Investigating EFL Teachers’ Difficulties in Connected Speech

Diterima: 28-08-2021 Disetujui: 28-09-2021 Abstract: English teachers must have adequtae knowledge of pronunciation like phonetics and phonology. They also must be able to perform a good speaking performance as they must be linguistic models for their students in the class. This study investiagted several phonological aspects of connected speech such as linking, elision, and assimilation by EFL English teachers. The result revealed that assimilation was the most difficult phonological aspects of connected speech rather than linking or elision. The major difference between English and native language of the teachers could be the main reason of their contrains related to connected speech aspects.

To begin with, one of the most complicated processes in communication is pronunciation. As a result, the way people pronounce each word in a sentence must be clear enough to be understood by others since it is a fundamental requirement to perform successful communication. Whether or not the listeners can catch the message spoken by the speakers is mainly determined by how good their pronunciation is even though research studies on the significance of clear pronunciation were quite rare (Jenkins, 2000). However, in recent years, there has been an increasing interest in studying pronunciation. Numerous studies on EFL learners have been conducted in the international context, covering several aspects of learners. A study shows that learners have not paid enough attention in pronunciation because teachers had fewer confidences on practising their pronunciation in the classroom. This may happen because of critical knowledge absence (Saalfeld, 2011). The researcher has only dealt with the learners' point of view rather than the teachers. Although teachers have very important roles as role models, he has not paid attention to teachers' perspectives in much detail.
The study of 28 experienced and well-qualified Uruguayan teachers related to beliefs and concerns about pronunciation teaching shows that most teachers lacked the confidence to teach pronunciation due to several reasons such as lack of training or knowledge. The reason behind their anxiety about using their pronunciation as a model was their status as a non-native English teacher. All these reasons lead to pronunciation being neglected in teaching (Couper, 2016). It indicates that there are several difficulties faced by teachers in pronunciation so research to identify teachers' difficulties in pronunciation will help them to tackle their lack of confidence, knowledge and decrease their anxiety in using their pronunciation as a model.
A study by Wong et al examines the role of the perception of reduced forms (e.g., contraction, elision, assimilation) of English words in connected speech comprehension and the phonological skills underpinning reduced forms perception. Sixty Chinese-speaking undergraduate students were tested with a battery of listening and phonological tasks. The result of this study indicates that the ability to perceive reduced pronunciation variants is important for L2 listening comprehension skills (Wong et al., 2017). It recommends that teachers should have the ability to differentiate the pronunciation between words in canonical forms and reduced forms since they are the role model of pronunciation in the classroom. Meanwhile, a study by Wong et al related to connected speech feature in the English oral production of 60 native Cantonese ESL learners revealed that the successful realisation of connected speech process leads to fluent speech acquisition. The significant difference between L1 and L2 is the main reason for difficulties faced by learners in demonstrating connected speech feature in native-like English speech (Wong et al., 2019). The result of this study proposes the need for a greater understanding of the source of errors by both students and teachers.
Another study examined the effectiveness of connected speech (CS) instruction in improving learners' decoding of spoken English. It follows the trend of a balanced listening session that focuses on both top-down and bottom-up modes. This study indicates that CS aspects pose a serious problem in decoding and segmenting speech for the participants (LAOUBI, 2019). The result of this study gives new insight that ability in decoding spoken English is vital for students' sake in communication.
Consequently, teachers should have adequate knowledge of connected speech. Therefore, investigating teachers' understanding and difficulties in connected speech is worth conducting.
Those studies mentioned above mainly investigate the difficulties related to connected speech features on the learners' side. However, the issue related to the difficulties related to connected speech features on the teachers' side is rare. This study is intended to unravel the difficulties in recognizing and performing connected speech faced by EFL teachers. In other words, this study attempts to listen to the voice of teachers and to seek the challenges that teachers related to practising the features of connected speech.

Pronunciation
Pronunciation is not only related to how to precisely understand and perform individual vowel and consonant sounds but also how the speed, pausing patterns, intonations, and combinations of speech in a spoken language. In other words, pronunciation is the language feature that can determine the mastery of both learners speaking and listening ability (Rezaei et al., 2015). Since the teaching of pronunciation has been often neglected especially when English is considered as foreign language, some teachers argue that pronunciation teaching should be delivered with little or no direct instruction. As a result, the students' fluency is not reached the native or native-like level. This creates difficulties for students to convey the spoken language when they communicate in a casual speech.

