average spectra2.
However, although the use of acoustic techniques ‘may provide important insights into the underlying configuration of the vocal tract because of the causal relation between
articulation and acoustic phenomena’ (Mennen et al., 2010: 26), it is often argued that with these kinds of measurements ‘one can never be sure whether acoustic differences in the speech signal are the result of differences in AS or differences in segmental targets’ (Gick, Wilson, Koch and Cook, 2004: 221). In other words, AS may be present throughout speech,
2
12
but it always underlies the individual speech sounds used in a language to form words and sentences and when attempting to measure AS during speech these segments may interfere. Therefore, it is proposed that to measure AS directly, without any segmental interference, it should be measured outside of actual speech.
The first study to perform direct measurements of AS was conducted by Gick et al. (2004). They linked AS to speech rest positions, ‘language-specific postures held during non-speech segments between utterances’ (Gick, Bernhardt, Bacscalfi & Wilson, 2008: 311). They assumed that in these speech rest positions the articulators remain in their language-specific AS, because the articulators are still in speech mode even though they are not actually articulating sounds. The speech rest position was proposed to be ‘the most representative, least biased configuration at which to measure the position of the articulators in order to infer a language’s AS’ (Wilson, 2006: 10).
Gick et al. measured the inter-speech postures (ISPs) of a number of articulators with the use of X-ray films of English and French speakers and found that the ISPs of English and French were significantly different and that these language-specific ISPs corresponded closely with at least the impressionistic description of the ASs of the two languages by Honikman. In addition, they concluded that ISPs, or ASs, are as ‘tightly specified as actual speech targets’, the articulators assume the same ISP in every speech rest period, and are not just ‘transition point[s] solely determined by immediately surrounding sounds’ (2004: 228). Later Wilson (2006) partially replicated the study by Gick et al. (2004) by measuring ISPs of English and Canadian monolingual and bilingual speakers using ultrasound imaging and Optotrak3 (a technique that captures and tracks movements of small markers that are glued onto the face of a speaker and allows researchers to follow the movements of the lips and jaw). He also
concludes that the ASs of English and French, ‘observed through the window of inter-speech posture’ (2006: 103), are significantly different and adds that these results ‘provide much-needed evidence to support the teaching of AS’ (2006: 98).
Recently, Schaeffler, Mennen and Scobbie (2010) have trialled different instrumental techniques as well as a number of different speech tasks in order to measure ISPs of German-English bilingual speakers. The techniques they used were electropalatography (EPG, see Schaeffler, Scobbie and Mennen 2008), ultrasound tongue imaging (UTI) and Vicon4, a
3
http://www.ndigital.com/lifesciences/certus-motioncapturesystem.php
4
13 motion capture system. The different speech tasks that were used were map tasks, sentence reading and picture naming. Although their research is ongoing, they have already shown that EPG might not be the best technique to measure AS because it seems to interfere most with the ‘natural’ rest position. In addition, the map task was also found not to be useful, since ‘an unprompted speech task will necessarily allow the speaker to plan ahead. Thus, it’s not just difficult to control segmental context, but also to control the pragmatic function of the
utterance, which might well have an impact on ISPs and their analysis’ (Schaeffler et al. 2010: 35). The other techniques are still under investigation, but Schaeffler et al. argue that ‘it is not a question whether articulatory settings exist, the question is how to best measure them’ (2010: 63).
2.3 Definitions and technical aspects
Although it has proven difficult, but not completely impossible judging from the studies of ISP mentioned above, to measure ASs and even to confirm their existence directly, most researchers have not doubted the assumption that there are indeed such settings. It is also generally agreed upon that ASs are language and even language variety specific, meaning that every language or dialect has its own AS which all native speakers of that language apply when speaking (although small individual differences do exist).
However, what is meant exactly by the term ‘articulatory setting’ differs from researcher to researcher and from account to account. Some researchers understand AS to mean the underlying settings of all speech organs, as Laver did in his early years when writing:
A recurrent feature of this sort constitutes in effect a tendency for the vocal apparatus to be subjected to a particular long-term muscular adjustment (Abercrombie 1967: 93) or ‘articulatory setting’ (Honikman 1964: 73). One example of such a setting would be a quasi-permanent tendency to keep the lips in a rounded position throughout speech. (…) Another would be the persistent choice of a characteristically ‘whispery’ mode of phonation. (1980: 2)
14 overall way in which the speech organs (i.e. lips, tongue, mouth and throat muscles, velum, larynx) are held throughout the speech process’ (2003b: 221). The more general term
‘phonetic setting’ is often used to refer to the settings of the supralaryngeal tract together with the phonation and tenseness settings.
The idea of AS is often mentioned in relation to segments, the minimal speech sounds of a language. Honikman says about AS that it ‘does not imply simply the particular
articulations of the individual speech sounds of a language, but is rather the nexus of these isolated facts and their assemblage, based on their common, rather than their distinguishing, components’ (1964: 73). So, while the individual sound segments of a language are produced by short movements of the articulators and are defined by their differences, the underlying AS, which is present in all utterances of a speaker, is used on a permanent basis and is defined by its common characteristics.
