<%BANNER%>

Speech Perception Test for Jordanian Arabic-Speaking Children


PAGE 1

SPEECH PERCEPTION TEST FOR JORDANIAN ARABIC SPEAKING CHILDREN By NADIA MOHAMED ABBAS ABDULHAQ A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by NADIA MOHAMED ABBAS ABDULHAQ

PAGE 3

This dissertation is dedicated to my parents Abbas and Ursula.

PAGE 4

ACKNOWLEDGMENTS I would like to thank my parents for their constant support and believe in me. They have been a bottomless source of support, inspiration and encouragement. My sisters Ute Nabila, Najat, Ulla Najwa, and brothers Mejdi and Uwe Maysarah, who where ever in the world they were made sure to keep in touch and encourage me through the tough times and share the joy of my success. My friend from Palestine, Lubna Shaheen, for all the times she has listened to my complaints and doubts and never let me go without giving me my confidence back. My good friend, Abdelqader Abu Awad, for his support and encouragement from the time I applied to the Fulbright program to the day of my graduation he has always been there for me. There is a group of people here in Gainesville without whom I would have felt very lonely. I would like to thank all, Debra Anderson, International student coordinator, for being a shoulder to lean on, a good listener and a friend in times of need. She has been my inspiration in many activities on campus that have brought diversity to my experience and made it so much richer in people, memories, and achievements. My friends in Gainesville, Abdullhatif Qamihieh for his readiness to help at any time what ever the task was. Maisa Haj-Tas for being a wonderful understanding roommate and her stimulating discussions and insight into research. Marah Al-Aloul, Ashraf Al-Qdah, Suha Abu Lawi, Shadi Krecht, Yaser Katanani, Enas Katanani and Hadil Faqih, for their support and for being my family away from home through out these past years, and my friends from the iv

PAGE 5

Fulbright program and international students for being a renewing source of hope, inspiration and fun. This project was made possible by the continuous support by the faculty members of the Departments of Communication Sciences and Disorders and Communicative Disorders at the University of Florida. I would like to thank in particular my first mentor Dr. Patricia Kricos for her guidance, support and generosity. Dr. Griffiths for his advise through the graduate program in general and mentoring my dissertation in particular, Dr. James Hall III, for his guidance, support and encouragement, Dr. Richard Harris for his hospitality, generosity and guidance through the research process, and Dr. Tom Oakland, for his honest opinions, believes in me and support. I would also like to thank Dr. Linda Lombardino, Dr. Alice Dyson, Dr. Keneth Gerhardt, Dr. Mary Kay Dykes, and Dr. Aida Bamia for their support. Funding for this project was made possible by the Gibson Dissertation Fellowship, offered by the Collage of Liberal Arts and Sciences at the University of Florida. The data collection and field research was made possible by the generous help and collaboration of Samar Al-Aghbar, the Jordanian ministry of education, and elementary schools of Amman, as well as the support of the Middle East Hearing Association and their team members, and Phonak hearing aid distribution office in Amman. Completion of this degree was made possible by financial support from the department of Communication Sciences and Disorders in the form of teaching assistantship and Grinter Fellowship award for four years, as well as the College of Liberal Arts and Sciences O. Ruth McQuown Scholarship for women, in addition to the v

PAGE 6

Fulbright scholarship that was the seed funding for this degree, and finally and most importantly, my parents financial support. vi

PAGE 7

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT.......................................................................................................................xi CHAPTER 1 INTRODUCTION........................................................................................................1 Background of the Study..............................................................................................1 Rationale and Purpose..................................................................................................3 Hearing Loss in Developing Countries.................................................................3 Hearing Loss in Jordan..........................................................................................4 Present Condition of Arabic Speech Audiometry.................................................5 Research Questions.......................................................................................................7 Experiment One.....................................................................................................7 Experiment Two....................................................................................................7 Hypothesis....................................................................................................................7 Experiment One.....................................................................................................7 Experiment Two....................................................................................................7 2 REVIEW OF THE LITERATURE..............................................................................8 Importance of Speech Audiometry...............................................................................8 Speech Audiometric Tools.........................................................................................11 Pediatric Speech Perception Materials................................................................13 Adult Speech Perception Materials.....................................................................14 Speech Reception Threshold Material.................................................................15 Full List and Half List Use..................................................................................16 Arabic Speech Audiometry.........................................................................................18 Comparison of Arabic Speech Reception Tests..................................................18 Use of Arabic Speech Reception Tests...............................................................20 Dialectal Differences in Arabic..................................................................................21 Special Considerations for the Present Study.............................................................23 Participants..........................................................................................................23 vii

PAGE 8

Speakers...............................................................................................................24 Stimuli for Dialectal Differences.........................................................................25 3 METHODOLOGY.....................................................................................................29 Participants.................................................................................................................29 Speakers......................................................................................................................30 Recording and Editing................................................................................................31 Instrumentation for Data Collection...........................................................................33 Study One...................................................................................................................34 Speech Material...................................................................................................34 Procedure for Data Collection.............................................................................35 Statistical Analysis..............................................................................................36 Speech Material...................................................................................................37 Statistical Analysis..............................................................................................37 4 RESULTS...................................................................................................................41 Study One...................................................................................................................41 Study Two...................................................................................................................43 5 DISCUSSION.............................................................................................................57 APPENDIX A PARTICIPANT RECRUITMENT.............................................................................64 B SPEECH MATERIAL SELECTION.........................................................................65 C LOGISTIC REGRESSION ANALYSIS....................................................................67 D LIST OF WORDS FOR DIALECT COMPARISON................................................68 E HALF WORD LISTS AND MEANING...................................................................69 LIST OF REFERENCES...................................................................................................71 BIOGRAPHICAL SKETCH.............................................................................................77 viii