Connected Speech
Words often pronounced differently in isolation and when is a part of a sentence. The pronunciation often changes and most of the words pronounced like a long word. The continuous stream of sounds created from pronouncing the words in a sentence, usually without any clear boundaries between words shows the existence of connected speech. In other words, as soon as we open our mouths, have conversation or talk, say something or even give a speech, connected speech is used.
Several things about connected speech need to be considered greatly in spoken language. The sounds may change to become more similar to sounds nearby called as assimilation, add to link words known as linking or intrusion, and even disappear under certain circumstances or elision (Ehrlich & Avery, 2013). Therefore, having such information related to how those two aspects that affect som modification of the stream of sounds, particularly in which way the English native-speakers link a word to one another in either phrase or sentence is mandatory for learners to be mastered. In other words, it is beneficial for both EFL learners' production and comprehension of spoken English when learners have adequate knowledge of these connected speech aspects.

Successful Communication
Being able to understand connected speech is one of the major factors affecting successful L2 spoken text perception and comprehension (Ito, 2006). However, as noted at the outset, having learned the dictionary forms of words pronounced in isolation (Brown, 2012), L2 learners are usually unable to understand authentic language regardless of the amount of formal instruction received for years. This is partly because they do not know how known words sound when they are put together in connected speech or even when they are familiar with the text and the vocabulary (Hagen, 2000). Learners cannot recognize the way sounds, syllables, and words change in spoken discourse because this requires word recognition and lexical segmentation skills. In other words, to be able to perceive and decode spoken input successfully, listeners should be taught how to distinguish word boundaries in spoken words (Kuo, 2012). Therefore, a better understanding of word recognition and lexical segmentation is critical because an inability to segment and recognize words is one of the most commonly encountered problems hindering L2 listeners' decoding and meaning building. A failure to understand connected speech is also considered to be a lexical segmentation problem.

METHODS
Since the aims of the research were to find out the types of connected speech which are difficult to be recognized by EFL teachers, the descriptive qualitative method was considered as the appropriate method because it could describe and interpret about events, condition, or situation of the present. It also provided an in-depth understanding of the ways people come to understand, act, and manage their day-to-day situations in particular settings. In this study, semi-structured interview was employed to gain data related to ELF teachers' difficulty in recognizing connected features.
Purposive sampling was used in the selection of participants in this study, which means that the researcher chose the subjects based on their typicality or specific criteria (Cohen et al., 2000). Fourteen high school teachers were selected for this study. In this case, the teachers who participated were high school English teachers who got a master scholarship from the ministry of religious affair 2019 in State University of Malang (UM). They had at least 2 years of teaching experiences and taught pronunciation to their students either directly or indirectly. The decision to select those teachers since they were the ones who had adequate ability and knowledge in teaching English. As these ten teachers came from different area, it provided more diverse and deep information related to their difficulties in recognizing connected speech. The teachers were in the age range of 21-43 years. All of them were asked to express their willingness to be the participants of the study. Their names were not exposed. Furthermore, the participants came from the various background: cultural background and educational background. They were also varied in terms of gender, kinds of research interest. The researcher believed that those kinds of diversity could form the various pattern of information.
The data of this research was utterance and pronunciation from the 14 English teachers in the form of recorded audio. The recordings from each participant were heard by the researchers to analyse what kind of connected speech feature that they failed to recognize and demonstrate. Later the data derived from the analysis procedure was gathered into some categories to determine the connected speech aspects that hard to be recognized by the participants. The last is the writer will summarize the finding and draw a conclusion of the study.
In qualitative research, validity specifically refers to the appropriateness of the data, while reliability refers to the consistency of data over time, location, and circumstances (Fraenkel & Wallen, n.d.). To improve the validity, the researcher made sure the appropriateness of the data by having member checking and triangulation of the data. The researcher applied member checking by taking the research findings back to the research participants to check whether they found some mistakes or incorrect information from the data. Besides, the data were also triangulated using data source triangulation and theory triangulation. The researcher collected the data from a different type of individuals intending to gain multiple perspectives. Furthermore, the result of the analysis was also compared with the theories or previous studies to generate data interpretation.
However, in the present study, the researcher did not have control to seek for the consistency of the data. Several researchers define reliability clearly as the replicability of research findings and whether they would be repeated if another study, using the same or similar methods, was conducted (Lewis et al., 2003). The idea of seeking reliability in qualitative research is often avoided by most qualitative researchers since the extent to which replication can occur in qualitative research has been questioned.