However, although AS is not about individual segments, it was generally assumed that the ‘articulatory setting of a language is determined, to a great extent, by the most frequently occurring sounds and sound combinations in that language’ (Honikman, 1964: 76). Thus, it was thought that the phoneme inventory used in a language determines, for a large part, the AS of that language. In other words, ‘individual segments are seen as being articulatory related to other segments in that a particular articulatory feature could be abstracted from the chain of segments as a shared property of all or most of the segments’ (Laver, 1980: 2). However, seeing AS as resulting from the individual speech sounds used in a language poses a problem. There are languages, and language varieties, that have very similar phoneme inventories but that still sound very different, because of their different ASs. When AS would be a by-product of the phoneme inventory used in a language, it would be expected that languages that have similar phoneme inventories would have similar ASs as well.
Recent research by Gick et al. (2004, discussed above) has linked the AS of a language to its inter-speech posture (ISP). This ISP is a relatively fixed posture that the articulators assume when they are not articulating individual segments but are in between utterances. The fact that a language’s AS is active during but also outside actual speech does not directly mean that it cannot be related to the most frequently used speech sounds occurring in that language, but it does open up other possibilities. Gick et al. argue that ‘if AS is an
15 information associated with [a] language’ (2004: 228). Since ISPs do function as specific targets and languages that have similar phoneme inventories do not usually have similar ASs, it does not seem likely that AS is purely functional and derived from segmental articulations, as Laver suggests when he envisages AS as the picture that emerges when all individual articulations are superimposed on each other (1980: 13).
2.4 Articulatory setting in education
It has become clear that the idea of AS has been explored in a number of different ways and has been applied in different fields of linguistic study. Phoneticians such as Abercrombie and Laver have explored the idea of setting in relation to their more general descriptions of voice quality. Others, such as Trudgill (1974) and Collins & Mees (1988) have used the idea of AS in their (sociolinguistic) studies of dialectal areas. In addition to the applicability of AS in these fields, another field in which the idea of a language specific setting has often been explored is the field of L2 acquisition. In his Primer of Phonetics, Sweet already stated that ‘[a] knowledge of the organic basis is a great help in acquiring the pronunciation of a language’ (1906: 74).
After Sweet, there have been a number of linguists who have explored the applicability of AS in education, as it fits perfectly in the holistic view of language that prevails at the moment and the top down approach of instruction that accompanies that view. Most of these researchers have written impressionistic descriptions of the settings of different languages and have compared settings in order to aid L2 learners who were struggling to get their pronunciation right. In more recent years some empirical studies have been conducted to establish the effectiveness of the use of AS in L2 pronunciation teaching. In this section, some of the explorations of AS in relationship to L2 teaching are discussed.
2.4.1 Impressionistic descriptions
16 textbooks, not as events in a moving continuum but as a manifold of detached articulations’ and therefore the student ‘so often fails to co-ordinate them satisfactorily’ (1964: 74).
In order to improve their pronunciation of English, Honikman’s students had to focus not on the segmental but on the supra-segmental differences between French and English. One of these segmental aspects of pronunciation is AS (other important
supra-segmental aspects of pronunciation are rhythm and word stress). The students had to let go of their native French AS and instead use the British English AS that is necessary to pronounce words in a British English way. Honikman states that:
[a]ll languages do not have identical articulatory settings: whereas one language may resemble another in this respect, others may differ considerably. Where two languages are disparate in articulatory setting, it is not possible completely to master the
pronunciation of one whilst maintaining the articulatory setting of the other. (1964: 74)
The students’ use of their native French AS to coordinate the English sounds they were making led to an un-English pronunciation.
In order to teach her students to use the British English instead of the French AS when speaking English, Honikman worked on an extensive contrastive analysis of the settings of both languages and came up with a list of advice for learners based on the most important differences between the settings of the two languages. For example, she advised her French students when speaking English to taper their tongues and still their lips. In her article, Honikman writes that she developed a number of exercises and drills to help her students modify their AS and that she used these techniques repeatedly to practise pronunciation. Unfortunately, she does not specify or give examples of what kind of exercises and drills those were.
17 sounds of a language either. Honikman ends her account with an advice to ‘establish the setting first, then the details of articulation’ (1964: 83).
Building on the descriptions of settings by Abercrombie, Laver and Honikman, Esling and Wong (1983) discuss the use of voice quality settings (of which AS is a part) in the teaching of pronunciation and give a concise description of the settings of American English. They immediately emphasise that ‘[w]hen a feature of voice quality figures prominently in the setting of an ESL student’s native language but does not occur commonly or to the same degree in English, it is a potential obstacle to intelligibility’. In addition, using the ‘wrong’ voice quality settings may result in lower social judgements (1983: 90). Thus, they argue (as Honikman did earlier):
[i]t may be that a segmental approach is not the most efficient way of introducing pronunciation in a second language, since if focuses on the specific rather than first directing attention to the general characteristics of accent. Whereas the child learning his/her native language acquires setting and segmental features as a mutually
combined system, the second language learner may impose the new phonemes of English on the old background posture of a non-English, and perhaps inappropriate, voice quality setting. (1983: 90)
They then propose a descriptive model of the American English setting which should
‘increase awareness of setting in English’ (1983: 91) and aid teachers in teaching L2 learners about the use of language-specific settings to improve their pronunciation.