PAGE 9

LIST OF TABLES Table page 2-1 Childrens speech reception tests in English............................................................27 2-2 Speech recognition word lists mean dB HL levels at 50% and slopes...................27 2-3 Speech reception threshold wordlists mean threshold dB HL levels and slopes between 20 and 80%................................................................................................28 3-1 Order of list presentation by participant and level of presentation..........................39 3-2 Word list and dialect presentation order..................................................................40 4-1 Normal Hearing participants age, gender, test ear and PTA...................................50 4-2 Half word lists and transcription in IPA...................................................................51 4-3 Logistic regression results for the full lists, the calculated slopes at the 50% and the 20-80% levels, the threshold dB level, and difference of threshold levels from the mean threshold...........................................................................................53 4-4 Logistic regression results for the half lists, the calculated slopes at the 50% and the 20-80% levels, the threshold dB level, and difference of threshold levels from the mean threshold...........................................................................................53 4-5 Selected characteristics of normal hearing participants...........................................54 4-6 Selected characteristics of hard of hearing participants...........................................54 4-7 Word recognition scores in percent correct for each subject per talker-talker and the average of scores by dialect................................................................................55 4-8 Mixed ANOVA results comparing dialects for all participants' scores...................55 4-9 Paired Sample T-tests comparing dialectal difference in the hard of hearing group and normal hearing group..............................................................................55 4-10 Paired Sample T-test results including all participants scores................................56 5-1 Descriptive statistics for Hard of Hearing and Normal Hearing based on dialect...63 ix

PAGE 10

LIST OF FIGURES Figure page 4-1 Average pure tone thresholds at all frequencies for all 20 normal hearing participants at 250, 500, 1000, 2000, 4000, and 8000 Hz........................................46 4-2 Half lists raw data scatter plot..................................................................................46 4-3 Psychometric function of 4 lists of 50 words based on calculated percent correct..46 4-4 Psychometric function of 8 half lists of 25 words based on calculated percent correct.......................................................................................................................47 4-5 Arabic monosyllabic psychometric functions for lists 1-4 (left) and half-lists 1A-4B (right)..................................................................................................................48 4-6 Average of pure tone thresholds of normal hearing participants at 250, 500, 1000, 2000, 4000, and 8000 Hz...............................................................................49 4-7 Average of pure tone thresholds of hard of hearing participants at 250, 500, 1000, 2000, 4000, and 8000 Hz...............................................................................49 x

PAGE 11

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SPEECH PERCEPTION TEST FOR JORDANIAN ARABIC SPEAKING CHILDREN By Nadia Mohamed Abbas Abdulhaq May 2006 Chair: Scott Griffiths Major Department: Communication Sciences and Disorders Comprehensive audiologic evaluation includes a variety of tests that provide a determination of the type of hearing loss. Among these tests are tests of word recognition. Many speech perception tests have been developed over the past half century to assess different aspects of speech. Management of hearing loss in developing countries and the USA differs significantly. The prevalence and demographics of the hard of hearing population is also different, and thus imposes different needs. According to the World Health Organization the number of deaf and hard of hearing individuals in developing countries is twice as much as in developed countries. There is a need for early identification and intervention for hearing loss in developing countries. At the same time there is a severe lack of equipment and highly trained professionals to provide such services. The focus of this study is speech audiometric measure for Jordanian Arabic speaking children. The first goal of this study is to develop four Jordanian Arabic 50-word lists appropriate to use for word recognition measure for Jordanian children age 6 to xi

PAGE 12

9 years. The second goal is to investigate the effect of using words recorded in Saudi dialect on the word recognition abilities of Jordanian normal hearing and hard of hearing children. Twenty individuals age 6 through 9 years participated in this study. A Jordanian speaker recorded 250 Arabic words selected for familiarity to elementary aged Jordanian children. The raw score data of all participants at the 10 intensity levels were compiled for each of the 250 words. Four equally difficult lists of fifty words each were derived from this experiment. In the second experiment, 3 Jordanian and 3 Saudi male speakers were selected to record 33 words. These were played back at a constant comfortable level for 10 children with normal hearing and 10 children with hearing impairment who were asked to identify the recorded words. Dialect produced a significant difference in performance for children with normal hearing, but not for children with hearing impairment. These results will lead to the development of improved techniques for assessing auditory performance in Arabic-speaking children. xii