FINDINGS AND DISCUSSION
To answer the research questions, the researchers divide this section into two issues related to phonological ascpects of connected speech, namely (1) the difficulties faced by EFL English teachers related to phonological ascpects of connected speech and (2) reasons or causes of the phonological ascpects of connected speech.

EFL Teachers' Difficulties in Connected Speech
There were three connected speech features investigated through this research. They were linking, assimilation, and elision. The linking feature had three sub-features. They were linking /r/, linking /j/, and linking /w/. Some people used the term insertion instead of linking. Then, the assimilation feature also had four sub-features. They were assimilation /t/ became /s/, assimilation /t/ became /p/, assimilation /d/ became /n/ and assimilation /n/ became /m/. The elision feature included two subfeatures. They were elision /d/, elision /t/. The term deletion sometimes used to replace the word elision.
The idea of linking happens when the final sound of one word blends into the initial sound of another. Most of the time, when the last sound of a word is a consonant and the initial sound of the next word is a vowel, the term linking occurs. For example, when people say " In an instant", they would connect the final consonant to the next vowel so this /n/ and this /a/ would be said together. In other words, these two sounds were pronounced smoothly without a pause. Meanwhile, when linking also happens when last sound of a word is a vowel and the initial sound of the next word is a vowel. It is also called as insertion or intrusive. As it is very difficult to connect two vowel sounds together, the sounds /r/ /j/ /w/ were added to make it easier to say.
Based on the data displayed in Figure 1, there were no participant who were able to pronounce the linking feature of connected speech such as linking /r/ and linking /w/. The phrase for linking /r/ feature recorded from read aloud activity were " here and " and " idea of ". The phrase for linking /w/ feature was " you enter". As all participants stated that they had unsufficient background knowledge related to English phonetics and phonology, they pronounced these three words relying on their native language. As a consequence, they pronounced these words as exactly as their phonetic transcriptions without applying the connected speech feature. Since there had been no phonological feature like linking /r/ and linking /w/ in their native language, this was appeared as the major reason why did they fail to apply the linking /r/ feature. This result might be explained by the fact that the influence of their native language pronunciation played a major role in their failure to perfom the connected speech feature (Subandowo, 2017). This result also supported the evidence from previous observation (Suliman, 2014) that learners still rely much on mother tongue in their English speech production.