In addition to discussing voice quality settings in relation to teaching and describing the American English setting, Esling and Wong mention a number of techniques that might help students to become aware of the settings they use when speaking their L1 and to change these settings into those of their L2. In one of their methods to raise awareness, students with different L1s have to say something in their respective L1s so that they may perceive
18 Two researchers who have contributed considerably to the subject of AS with their descriptions of the settings of languages and accents such as American English, RP English, Cardiff English, Standard Dutch, and Danish are Collins and Mees. In addition to giving impressionistic descriptions of the ASs of these languages, Collins and Mees have stressed the usefulness of employing the idea of AS in the teaching of L2 pronunciation. By writing
contrastive analyses of the settings of pairs of languages, they have attempted to aid L2 learners in effectively learning the pronunciation of their L2. After describing the settings of a pair of languages, for example RP English and standard Dutch in The Phonetics of English and Dutch (2003b), Collins and Mees give a list of advice to the L2 learners in order to optimise their L2 AS and, to take up the example, to make sure that a Dutch learner actually sounds English when speaking English.
Collins and Mees emphasize that because AS is language-specific, advice given to L2 learners will have to be language-specific as well. They argue as follows:
[s]ince AS varies from language to language and from accent to accent, advice to learners acquiring a second language will depend crucially on the differences between the AS of the particular variety of their first language (L1) and the pronunciation model of the target language. (Collins & Mees, 1993: 7)
Consequently, two different sets of instructions will have to be given to learners with different L1s when learning the same L2. For example, while Danish learners of American English would be advised to ‘[a]dopt generally tenser setting of the body of the tongue, with firmer closures and narrowing for stops’ (Collins & Mees, 1993: 7), Dutch learners of American English would be advised to do the opposite and make ‘looser contacts for plosives, particularly in order to achieve aspiration for initial [p t k]’ (Collins & Mees, 1993: 8).
Collins and Mees argue that teaching L2 learners the AS of the language they are learning is a more effective and economic way of teaching L2 pronunciation and often produces better results than an approach based solely on the production of segments (Collins and Mees, 1993; 2003b). Modifying an AS means that all segmental articulations are
19 of the target language’ (Collins & Mees, 1993: 9). In other words, instead of learning all the small differences between the segmental articulations of an L1 and an L2, time can be saved by learning the AS underlying those segmental articulations. Thus, Collins and Mees argue, ‘[w]hilst AS prescriptions cannot ever replace all traditional teaching techniques based on segmental contrastive analysis, we believe that they can nevertheless be a very valuable aid to expediting such procedures’ (1993: 10).
Although Collins and Mees do not accompany their impressionistic descriptions with practical information about how to teach or learn a new AS, in the form of exercises for example, they state that, ‘[j]ust as in any other form of articulatory pronunciation training, the ease of providing effective AS teaching prescriptions varies greatly according to the location within the vocal tract’ (1993: 9). The different settings of the lips and jaw can easily be observed by teachers and learners alike and are therefore not so difficult to teach or learn. The settings which are not as easily observed, however, are the settings of the tongue and other internal speech organs. The further back in the vocal tract the speech organ is located, the more difficult it is to get a good sense of what exactly is going on with the articulators and to come up with effective exercises (Collins and Mees, 1993).
At the University of Amsterdam, the Netherlands, Van Buuren (1990) developed a system of generative phonetic rules for English which he used in teaching his students about the pronunciation of English, theoretically and practically. In addition to treating other aspects of pronunciation such as air stream, pitch and phonation, he pays attention to the English AS and problems that Dutch students might face when speaking English. Through a system of rules, Van Buuren describes the AS of British English in every possible context and with that provides his students with a step by step guideline of how to use their articulators when speaking English. Through several analyses of the settings used by other speakers and much practice with the rules, his students learn to control their English pronunciation.
Unfortunately, although Van Buuren states that ‘teaching practical control of all the individual phonetic features (…) can be done quite quickly and efficiently by an experienced teacher’ (1990: 34), he also does not give any examples of exercises or techniques.
2.4.2 Techniques to teach AS
20 watch videos of people speaking different languages after which they have to identify AS differences and exercises in which languages are purposefully spoken with a ‘wrong’ AS so L2 learners may perceive the effect of an incorrectly used AS. Wrembel describes more specific techniques used to practise a new AS, such as ‘speaking with a light object placed between the lips’ (Wrembel, 2001: 63) to indicate the relatively small opening of the jaw and lips in English, ‘exaggerating the length of long open vowels and diphthongs produced in context’ (Wrembel, 2001: 63) to achieve more muscular laxity, and ‘practising rhythmic sentences with frequent /t/, /d/ or /n/ sounds to activate the tip of the tongue to make contact with the alveolar ridge’ (Wrembel, 2001: 64).
Another technique that might prove useful to the teaching of AS is the use of ultrasound tongue imaging (UTI) to provide direct visual feedback on pronunciation.
Although this technique has not yet been applied to the teaching of AS, some successes were booked when the technique was used to improve learners’ productions of a number of speech sounds such as /r/ (Gick et al., 2008). In addition, the technique is currently used to
21 .
2.4.3 Empirical evidence
Having dealt with the most important impressionistic accounts of AS in relation to L2 pronunciation teaching and some techniques that might prove useful in the teaching of AS, it is time to explore some applied studies into the teaching of AS. However, although many impressionistic accounts are available that advocate the use of AS in pronunciation teaching, only few empirical studies into the teaching of AS have been conducted, of which only one5 is a quantitative study
Jones and Evans (1995) describe a project in which they taught English pronunciation to groups of Cantonese speaking students with the use of AS. AS was first used to train students to convey and recognize different emotions in English rather than to distinguish between their L1 and L2 pronunciations, but was later also used to sensitize the students to the differences in settings between their L1 and intended L2. The results of this method of
teaching English pronunciation were perceived by the researchers to be positive. However, since there is no objective evidence for their assumption, this study is no more helpful in assessing the use of AS in education than the earlier impressionistic accounts.