PAGE 13

CHAPTER 1 INTRODUCTION Background of the Study An audiological evaluation typically includes measures of tympanometry, acoustic reflexes, otoacoustic emission, and pure-tone (air and bone conduction) threshold and speech audiometry. Results of these tests are used to diagnose hearing loss and determine the type of hearing loss. Hearing testing has come a long way from estimating hearing from the distance a person can hear a voice. Since the early 1900s, pure-tone audiometry has been instrumental in measuring hearing sensitivity. The work of Harvey Fletcher on the perception of speech in relation to sound level and noise effect laid the foundation for decades of speech perception research (Fletcher, 1995). Speech perception measures in the present format in the United States of America (USA) have been used since the late 1940s and early 1950s (Egan, 1948; Haskins, 1949; Hirsh, Davis, Silverman, Reynolds, Eldert, & Benson, 1952). Many speech perception tests have been developed and improved over the past half century to assess different aspects of speech; e.g., speech reception thresholds, speech pattern identification, and speech reception in noise (Elliot & Katz, 1980; Erber, 1974; Goldman, Fristoe, & Woodcock, 1970; Jerger, Lewis, Hawkins, & Jerger, 1980; Ross & Lerman, 1970; Tillman & Carhart, 1966). Speech reception tests that mostly are used clinically include the Central Institute for the Deaf W-22 (CID W-22) by Hirsh et al. (1952), Northwestern University Auditory Test No. 6 (NU-6) by Tillman & Carhart (1966), Phonetically Balanced Kindergarten 1

PAGE 14

2 Test (PBK-50) by Haskins (1949), and Northwestern University Childrens Perception of Speech (NU-CHIPS) by Elliot & Katz (1980). Common characteristics among these tests are that they are based on monosyllabic frequently used words which are familiar to the target population (at least at the time of test development) and were developed based on measures of correct responses as a function of intensity of presentation. The articulation function is a common representation of speech audibility result, which forms an ogive or S-shaped curve and indicates the degree a persons hearing ability improves given increases in the intensity of the material presented (Carhart, 1951). The shape of the ogive and the 50% point (threshold point) are affected by the speech material presented as well as the speaker. That is, different speech materials produce different ogives. If different speakers recorded the same speech material, the result would yield different ogives (Beattie, Edgerton, & Svihovec, 1977; Beattie, Svihovec, &Edgerton, 1975; Carhart, 1965; Doyne & Steer, 1951; Hirsh, Reynolds, & Joseph, 1954; Kruel, Bell, & Nixon, 1969; Wilson & Carter, 2001; Wilson & Oyler, 1997). Results from adult speech recognition test (NU-6) and childrens tests (PBK-50 and WIPI) were compared (Sanderson-Leepa & Rintelman, 1976). The adult test was found to be inappropriate for use with young children age 3.5 to 9.5 years while WIPI was most appropriate for young children ages 3.5 to 5.5, and NU-6 was appropriate but more difficult than the PBK-50 for children age 7.5 to 11.5 the. That is, speech material needs to be age appropriate, and words need to be familiar to the target group. Carhart (1965), and Kreul et al. (1969) recommend the use of a limited set of recorded materials to establish normative measures because different recordings by the same speaker can result in different articulation functions.

PAGE 15

3 Studies of speech perception in other languages (e.g., Arabic, Polish, Korean, and Chinese) have followed Egens (1948) and Hirsh et al. (1952) lead in developing speech reception material (Alusi, Hinchcliffe, Ingham, Knight, & North, 1974; Zakrzewski, Jassem, Pruszewicz, & Obrebowski, 1976; Ashoor & Prochazka, 1982; Ashoor & Prochazka 1985; Harris, Kim, & Eggett, 2003a; Harris, Kim, & Eggett, 2003b; Nissen, Harris, Jennings, Eggett, & Buck, 2005a; Nissen, Harris, Jennings, Eggett, & Buck, 2005b). Special attention was paid to the familiarity of words, the equivalence of word lists, and in some cases the phonetic balance of word lists. The resulting lists have had similar articulation functions to the English word lists in shape and slope. Thresholds of speech reception and word recognition were more varied across languages. Rationale and Purpose Hearing Loss in Developing Countries The management of hearing loss in developing countries and the USA differs significantly. The prevalence and demographics of the hard of hearing population also differs and thus imposes different needs for services. According to the World Health Organization (WHO), the number of deaf and hard of hearing individuals in developing countries is twice that in developed countries (Smith, 2001). According to the American Academy of Audiology an average of 3 in every 1,000 healthy newborns in the United States has severe sensorineural hearing loss. In Jordan 6 in every 1,000 healthy newborns have hearing loss (Al-Masri, 2003). For comparison purposes, the WHO reported that 4 to 5 children under the age of 18 in every 1,000 children have sensorineural hearing loss in the South-East-Asian Region (Smith, 2001). This number is reported to be inaccurate due to the lack of epidemiology surveys in developing countries; the actual numbers are projected to be double what has been reported (Smith, 2001). Globally, the majority of

PAGE 16

4 children are living in developing counties, which indicates that the majority of children with hearing loss live in those countries as well (Jauhiainen, 2001). The WHO identified the major causes of hearing loss to be chronic otitis media, genetics, maternal and perinatal problems, and ototoxicity (Smith, 2001). The rates of complications due to otitis media are 1/1000 in developed countries and 60/1000 in developing countries. The incidence of deaths due to ear infections complications are 1/100,000 in developed countries and 1 /100 in developing countries (WHO, 1998). In spite of the higher prevalence of hearing loss in developing countries, services and technology generally are limited or lacking, especially in rural areas (WHO, 1998; Jauhiainen, 2001). The prevalence of Ear, Nose and Throat doctors ranges from 1/ 30,000 to 1/150,000 in developed countries and 1/2,000,000 in the less developed countries in Africa (excluding South Africa and Egypt) (WHO, 1998). Early identification and intervention for hearing loss in developing countries are need. At the same time there is a severe lack of equipment and highly trained professionals to provide such services. In spite of this general statement, services in some developing counties are more advanced than others, and some individuals have the means to afford world class services (Jauhiainen, 2001). The focus of this study is on audiology in Jordan, specifically speech audiometric measure for Jordanian Arabic speaking children. Hearing Loss in Jordan Pilot data from screenings conducted by the Middle East Hearing Association (MEHA) suggest estimates of the hard of hearing and deaf population in Jordan to be 64,000, with 2,200 infants with severe hearing loss born yearly (Al-Masri, 2003). Some institutions provide limited services, including the Speech and Hearing Clinic at the