Figure 1. Linking Feature of Connected Speech
Meanwhile, 30% of participants were able to produce intrusive sound /w/ in the phrase " go on". It was somewhat surprising that some participants slightly inserted the sound /w/ to connect the final and initial vowel sounds of the words. The reason for this was not clear but it might be that these participants benefitted from either their experiences or exposures after having communication with English native speakers frequently. As one of the participants had some friends of English native speakers and had in touch with them frequently, this factor might explain why did some participants insert the sound /w/ while pronouncing the words.
On the other hand, 90% of participants were able to apply the linking /j/ feature in the phrase " say it ". They clearly pronounced a very short sound /j/ inserted between the final vowel sound of the first word and the initial vowel sound of the second word. It is difficult to explain this result, but it might be related to the result of previous pilot study (Alameen, 2007) showed that vowel to vowel linking, especially a word ends with final sound /ɪ/ followed by a word with initial sound /ɪ/, was easily pronounced by non-native English speakers. They will spontaneosly inserted /j/ sound to link the vowels sound. Moreover, when a non-native speaker pronounce the phrase " say it " faster a a single stream of sound, they tend to read the phrase as it written. It could be assumed that linking /j/ was easily produced rather than other linking features.
It could be shown from the data displayed in the Figure 2 all participants could not produce the assimilation feature of connected speech such as assimilation /t/ becomes /p/, assimilation /t/ becomes /s/, and assimilation /n/ becomes /m/. The phrase for the following assimilation features recorded from read aloud activity were " that person ", " that side ", and " ten pounds". In accoedance with the present result, previous studies have demonstrated that having acceptable pronunciation was a serious challenge for foreign language learners. This is partly because of the difference of both phonetic and phonological rule between the native language and the targeted language (Rezaei et al., 2015). As all participants pronounced these three words as they were in isolation form, it could be assumed that EFL learners phonetic and phonlogical difference made them difficult to apply the following assimilation features. Futhermore, the lack of knowledge related to assimilation aspect of connected speech as well as the lack of practising them could be the main reasons of the difficulties in performing assimilation (MEGHLAOUI & Meriem, 2017). The finding of the present research indicated a need of training program to improve the pronunciation ability. Meanwhile, 14% of participants unexpectedly succeded to pronounce the words " in the" by assimilating the initial consonant sound /t/ of second word to the final consonant sound /n/ of the first word. This finding was unexpected since regressive assimilation did exist in their native phonological rules. This present finding was consistent with the previous study (Darcy et al., 2007) that advanced English learners succeeded to apply the phonological rules of assimilation with a little interference of their native phonological processes. Therefore, it could be assumed that the participants who were able to perform regressive assimilation have already reached the advanced level of English proficiency.

Figure 2. Assimilation Feature of Connected Speech
On the other hand, 36% of participants succeeded to assimilate the final consonant sound /d/ of the word " good " to initial consonat sound /n/ of the word " night ". As the phrase " good night " was heard and spoken regularly by them in their classroom environment. It could be described as the most possible factor which helped them to apply the assimilation /d/ becomes /n/ successfuly. This finding supported the evidence of the previous study (Brand & Ernestus, 2018) that English learners were able to comprehend and pronounce the words or sentences that they frequently used or heard in daily life with accepted pronunciation. Therefore, it could be assumed that the more familiar the learners with the words, the more better pronunciation they could produce.

Figure 3. Elision Feature of Connected Speech
As seen from the figure 3, that all participants were able to produce the elision /t/ feature of connected speech. The phrase for the following elision features recorded from read aloud activity were " next year ", " last chance ", " most popular". All participants simply deleted the final consonant sound /t/ of each first words from these phrases when pronouncing them. However, the finding of the currecnt study did not support the previous research that confirmed some connected speech features such as linking and elision became the most difficult phonetics problems for EFL learners (Cho, 2021). Moreover, it had been claimed that the acquisition of English /t,d/ elision patterns in word-final consonant clusters has not yet been mastered by second language learners (Edwards, 2011). On the other way around, this current finding broadly had some similarities with the work of other studies that Second language learner of English could recognize the connected speech feature of assimilation and elision more easily when final consonant sound /t/ was in the word (Vidal, 2019). As finding of some previous research supported the current finding and some others were not, future research on the current topic are therefore recommended. Surprisingly, there were two different result of the same connected speech feature of elision /d/. 86% of participants succeeded in pronouncing the phrase " speaks and just" with simply deleted the final consonant sound /d/ from the word " and ". The present finding supported evidence from previous research (Field, 2003) that foreign language learner mostly avoid the final consonant sound which was hard to be pronounced by simply eliding the sound. As the word " and " categorized as function word that was rarely stressed, it would be easier to left out the final consonant /d/ to economize the effort in pronunciation.
On the other hand, there were no participants who were able to delete the final consonant sound /d/ from the phrase " old man ". The final consonant sound /d/ was highly voiced by all participants when they pronounced the word " old ". As the word " old " categorized as content word that was frequently stressed, it would be difficult to elide the final consonant /d/ to perfom elision /d/ feature. This finding was consistent with the work of previous research (Dalton, 2002) that language teachers lacked of background knowledge related to articulary phonetics and phonology. In other words, it could be assumed that they did not delete the final consonant sound /d/ because they relied more on grammatical features than the aspect of phonetics and phonology in speaking.