Another study into the use of AS in pronunciation teaching was reported on by Kerr (2000). She describes a case study in which one speaker of Cantonese underwent a number of training sessions to improve his intelligibility in English with the use of AS. This study also showed a positive perceived outcome evident from Kerr’s conclusion that ‘this case study suggests that, even in the presence of numerous segmental errors, changes to articulatory setting may produce a noticeable improvement in intelligibility’ (2000: 11).
The first quantitative study into the use of AS in pronunciation teaching was
conducted by Mompeán González (2003). The study was set up to ‘provide evidence on the usefulness of an AS training programme’ (Mompeán González, 2003: 1604), using a
combination of techniques (found in the literature) for the teaching of AS. Seven female first-year students of English Philology participated in the study. The participants were asked to listen to an English text, after which they were presented with a written copy of the text and were recorded reading it. Then they had to attend 10 training sessions in which different techniques were applied to teach them about the English AS. These sessions took place in the space of one month and took up 18 hours in total. After the participants had completed all 10 training sessions, they were again recorded reading the English text. Three native speakers of
5
22 English then rated the pre and post instruction recordings on ‘degree of native accent’, ‘degree of good pronunciation’ and ‘degree of foreign accent’ (Mompeán González, 2003).
The outcomes of the study were statistically analysed and showed that the Spanish students had obtained a more native like and less foreign accent as well as a better
pronunciation. Mompeán González concludes that although it is not possible to evaluate the effectiveness of the different techniques for the teaching of AS, the combination of techniques brought together in the training programme used in this experiment has proved its value. ‘This suggests that AS training programmes can actually benefit students in their attempt to
improve their spoken language and the image they project of themselves,’ which many researchers before declared, but which was not quantitatively tested until now (2003: 1605). However, since Mompeán González did not use a control group, the improvement in
pronunciation proficiency of the Spanish students might be due to other factors as well.
2.5 Present study 2.5.1 Rationale
Previous impressionistic accounts of AS have often advocated the use of AS in the teaching of L2 pronunciation and claimed that the teaching of AS to L2 learners might be more effective and more economical than the teaching of individual speech sounds. However, apart from describing some of their success stories, those theoretical accounts have not come up with any actual evidence to prove their claims. A small number of empirical studies have come up with some tentative evidence that the teaching of AS to L2 learners can improve their
23 2.5.2 Pronunciation in Dutch secondary education
Derwing, Munro, and Wiebe state that the ‘interest in pronunciation instruction for ESL learners has increased dramatically in the last decade’ (1998: 394). However, in the Dutch secondary school system, this does not seem to be the case. It appears that with the transition from a focus on correctness to a focus on communication in the teaching of foreign languages, the interest in pronunciation has almost disappeared. In the Dutch goals for teaching foreign languages at the lower secondary school, the greatest importance is given to the receptive skills; reading and listening (Staatsen, 2007). When the productive skills (speaking and writing) are concerned, the focus lies on the pupils’ abilities to make themselves understood by listeners/readers. In speaking, there is only little attention to pronunciation. Indeed, the only mentioning of pronunciation in the attainment targets for conversation skills valid from 1998 is that ‘at the end of the lower secondary school students are required to express
themselves in such a way that their intention can be understood without great effort; a certain degree of correctness of pronunciation is necessary’ (Koet, 2007: 19).
What exactly is meant by ‘a certain degree of correctness’ and how to achieve that ‘goal’ as well as what level of intelligibility is targeted is left open for every secondary school to decide for itself. In practice, there is little time to focus on pronunciation and if any specific attention at all is paid to pronunciation it often involves the teaching of segments such as the th-sounds, while in order to improve comprehensibility or intelligibility the teaching of suprasegmental features, such as prosody and stress patterns, has proven to be more useful (see for example Derwing et al., 1998). Munro argues as follows:
because teachers and learners do not have unlimited time for instruction in
pronunciation (any more than in any other skill), it is important to establish a set of priorities for teaching. If one aspect of pronunciation instruction is more likely to promote intelligibility than some other aspect, it deserves more immediate attention. (Munro, 2008: 197).
24 2.5.3 The articulatory settings of Standard Dutch and British English
The AS of Standard Dutch and especially that of British English have been discussed in depth several times in many of the impressionistic accounts of AS. In a few of these studies both the AS of Standard Dutch and that of British English were described as well as contrastively analysed. This section will combine a number of previous descriptions of the ASs of both languages and will provide a list of most important setting differences as well as one with advice for Dutch learners of British English and will be used as a frame of reference in the remaining part of this study.
With regard to what Honikman calls the external setting, the settings of the lips and jaws (1964), most researchers agree that in British English they have a quite neutral setting. The lips are loosely spread (Collins & Mees, 2003b) and usually rounded only to produce specific sounds such as /w/. They are usually ‘slightly and loosely apart’ (Honikman 1964: 75) and not very active. Hence the use of the slightly exaggerated term stiff upper lip to describe the articulatory movements of an Englishman (Lowie, 2004) and the observations made by many learners that the English do not open their mouths when they speak. The setting of the jaws is usually described as closed, but loosely so, without any tension (van Buuren, 1990; Jenner, 1987).