PAGE 17

5 University of Jordan, Al-Ahliyya Amman University, Holyland Institute for the Deaf, King Hussein Medical Center, and some private otolaryngology and audiology clinics and hearing aid providers. However, diagnostic and rehabilitative services and professional training are seriously lacking. There are initiatives by the Jordanian government, the royal family in particular, to develop such services. Under the patronage of His Royal Highness, Prince Firas Raad, MEHA was established in 1998 in cooperation with the Canadian International Scientific Exchange Program, and a new center was opened recently to provide services to the hearing impaired and deaf population. Projects being implemented through this organization include newborn hearing screening, genetic hearing loss research, audiologic evaluation, hearing aid fitting, audiologic rehabilitation and follow up for children. Equipment needed for such services is available and some basic audiometric measures are being performed. An important part of evaluation, speech perception audiometry, is missing from the diagnostic battery. Speech perception materials suitable for testing Jordanian Arabic speaking children are unavailable. The purpose of this study is to develop a speech perception test for Jordanian Arabic speaking children. Present Condition of Arabic Speech Audiometry Carhart (1951) emphasized the importance of using familiar words that are in the listeners native language. Tests developed in Arabic have been recorded using Maroccan (Messouak, 1956), Iraqi (Alusi, 1974), Egyptian (Soliman, 1976; Soliman, Abd El-Hady, Saad, & Kolkaila, 1987; and Soliman, Fathallah, & El-Mahalawi, 1987), and Saudi (Ashoor et al., 1982; and Ashoor et al. 1985) dialects. The question arises as to whether the different dialects have an effect on speech reception. Although all Arabic countries share one standard Arabic language that is taught formally in school in the form of the

PAGE 18

6 written language (Altoma, 1969; Al-Kahtani, 1997; Fatihi, 2001; Ferguson, 1956) dialectal differences exist. Studies conducted to measure the dialectal effect on speech reception in Arabic could not be located. Thus, one may question the effect of the dialect related to the words used in that dialect as well as the articulation and voice characteristics of the speaker. Recordings by different speakers result in different word recognition scores, and the use of different speech material has the same effect (Beattie et al., 1977; Beattie et al., 1975; Carhart, 1965; Doyne & Steer, 1951; Hirsh et al., 1954; Kruel et al., 1969; Wilson & Carter, 2001; Wilson & Oyler, 1997). This suggests a need to develop speech audiometric material based on one recording to ensure reliable results. Recordings reported in other studies (e.g. Alusi et al., 1974; Ashoor & Prochazka, 1985) have not been available for wide use. One intended outcome of this study is to make available a high quality digital recording of speech material appropriate for use in testing word recognition loss of Jordanian children. Audiologic testing in Jordan currently is based on non-speech related audiometry, including pure-tone audiometry, otoacoustic emissions, and auditory evoked potentials. This study has two main goals. One is to develop four Jordanian Arabic 50-word lists appropriate to use in word recognition measures for Jordanian children ages 6 through 9. Another goal is to investigate the effect of using words recorded in Saudi dialect on word recognition abilities of Jordanian children who display normal hearing and hearing loss.

PAGE 19

7 Research Questions Experiment One 1. Is there a significant difference in word recognition abilities of Jordanian Arabic speaking children, given increases in intensity of presentation? 2. Can four parallel word lists be developed in Jordanian Arabic language (e.g., their psychometric qualities of the word lists do not differ significantly)? Experiment Two 1. Do word recognition abilities of Jordanian Arabic speaking children differ when listening to words presented in Jordanian versus Saudi dialects? 2. Do word recognition abilities of Jordanian Arabic speaking children who display normal and hearing disabilities differ when hearing words presented in a Jordanian versus Saudi dialect? 3. Does word recognition ability differ when hearing speakers using the same dialect? Hypothesis Experiment One 1. An increase in presentation level will increase word recognition scores. 2. Four lists of 50-words with similar characteristics can be created to be used in measures of word recognition abilities. Experiment Two 1. Jordanian children will produce higher word recognition scores when listening to words presented in Jordanian dialect than when listening to words presented in Saudi dialect. 2. Jordanian and Saudi dialects will have the same effect on word recognition scores of Jordanian children with normal hearing and with hearing loss. 3. Word recognition will differ for two talkers with the same Arabic dialect.