Causes of the EFL English Teachers' Difficulties in Connected Speech
Based on the data from the interview, it could be concluded that there were several causes which made teachers had difficulties in connected speech. First, they did have both phonology and phonetics background but never learnt about suprasegmental features like connected speech. In accordance with present result, the previous research have demonstrated that the lack of initial knowledge of both phonetics and phonology became the main reason of teachers' difficulties in connected speech feature (Couper, 2016). Since those aspects have played a major role to help the learners in pronouncing words both in boundary and in a casual speech. It is strongly recommended for teachers to improve their knowledge by joining several trainings, short courses, or even pursuing master degree.
Second, Having unsufficient words choice, feeling unconfidence, and speaking anxiety have hindered the teachers to provide a proper example of pronunciation for their students. This finding also accorded with earlier studies which showed that majority of English teachers were not having adequate confidence in pronouncing word or sentences (Uchida & Sugimoto, 2020 Moreover, their lack of confidence was not only a result of insufficient background knowledge of connected speech but also their status as non-native English teachers (Bai & Yuan, 2019). It is suggested that they needed to have more faith in theirselves related to their pronunciation ability so they could overcome that problem. Third, there was only a small gap between of language proficiency level betwwen the tecahers and the students. This finding broadly supported the work of previous research that some non-native English teachers had low language proficiency since they also had less opportunity and supportive environment to demonstrate their ability (Elder & Kim, 2013). Futhermore, the dependency to traditional strategies to teach pronunciation such as reading aloud, repetition, using dictionaries became main hinderance as these strategies did not provide proper understanding regarding pronunciation aspects.
Finally, the over used of the mother tongue by the teachers in teaching and learning process was significantly contributed to the difficulties in comprehending connected speech feature in a casual conversation through classroom interaction. This finding is contrary to the result of previous research that utilizing the use of native language in teaching was recommended to maximize target language learning (Shin et al., 2020). On the other hand, this current result matched with the work of earlier study that the over use of learners' native language in teaching could obstruct the learners' exposure and acquisition of the target language (Harmer, 2001). The different result of the previous studies showed the uncertainty whether the use of mother tongue in English teaching was beneficial or not. However, the present finding was more relevant with the idea that using learners' native language could hinder their progress in learning the target language.

CONCLUSION
Firstly, the assimilation feature was found to be the most difficult feature for EFL teachers in connected speech. There were 5 sub-features of assimilation namely assimilation /t/ to /p/, /t/ to /n/, /t/ to /s/, /d/ to /n/, and /n/ to /m/. Based on the analysis of the recording from read aloud activity, it was found that only 14% of teachers were able to use the assimilation /t/ becomes /n/. Meanwhile, 36% of teachers succeeded to perfom assimilation /d/ becomes /n/. However, none of teachers succesfully pronounced the rest assimilation feature. Overall, only 10% of teachers could apply the assimilation feature. Secondly, the linking feature was found to be the second most difficult feature for EFL teachers in connected speech. There were 3 sub-features of linking namely linking /r/, /j/, and /w/. It was discovered that 90% of teachers managed to produce linking /r/. At the same time, 30% of teachers succeded to use linking /w/. Nevertheless, all teachers were unable to use linking /r/ feature. In general, only 24% of teachers could use the linking feature. Thirdly, the elision feature was found to be the least difficult for EFL teachers in connected speech. There were 2 sub-features of elision namely elision /t/ and /d/. It was revealed that all participants could apply the elision /t/ and 86% of teachers managed to use the elision /d/. Altogether, 37% of teachers succesfully produced the elision feature. Lastly, to answer the second research question, the semi-structured interview was conducted. It was figured out that there were several reasons that contributed to teachers' difficulties in connected speech. They were lacking of both phonetics and phonological background, feeling unconfident, anxiety in speaking, unsufficient language proficiency, and the overusing of native language. Future researchers are invited to conduct research on both teachers and students' difficulties in pronunciation not only in the area of connected speech but also in the unobserved areas within these topics. It is also recommended to investiagte the proposed solutions, techniques, or even the use of advance technology to solve the mentioned problems and their causes.