The internal setting of British English is also characterised by a laxness of the speech organs. Honikman describes the tongue to be ‘tethered laterally to the roof of the mouth by allowing the sides to rest along the inner surface of the upper lateral gums’ (1964: 76). The setting of the tongue root is quite neutral and lax; it is little used and when used only for back vowels and velars. But the tip of the tongue is very active, vigorously moving upwards and downwards to the alveolar ridge to form the frequently used alveolar consonants /t, d, n, r, s, z/. According to van Buuren (1990), the tip of the tongue is tapered, or pointed, resulting in a small area of contact between the tongue tip and the alveolar ridge (Collins & Mees, 2003b). Using Laver’s range of possible lingual tongue settings on page 10, the tongue setting of British English can be characterised as 2 + 5, ‘predominantly alveolarisation plus a degree of velarisation’ (Collins & Mees, 1993: 6). Findings from the study by Gick et al. (2004) support the above mentioned impressionistic descriptions of the English tongue setting (although these findings are based on the Canadian English AS, which differs somewhat from the British English AS). The muscles of the pharynx are relaxed and the larynx is usually lowered somewhat (Collins & Mees, 2003b; Jenner, 1987). Although Laver states that nasality
Collins & Mees (2003b) and Jenner (1987) argue that a rather firm velic closure is part of the British English AS.
Although not nearly as often described as the AS of British English, the AS of Standard Dutch has also been discussed by phoneticians such as Collins and Mees (1993; 2003b) and Jenner (1987). The external setting of Standard Dutch is usually described as tense. The lips are ‘held firmly and drawn back at the corners’ (Collins & Mees, 2003b). They have inner rounding and can almost be described as being pursed (Lowie, 2004). The jaws are slightly open, but again very tense (Jenner, 1987).
With regard to its internal setting, Standard Dutch can also be characterised as rather tense. In contrast to British English, Standard Dutch has a tense and ‘blunter lingual setting’ (Collins and Mees 2003b: 221); the blade or front of the tongue is used while the tip is usually inactive (see figure 2). In addition, the root of the tongue is somewhat drawn back and quite active; Collins and Mees argue that the ‘centre of gravity for tongue activity is posterior’ (2003b: 222), resulting in a setting which can be described as 6/7 + 3,
uvularisation/pharyngealisation + palate-alveolarisation, on Laver’s range of tongue settings on page 10. The setting of the throat muscles is tense, ‘often with some pharyngeal
constriction’ (Collins & Mees, 2003b: 221) and the larynx is usually raised. In contrast, the setting of the velum is lax, resulting in ‘semi-continuous background nasalisation’ (Collins and Mees, 2003b: 223).
Figure 2. The tongue settings of British English and Standard Dutch (Collins, den Hollander & Rodd,
1987)
The most important differences between the settings of British English and Standard Dutch can be found in table 1 below.
26 Table 1
The ASs of British English and Standard Dutch compared
BRITISH ENGLISH STANDARD DUTCH
Jaws loosely closed slightly open, but tense
Lips loosely spread, not very
active
inner rounding, tense
Tongue root neutral + lax
tip tapered + very active
root drawn back + active tip inactive
blade/front active
Larynx lowered raised
Velum firm closure lax
Based on these differences between the British English and Standard Dutch settings, a list of advice for Dutch learners of English can be put together. Dutch learners should attempt to slightly spread their lips when speaking English (Lowie, 2004) and only round and protrude them for specific sounds such as /w/ (Collins and Mees 2003b). Furthermore, they should adopt a tapered tongue setting for tip articulations as well as relax the tongue and lips (which makes it easier to produce sounds such as /S/ and /C/ and to achieve aspiration for initial voiceless stops) (Collins et al., 1987; Collins and Mees, 2003b). Finally, learners can be advised to avoid retracting their tongue root into their pharynx (to improve their production of British English dark [4] and /P/) and to lower the larynx (Collins & Mees, 2003b).
2.5.4 Rating pronunciation
27 Currently, however, it is believed that although a (heavy) foreign accent in an L2 learner’s speech may interfere with a listener’s understanding of that speech, it does not necessarily have to (Munro & Derwing, 1995a; Van Heuven, 2006). Thus, whether a speaker is intelligible or not depends on far more than just accentedness. Because the aims of L2 teaching have shifted from a focus on correctness to a focus on communicative competences, at least for most learners, it is now recognized that improving a speaker’s intelligibility is a more important initial goal of pronunciation training than improving his/her foreign
accentedness. Therefore, by improving pronunciation is usually meant improving
intelligibility. In any case, it has to be emphasised that a speaker’s foreign accentedness and his/her intelligibility are not mutually related and therefore cannot be measured
simultaneously.