PAGE 20

8 CHAPTER 2 REVIEW OF THE LITERATURE Importance of Speech Audiometry Non-speech audiometric procedures provide valuable information needed for description of hearing loss and estimation of amplification benefit. These tests, however, cannot measure the effect of hearing loss on speech (Carhart, 1951; Cramer & Erber 1974; Davis, 1948; Doyne & Steer, 1951; Erber, 1974; Hirsh, et al. 1952). Speech audiometry requires a language related test material, and thus may be influenced by the phonetic, melodic, and intonational differences between languages (Carhart, 1951). Different languages require speech tests that consider the features unique to the language. Distinctive features of languages can result in different auditory requirements and affect the auditory capacities evaluated. To provide comprehensive audiologic services, it is necessary to use speech audiometry for measurement of hearing and the outcome of management. Speech perception is important in facilitating cognitive development and normal language acquisition (Cramer & Erber 1974). Hearing loss is an obstacle of language and speech acquisition. Pure-tone thresholds provide information about detection of sound at specific frequencies, yet provide little information about the perception of complex signals such as speech. Knowledge of a persons ability to perceive speech can provide information as to the extent hard of hearing children are able to communicate effectively and, in addition, how they are likely to learn language. Speech audiometry is a clinical

PAGE 21

9 approach in which well-defined speech samples are presented using a calibrated system to measure an important aspect of hearing ability. Measurement of speech perception different intensities results in an articulation function taking the shape of an ogive or S-shaped curve, that indicates the degree a persons hearing ability improves with an increase of the intensity of the material presented (Carhart, 1951). Over 50 years ago, Carhart (1951) reported on the importance of speech audiometry in hearing assessment. He considered it to be the most useful contribution to hearing testing after the introduction of pure-tone audiometry. Carhart emphasized the value of well-defined speech audiometry to provide finer classification of hearing loss, and its importance in providing measures for educational and rehabilitation purposes. Another early hearing researcher, Hallowell Davis (1948), acknowledged that there is no simple interpretation of the pure-tone audiogram to express the patients ability to hear speech. It is important to measure the patients ability to discriminate speech just as it is important to measure the hearing loss in decibels. Fletcher noted, The process of speaking and hearing are very intimately related, so much so that I have often said that, we speak with our ears. We can listen without speaking but we can not speak without listening (p.A1 Fletcher, 1995). Speech perception and language acquisition are two closely related processes. Hearing loss usually disrupts the process of speech perception and thus delays language acquisition. Therefore, Cramer & Erber (1974) emphasized the need for an accurate and valid measure of speech perception for hearing impaired children. Arlinger (2001) emphasized the importance of using speech recognition to measure hearing aid benefit in children, and considered periodic evaluation of hearing aid benefit of great importance for children

PAGE 22

10 to monitor their language and speech development. Blamey, Sarant, Paatsch, Barry, Bow, Wales, Wright, Psarros, Rattigan, & Tooher (2001) used a measure of speech perception to evaluate benefit of amplification, and the relationship between speech perception and speech production, language, hearing loss and age in 87 children age 4 to 12 years. According to this study, the authors expected the language delay to be 4 to 5 years by the time children are 12 years old. Speech perception scores are expected to improve significantly with the improvement of language; specifically they expect children to score 90% on the open set Bench-Kowal-Bamford (BKB) sentence test when they reach the level of language proficiency of a 7 year old. ster (2002) examined the relation between audiological measures and speech intelligibility for eleven profoundly deaf Swedish teenagers (age 15 to 17 years). Their pure-tone averages (PTA) ranged from 90 to 108 dB HL. Correlation analysis was assessed for the intelligibility of the childrens speech and their pure-tone average, the shape of the audiogram, and residual hearing use. Results showed that there was great variation in their speech intelligibility in spite of the narrow range of PTA, indicating that the speech intelligibility cannot be estimated based on PTA. The correlation between speech recognition scores and speech intelligibility scores resulted in a positive correlation of 0.73, confirming a high correlation between residual hearing use and speech intelligibility. ster (2002) concluded that a simple speech test can be used as a predictor of prelingually deaf childrens ability of developing intelligible speech. Laukil & Fjermedal (1990) researched the reproducibility of bone conduction thresholds and the speech recognition thresholds. The results show low variability and no significant difference between the two measures making the speech recognition threshold

PAGE 23

11 measure a reliable one for Norwegian Spondees. These studies indicate that speech perception measures are very important and reliable clinical tools. Speech Audiometric Tools Speech audiometry is an important part of an audiologic evaluation battery (Arlinger 2001; Carhart 1951; Cramer & Erber, 1974; Curry 1949; Davis 1948; Erber, 1974). There are two essential measures of speech audiometry: the measure of speech recognition threshold (SRT) and the measure of speech recognition at suprathreshold levels (Carhart 1951; Davis, 1948). Davis (1948) has identified these measures as two dimensions of hearing loss; one is hearing loss (dB level) and the second is discrimination loss (word recognition score). Discrimination loss relates to the loss of the ability to recognize words even when they are made audible. Several tests have been developed to measure these two dimensions of hearing loss. The most frequently used ones for speech perception are speech recognition threshold (SRT) test and word recognition scores (WRS). An example of the SRT test is the Central Institute for the Deaf spondee word test (CID W-2) by Hirsh et al. (1952). Many word recognition materials lists have been developed over the years including the Phonetically Balanced 50 word lists (PB-50) described by Egan (1948), the CID W-22 (Hirsh et al., 1952), the Northwestern University Test No.6 (NU-6) by Wilson and Oyler (1997), the PB Kindergarten word test (PBK) by Haskins (1949), the Word Intelligibility for Picture Identification Test (WIPI) by Ross & Lerman (1970), and the Northwestern University Childrens Perception of Speech (NU-CHIPS) by Elliot & Katz (1980). Egan (1948) specified the criteria of selecting word lists for the word recognition tests as follows: 1. Monosyllabic structured words 2. Equal average of difficulty between lists 3. Equal range of difficulty within lists

PAGE 24

12 4. Equal phonetic composition between lists 5. A composition representative of spoken English 6. Commonly used words The above listed speech perception tests were developed following Egans guidelines. Speech audiometric tests have been developed in many different languages, each to fit the requirements of measuring speech perception based on the specific features of that language. Measurement of speech perception tests included the articulation function that defines word recognition scores at different intensity levels. The measurement starts at a very low intensity level where the material is unintelligible. As the signal intensity level is increased, the listener is able to identify correctly more of the stimuli up to a point at which the intensity is high enough for the listener listening to his native language to identify all the material without error. Fletcher (1929) demonstrated that the shape of the curve changes from one material to another and, using the same material, measurement of speech recognition with different speakers results in different shaped curves. With speech discrimination loss, the articulation curve not only shifts at the dB axis but it also changes in shape. The curve reaches a plateau at percentage correct levels below 100% (Davis, 1948). Davis concludes that this effect results from the loss of sensitivity, especially in the high frequencies, which is important for consonant recognition and clarity of speech. The shape of the curve is again different for each hearing impaired individual. It might be of normal shape and shifted to the right indicating higher intensity levels, or it might have a different shape and not reach the 100% correct identification.

PAGE 25

13 Pediatric Speech Perception Materials Childrens speech perception tests have been developed with different materials and different tasks appropriate for different age and ability levels. Some tests are open set (i.e. the listener has no knowledge of the category of word or any contextual cues) and require verbal response (e.g. Phonetically Balanced Kindergarten 50 Word Test, Haskins 1949), whereas others are closed set (i.e. the listener is provided with a set of 3-6 options to chose from or the category of words is specified to provide some cues) requiring picture pointing (e.g., Word Intelligibility by Picture Identification, Ross & Lerman, 1970). Following is a detailed description of the most commonly used children speech perception tests (see table 2-1 for list). Haskins (1949) developed The Phonetically Balanced Kindergarten 50 Word Test. Though the Haskins lists appeared on in the authors masters thesis at Northwestern University, they have been widely used.The four lists were developed based on the phonetically balanced word lists (PB-50) used by Egan (1948) for assessment of speech perception in adults Haskins selected words that were among the 2500 words most used by kindergarten children (The International Kindergarten Union, 1928). Measurement of the psychometric function and the equivalence of wordlists were completed with adult normal hearing participants with one randomization of lists presented at 5 intensity levels. As a result, lists 1, 3, and 4 were found to be equivalent while list 2 was easier. The slope of the psychometric function between 20 and 80% word correct was 4%/dB, as reported in Mayer & Pisoni (1999). In spite of the wide use of these lists, no formal data collection and analysis was completed with pediatric populations. Other speech perception audiometric tests for children were mostly developed for closed sets of words (Elliot & Katz, 1980; Erber, 1974; Erber, 1980; Goldman et al.,

PAGE 26

14 1970; Jerger et al., 1980; Ross & Lerman, 1970), and sometimes with groups of hard of hearing children as the only participants (Erber & Alencewicz, 1976; Ross & Lerman, 1970). Several closed set picture presentation tests are used regularly in audiology clinics. These tests include the Word Intelligibility by Picture Identification (WIPI) by Ross and Lerman (1970), the Northwestern University Childrens Perception of Speech (NU-CHIPS) by Elliott and Katz (1980), Goldman-Fristoe-Woodcock Test of Auditory Discrimination (Goldman et al., 1970), and The Pediatric Speech Intelligibility test (PSI) by Jerger et al. (1980). Erber & Alencewicz (1976) suggested a picture pointing closed test to evaluate the word recognition ability of children with hearing loss that provides a distinction of word recognition and word pattern recognition. Adult Speech Perception Materials Beattie & Warren (1983) described for adult word recognition tests in English, an increase in intelligibility with the increase of intensity equivalent to 4.5%/dB in the range of 20 to 80% scores, with an approximation of maximum intelligibility at level s of 25 dB SL. Wilson, Zizz, Shanks, & Causey (1990) reported the NU-6 word recognition threshold when spoken by a female speaker to be 4.5%, similar to other studies, while the intensity level of 50% correct recognition was shifted 5dB to the right (higher than previous studies). Wilson & Oyler (1997) compared the psychometric function of the CID W-22 word lists and the Northwestern University No.6 (NU-6) as spoken by the same talker and found the 50% score level to be at 15.6dB HL for the W-22 and 13.4 dB HL for the NU-6. The slopes between 20% and 80% points were 4.8%/dB for the W-22 and 4.4%/dB for the NU-6. These results are comparable to speech audiometry data for other languages For example, Harris et al. (2003a) studied the psychometric function of wordlists spoken by males and females and found the mean 50% level in Korean to be at