Related to the issue of what exactly is meant by improving pronunciation, is the discussion of how an improvement in pronunciation might be measured. Munro discusses three approaches to speech assessment, namely ‘(1) responses from unsophisticated listeners, (2) impressionistic analyses from expert evaluators, and (3) acoustic phonetic analyses’ (2008: 200). Munro prefers the first and argues that ‘unsophisticated listeners’ judgments are especially important because they may provide insight into how understandable L2 speakers are when they interact with other members of their community’ (2008: 200). Expert
evaluators (such as teachers and linguists) that are familiar with the speech of L2 learners may assess their speech differently than outsiders. They may, for example, be better able to
identify problem areas, but they may also understand L2 utterances that other, not
phonetically trained, listeners do not and they may not be able to estimate in what degree an L2 learner’s speech utterances are intelligible to an unsophisticated listener (Munro, 2008). The use of acoustic analysis is also problematic as a way of assessing pronunciation, since instrumental measurements may ‘reveal differences between native and non-native speech that are not noticed by listeners’ (Munro, 2008: 200). Munro therefore argues that, ‘[f]rom the standpoint of communication, there is no useful way to assess accentedness and
comprehensibility, except through listener responses of some sort’ (2008: 200).
These listener responses can then be obtained in several ways. Usually the foreign accentedness of an L2 learner’s speech is assessed by listeners who rate a speaker’s
28 foreign accent’ (1) to ‘very strong foreign accent’ (9). The perceived comprehensibility of an L2 learner’s speech can also be assessed by listener ratings. However, as Munro argues,
rating data are of limited use in evaluating how much comprehension has actually taken place, because listeners sometimes mistakenly believe that they have understood an utterance, and may therefore rate it highly, when they have not understood it well at all. (2008: 200-1)
Consequently, researchers have come up with different ways of assessing
intelligibility, ‘the degree to which a speaker’s utterance is actually understood by a listener’ (Munro, 2008: 200). In the past, listener tasks were used in which listeners had to find pictures corresponding to L2 speakers’ utterances, complete cloze tests and answer comprehension questions (Munro, 2008). These tasks were then scored by the researchers after which the level of intelligibility was established. A now common way to assess intelligibility is the dictée task, in which listeners make orthographic transcriptions of L2 speakers’ utterances. These data can then be scored for total numbers of words that are transcribed correctly, percentages of key words that are recognized, or number of deviations between the transcripts and the intended utterances, among others.
2.6 Statement of purpose
29 METHOD
Two training programmes, one based on AS differences between Standard Dutch and British English and one based on segmental differences between the two languages, were developed to teach two classes of secondary school pupils about the correct pronunciation of British English in relation to Standard Dutch. In order to test whether the teaching of the British English AS actually improved the pupils’ English pronunciation, pre and post instruction tests were set up in which all participants were asked individually to describe a painting in English. Short audio fragments were composed by cutting and pasting words and sentence fragments from each of the recordings made during the pre and post tests and were sent to native speaker judges who rated these audio fragments in a pronunciation judging task.
3.1 Participants
The experiment was started with 44 pupils (20 female and 24 male) in two Havo-two classes at a secondary school in the Netherlands (the Hondsrugcollege in Emmen). The pupils were told that participating in the experiment was not obligatory, but that they did have to follow the four instruction sessions. Participants decided to take part in the experiment voluntarily and were not paid for their participation. The choice was made to use pupils at the beginning of their secondary education, because it was expected that they would have a basic knowledge of English, they would at least have had one and a half years of English at their secondary school and probably some before, but would still be relatively open to a new method of pronunciation teaching. The classes were each made up of 22 pupils with ages ranging from 13 to 15. At the time of the experiment, both classes were taught English by the same teacher. At the start of the experiment, the two classes were randomly chosen to function as either test- or control group. The test group consisted of 15 male and seven female pupils and the control group consisted of nine male and 13 female pupils.
Because, in order for the experiment to be valid, all participants had to be present at six of the six meeting times during the experiment, a number of pupils dropped out during the experiment due to illness, another kind of absence, or unwillingness to participate in part of the experiment. In addition, because all non-native speakers of Dutch were filtered out because of possible interference of their L1s, three more pupils were lost as participants. Consequently, 27 of the 44 pupils that started the experiment also completed it and are
30 and 10 male) completed the whole experiment and in the control group 14 pupils (eight
female and six male) completed the whole experiment. All participants were native speakers of Dutch and stated to have begun learning English as a second language at ages ranging from six to 12.
3.2 Materials
3.2.1 Picture description task
Before starting the actual experiment, a small scale pilot study was conducted in which two possible methods of collecting speech material from the pupils were tested. The methods that were tested were a delayed repetition task and a picture description task. During the pilot study it became clear that in the delayed repetition task each pupil produced an equal amount of speech material and that variables could be carefully controlled. However, the speech material elicited from the pupils in this task was unnatural and can probably be defined as better than their normal English output because of they had the opportunity to imitate the speech model.
Because the objective of the present study was to find out whether the teaching of AS would improve the pupils’ English pronunciation, and the speech output elicited with a delayed repetition task might not be representative of the pupils’ actual English pronunciation skills, it was chosen not to use this kind of task in the experiment. Instead, it was opted to use a picture description task, because the pilot study showed that such a task forced the pupils to form their own sentences from words that they were familiar with, resulting in more natural speech output and pronunciation patterns.
However, the pilot study showed that there are also a number of downsides to using a picture description task. Firstly, by making the pupils describe some sort of picture in their own words, they may produce ungrammatical speech which might very well influence pronunciation ratings. Secondly, in a picture description task the speech output of the pupils cannot be controlled as easily and as effectively as in a delayed repetition task. This might result in very different, incomparable, outputs from the individual pupils. And thirdly, the use of a picture description task might not provide enough data from each of the pupils, since pupils might not be able to describe certain parts of the picture because their, active, vocabulary is not large enough.