PAGE 27

15 11.4 dB HL for male speakers and 10.7dB HL for female speakers with mean slopes between 20% and 80% points of 4.4%/dB for male and female speakers. Niessen et al. (2005a) found a threshold level for Chinese Mandarin speech materials of 5.4 dB HL for male speakers and 2.3dB HL for female speakers, and mean slopes between 20% and 80% points were 6.3%/dB for male speakers and 7.1% for female speakers. The difference in the Chinese word lists might have been because disyllabic words were used in the composition of the lists. Alusi et al. (1974) developed equivalent lists for speech recognition in Arabic with a threshold level of 22.5 dB HL and a slope of 5%/dB. In summary, studies of monosyllabic lists in three languages (English, Korean, and Arabic) show similarity in psychometric function slopes (ranging from 4.4% to 5.1%) whereas the study of disyllabic Chinese word lists showed steeper slopes (6.3% and 7.1%). Table 2-2 lists the thresholds and slopes of word lists in the different languages listed above. Review of data displayed in Table 2-2 shows that there is a difference in threshold among languages, with the Chinese disyllabic words having the lowest threshold (2.3 dB HL) and the highest for Arabic monosyllabic word lists (22.5 dB HL). Speech Reception Threshold Material Studies of speech recognition thresholds (SRT) lists in different languages revealed steeper slopes than those reported for word recognition lists. SRT measures in English are composed of disyllabic spondees. Hirsh et al. (1952) reported a psychometric function slope between 20% and 80% of 8%/dB. Young, Dudley, & Gunter (1982) reported a slope of 10%/dB, and Wilson & Strouse (1999) reported a slope of 7.4%/dB. In studies of languages other than English, such as the study of trisyllabic Chinese Mandarin materials, Nissen et al. (2005b) found a slope between 20 and 80% of 9.7%/dB for a male speaker and 10.5%/dB for a female speaker. According to Nissen (2005b), slopes of SRT tests in

PAGE 28

16 other languages are at similar levels, i.e., Polish = 10.1%, Spanish = 11.1%, and Italian = 7.3%. Harris et al. (2003b) in their study of Korean disyllabic words found a slope of 10.3% for male speakers and 9% for female speakers. Ashoor & Prochazka (1982, 1985) reported a slope of SRT word lists of 5% for both adults and childrens wordlists. Slopes of Ashoors lists are less steep than those reported in other languages, but similar to slopes of word lists used in Alusis word recognition scores for adults. Siegenthaler, Pearson, & Lezak (1954) investigated the speech reception threshold for children using monosyllabic words and found the slope to be 8.6%/dB between 20% and 80% correct word recognition. Ashoor & Prochazka found the threshold of word recognition to be at 2.2dB HL for adults (1982) and at 0 dB HL for children (1985). Table 2-3 lists speech reception threshold wordlists in different languages. Full List and Half List Use The use of full lists or half lists depends on the patients performance on the test. Studies by Beattie and Warren (1983), Dubno Lee, Klein, Matthews and Lam (1995) and Thornton and Raffin (1978) investigating the confidence intervals of using full lists of 50 words and half lists of 25 words in initial testing and retesting of patients performance on speech reception. Thornton and Raffin (1978) described the variability in speech discrimination scores based on the CID W-22 test, and highlighted the differences in variability between using full lists and half lists. Their results show the fact that the closer the scores are to either end of the spectrum of scores (0 or 100) and the more words are included in the list the less variability there is in scores and the smaller the confidence interval. For example if a patient scored 96% on a 50 word list the confidence interval is between 86-100. In other words if the patient scored 96% the first time and was retested, a score between 86-100 will be considered not different from the

PAGE 29

17 first score. While if the patient scored 96% on a half list of 25 words, the confidence interval is between 80 and 100. With lower scores, closer to 50%, the confidence interval grows larger more so for half lists than for full lists. This can be used as an indication for the need to use a full list versus a half list. That is if the score is closer to 50%, the use of 50 word lists would provide a more accurate measure. In comparing the test retest results from hearing impaired participants, Beattie and Warren (1983) found the standard deviation in test retest results using 25 word lists to be 10%, which was reduced to 8% using 50 word lists and to 6% when using 100 word lists. In Beattie et al.s judgment, this difference was not significant to increase the size of test material and was satisfied with the 25-word list size. Dubno et al. (1995) studied the correlation between the degree of hearing loss and word recognition scores. They studied the word recognition scores from 407 ears with normal hearing and mild to severe hearing loss, with the goal of providing data for confidence limits of scores on 25 and 50 NU-6 word lists in relation to the PTA. The authors provided tables of scores for 25 and 50 word lists corresponding to the 95% confidence limit of best performance (PBmax). They found a correlation between word recognition score and PTA, where a lower PTA resulted in higher scores. These findings are intended to help in diagnosis decisions on whether the score is considered within expected range for the degree of loss or whether it is poorer than expected and thus requiring additional testing. Dubno et al. (1995) cautioned about the use of these tables to generalize to other lists since different material would have different results. Still this gives an indication for clinicians and researchers to be cautious when using word lists at one presentation level and to keep this data in mind when making clinical decisions.