31 tackle the first problem, it was decided to cut and paste together short fragments of the pupils’ utterances before sending them to the English native speaker judges who would rate the pronunciation of these utterances. Now, any grammar mistakes present in the recordings of the pupils’ picture description tasks could be excluded from the sample and would therefore not be an issue any more. The fact that a picture description task elicits very different
utterances from different pupils does not seem to be a serious problem since the utterances do not have to be compared between pupils. And even though the pronunciations of the
individual pupils do have to be compared at different times, the utterances do not have to be the same, because the focus of this study lies not on specific speech sounds, but on the underlying setting which should be present continuously throughout the pupils’ speech. In order to stimulate the pupils to say as much as possible about the picture used in the task, it was decided to provide the pupils with a list of words of objects on the picture (appendix I). This list could be used to refresh their memories and give them the opportunity to use words that they were passively familiar with, but could not actively enter when describing the picture.
The picture that was used to elicit English speech from the pupils both during the pre- and post test was a painting by Pieter Brueghel titled the Massacre of the Innocents (1565-7) (appendix II). It was colour printed on A3 paper so that it would be clear and easy for the pupils to view and describe. This specific painting was chosen quite randomly because it was available on the internet in a good quality (so that it could be enlarged and still be very clear) and because there is much activity in the painting as well as many objects which could be described by the pupils. Consequently, it was assumed that there would always be something that the pupils were able to describe even when they did not know what to say about most other objects or activities.
3.2.2 Training programme AS
32
subsequently focussing on the most important6 differences between the Dutch and British English settings, namely lip setting, tongue setting and tension settings of the supralaryngeal vocal tract.
The first lesson was designed to give the pupils an introduction to the subject of AS. The lesson started with a short exploration of the meaning of AS, to make the pupils become aware of the idea of AS. Secondly, an exercise to discuss the Dutch and British English settings was designed on the basis of short YouTube films of the Dutch prime minister speaking Dutch, the Prince of Wales speaking British English and actress Amy Walker doing a number of accents. A last exercise was developed to shortly explore the function of the speech organs and the process of making speech sounds. A homework exercise was added to this lesson in which the pupils were asked to pay attention to and write a short report on the ASs that were used by speakers of different languages on television.
The second lesson was designed with a focus on the British English lip setting.
Exercises were designed in which the students were stimulated to distinguish the different lip settings used in English and Dutch by observing short YouTube films of Prince Charles, David Beckham and Prince Harry speaking English, by observing a partner speaking Dutch and by self observation through little mirrors. Other exercises to explore the British English lip setting were borrowed from the accounts of Mompeán González (2003) and Wrembel (2001), i.e. putting the fingertips on the corners of the mouth to follow the slightly moving and relatively closed lips, and speaking with a pencil loosely held between the lips. In order for the pupils to become really aware of the differences between the British English and Dutch lip settings, an exercise in which the lip setting of English was used to speak Dutch and vice versa was added after Lowie (2004). A last exercise was added to the lesson so that the pupils could practise what they had learned; the pupils were asked to read a short text in pairs and later in class while maintaining an English lip setting.
The third lesson was designed with a focus on the British English tongue setting. Exercises to explore the differences between the Dutch and British English tongue settings as well as a number of exercises to practise using the tip of the tongue when speaking English were used. One exercise by Wrembel was included in which rhythmic sentences with many /t/, /d/ and /n/ sounds were used in order to ‘activate the tip of the tongue to make contact with the alveolar ridge’ (2001: 63-4). In a subsequent exercise a number of tongue twisters with the
6
33
same sounds had to be uttered. Halfway through the lesson, an exercise to raise pupils’ awareness of the effects of the use of different ASs when speaking English was added. The last few exercises were aimed at letting the pupils practise using the English AS when
speaking English. They were advised to start speaking English with the right AS by repeating Missisipi three times (Lowie, 2004).
The last of the four lessons on AS was designed with a focus on the tension settings of the speech organs. The first exercise was meant as a warming up and relaxation exercise for the muscles of the speech organs. In the second exercise, the length of a number of long open vowels was exaggerated since ‘long vowels in English are closely connected with muscular laxity’ (Wrembel, 2001). Another exercise was added to relax the muscles of the lips and tongue when saying the names Piet, Kim and Tom after an exercise in Collins and Mees (2003b: 166). Again an exercise to raise the pupils’ awareness of the differences between different ASs was added in which a number of audio fragments of speakers of different L1s (Scottish, French, British English, Australian, Indian and Italian) speaking English and one English person speaking Dutch were used. The remaining part of lesson four was designed to practise the complete English AS in some speaking exercises.
3.2.3 Training programme segments
A second training programme, based on the teaching of segments, was developed in order to teach the pupils in the control group about British English RP pronunciation in a more traditional way. This training programme (appendix IV) was also built up of four 45 minute lessons. They were designed to treat a number of important7 segmental differences between Dutch and British English. The first lesson, as in the AS training programme, was designed to give the pupils an introduction to the subject of segments. One exercise was designed to explore the idea of segments and sounds and a following exercise treated the phonetic alphabet of British English with the use of a list of sounds and magnetic phonetics8 (Cascadilla Press).
The second lesson was designed with a focus on the English th-sounds /S, C/. In the first exercise of this lesson, however, the production of speech and possibilities of the speech organs were treated. After that an exercise to distinguish /S/ and /C/ from /t, d, s, z, f, v/,
7
Based on lists of gravity of errors by Collins et al. (1987); Gussenhoven and Broeders (1981); Collins and Mees (2003b).
8
34
which are used by Dutch people to represent /S/ and /C/, was added using descriptions of these segments by Collins and Mees (2003), small mirrors for self observation and a number of animations from the University of Iowa Phonetics Flash Animation Project9. An exercise to practice the production of /S/ and /C/ as well as an exercise in distinguishing different segments were added to the lesson.
The third lesson of the segment training programme was designed with a focus on the English /z/ and /D/, which are often both pronounced as /D/ by Dutch speakers, because /z/ is not used in Dutch. The first exercise was designed to make the pupils aware of the difference between /z/ and /D/ and the fact that Dutch people usually do not use these sounds correctly. Hereafter an exercise to practise the production of /z/ and /D/ was added using elements from exercises by Catford (2001: 122) and Gussenhoven & Broeders (1981: 53), and the use of small mirrors. An exercise to distinguish between the sounds was added as well as an exercise to practice producing a number of minimal pairs containing /z/ and /D/. A last exercise was used in which pupils had to give and receive directions using a map with minimal pair street names after Sicola (2009).
Lesson four was designed with a focus on plosives (/p, b, t, d, k, g/) used in word final position. Because in Dutch all plosives at the end of words are voiceless, Dutch people often do not use voiced plosives when speaking English either. Exercise one was designed to remember the pupils how to make these plosives. The following exercise was developed to make pupils aware of the Dutch and English situations by careful listening and watching. Another exercise was designed to encourage the pupils to search for minimal pairs such as bad and bat. The lesson was finished by and exercise in which pupils were encouraged to make a number of sentences using more than one of the minimal pairs they had come up with earlier.
3.3 Procedures
Pre and post instruction tests were conducted in a small non sound proof room in one of the school buildings of the Hondsrugcollege. This location was chosen for its accessibility, not for its perfect conditions. The pupils individually participated in both the pre and post tests which both took about six minutes per participant. The participants took part in the tests in random order.
9
35 The pre instruction tests were conducted on two subsequent days three weeks before the start of the training programme at the beginning of February 2010. The instructions belonging to the pre test were given in Dutch, to assure the participants’ full understanding of what was expected of them. In short, participants were instructed to describe the painting in front of them in English. They were asked to take their time and describe what happened on the painting as well as what sorts of objects they saw. They were asked to say as much as they possibly could about the painting without paying attention to grammatical structures as the recordings would be cut and pasted into short fragments in which grammatical correctness played no role. Participants were then told that they were allowed to consult a list of English words connected to the painting, but were not obliged to, and were asked to use complete sentences as often as possible. They were then asked whether they had any questions and if they had questions those were answered. After that they were told to switch into English and start talking. When participants stopped speaking after having said too little in the
researcher’s opinion they were encouraged to continue describing the painting until they had produced enough speech material. The participants were then complimented with their performance and thanked for cooperating.
The post instruction tests were conducted on two subsequent days one week after the last sessions of the training programmes at the end of March 2010. The instructions for the post test were similar to those for the pre test. Just before starting the description the participants were told either to use their English AS and start talking (for the test group) or think about what they had learned in the pronunciation lessons and start talking (for the control group). The participants’ descriptions of the pictures both during the pre and post tests were recorded as wav files (44.1 kHz sampling rate, 16 bit resolution) onto a solid state audio recorder (Marantz PMD-670) with an e865 Sennheiser microphone.
36
3.4 Design and analyses
To test whether the AS training programme led to an improvement of the Dutch participants’ English pronunciation, a judging task was set up in which the participants’ pronunciation would be judged on accentedness, comprehensibility and intelligibility. Nine adult native speakers of British English, selected on the absence of a strong regional accent and familiarity with the British RP accent, were found willing to perform the judging task. Four of them were female and five were male. Their ages ranged from 32 to 73. None of them had any
considerable experience with Dutch or Dutch English.
For the judging task, the nine judges and the 27 participants were divided into three groups of three judges and nine participants each. The participants were divided over the groups in such a way that a group consisted of four pupils from one class and five from the other (Group A and C consisted of four pupils from the test group and five from the control group, group B of five from the test group and four from the control group). Now, all of the judges were asked to evaluate four 10-second audio fragments belonging to each of the pupils in their group; in total they would evaluate 36 audio fragments. This measure was taken to ensure that the judges would not have to spend more than one hour approximately on their judging tasks, since they would participate in the judging task voluntarily and would not be paid for their efforts.
The audio fragments were composed by randomly cutting and pasting individual words and parts of sentences from the participants’ picture description tasks using PRAAT software (Boersma & Weenink, 2010). For every pupil, two fragments were composed of parts of the pre test recordings and two of parts of the post test recordings. The 36 audio fragments in each of the judging tasks for the three judging groups were ordered so that the first, 10th, 19th and 28th audio fragments belonged to the same participant and so did the second, 11th, 20th and 29th, etcetera. The order in which the four audio fragments belonging to each of the participants appeared in these positions was randomly assigned.