PAGE 30

18 Arabic Speech Audiometry Several speech perception tests are available in the Arab countries. These include speech recognition tests for adults in Moroccan (Messouak, 1956), Iraqi (Alusi et al., 1974), Egyptian (Soliman, 1976) dialects, and an SRT test (Ashoor & Prochazka, 1982) in Saudi dialect. Pediatric speech audiometry tests have been developed including an SRT test in Saudi (Ashoor & Prochazka, 1985) and Egyptian (Soliman et al., 1987b) dialect, an Arabic word intelligibility (recognition) by picture identification in Egyptian dialect (Soliman et al., 1987a), and an Arabic speech pattern contrast (ArSPAC) test developed in Israel (Kishon-Rabin & Rosenhouse, 2000). Some of these publications are inaccessible (Messouak, 1956; Soliman, 1976; Soliman et al., 1987a; Soliman et al., 1987b). Insufficient details are found in the literature to allow for in depth discussions of these tests. Recordings of the test materials are not widely available, even though the word lists are printed in the publications (Alusi et al., 1974; Ashoor & Prochazka, 1985; Kishon-Rabin & Rosenhouse, 2000). Allusi et al.s (1974) and Ashoor & Prochazkas (1982 and 1985) studies address word recognition comparable to the present study; therefore these studies are discussed in detail. Comparison of Arabic Speech Reception Tests Alusi et al. (1974) and Ashoor & Prochazka (1982) used monosyllabic words in their word lists. The structure of monosyllabic words was CVC, CVCC, CVVC, and CVVCC, with all 28 consonants and 6 vowels of standard Arabic represented. Alusi et al. (1974) divided the 150 monosyllabic words into 6 phonetically balanced lists of 25 words. Ashoor & Prochazkas adult lists (1982) contained 120 words divided into 6 phonetically balanced lists of 20 words, and their childrens lists (1985) included 80 words divided into 8 lists of 10 words. Both Alusi and Ashoor based their phonetic

PAGE 31

19 balance of lists on the frequency of consonant and vowel occurrence compared to continuous text, based on counts conducted at the time of study. Ashoor & Prochazka (1985) in his childrens lists focused on equal distribution of syllable structure more than the phonetic representation, while he kept the overall phonetic balance in all 80 words. To ensure word familiarity, both researchers used standard Arabic, the main teaching language in schools and universities as well as the language of mass media (i.e., newspaper, radio and TV broadcast). The word sources of choice were elementary school books, childrens stories, and daily newspapers. Absurd words and technical vocabulary were excluded. In addition both researchers chose words that are similar in standard Arabic and colloquial. Ashoor & Prochazka (1982 and 1985) ensured the word familiarity by collecting ratings from a large number of participants coming from 14 different regions of Saudi Arabia. Lists were recorded at voice intensity of 70 to 75dB SPL, fluctuation was limited to + 5 dB on volume meter in sound treated booths, and ambient noise did not exceed 30 dBA SPL. Neither researcher used carrier phrases. Alusi et al.,s recording (1974) was in standard Baghdad dialect while Ashoor and Prochazkas recordings (1982 and 1985) were in standard Saudi dialect. The rates of recording varied. Alusi et al. (1974) recorded 8 words per minute, whereas Ashoor and Prochazka recorded 12 words per minute for adults and 6 words per minute for children (Ashoor & Prochazka, 1982, and 1985). The intelligibility tests were performed with somewhat different criterion. Alusi et al. (1974) included 17 participants age 20 to 38 years representing several Arab countries, since his goal was to develop a test that could be used in different Arab countries. Ashoor & Prochazka (1982) enrolled 74 adult participants age 20 to 35 years representing 14

PAGE 32

20 rural and urban areas in Saudi Arabia, and 100 children age 4 to 9 years representing most Saudi dialect areas. Both studies (Alusi et al., 1974; and Ashoor & Prochazka, 1982) presented the stimulus words via headphones to their adult participants, while Ashoor and Prochazka (1985) presented the stimulus words via sound field to their pediatric participants. In all studies the authors attempted to measure the difference in inter-list difficulty and also the difference associated with presentation method. The word lists were presented at different intensity levels ranging from speech detection thresholds, increasing by 5dB steps, up to the point where 100% of the words were identified correctly. There was no difference between lists in terms of difficulty or the order of stimuli presentation in ascending or descending dB level, within each study. Resulting articulation function curves from all three studies were similar to those of other languages, including English, in shape and slope. The findings suggest the lists are suitable as speech recognition measure. Ashoor & Prochazka (1982) found a slight difference between the adult groups of students and non-students, i.e., a curve shift further to the right for the latter group indicating higher threshold levels (in dB). In addition, he found a difference between two age groups (4 to 5 years and 5 to 9 years). The younger group needed higher intensity levels than the older group to reach threshold level of 50% word recognition, a finding that may be related to maturation and knowledge of the language. Use of Arabic Speech Reception Tests Though Alusi et al. (1974) and Ashoor & Prochazka (1982 and 1985) recorded their material, these recordings were not marketed for wide distribution. The limited distribution may have several explanations. One is the small number of audiologists and audiological services in Arab countries. For example, Saudi Arabia has one of the most

PAGE 33

21 advanced services for communication disorders among the Arab countries. With a population of 25 million, there are 14 registered audiologists at the Saudi Speech Pathology and Audiology Association and five facilities that provide audiological services (SSPAA, 2004). The second possible explanation for the limited distribution of speech materials is the difference in dialects between Arab countries. Although Arab countries share the standard written Arabic language, there is a wide range of dialects (Fatihi, 2001). Published speech recognition tests are in Moroccan, Baghdadi, Egyptian or Saudi standard dialect. The possibility of using one test across the Arab countries has not been investigated. Alusi et al. (1974) has suggested the possibility of using the word lists he developed in all Arab dialects since the words were taken from standard Arabic. However, Alusis speech materials were recorded in a Baghdad standard dialect. In developing the speech test, Alusi had a limited number of participants (17) representing several Arab countries (the author did not specify which countries), who were young educated adults. The sample did not necessarily represent the large Arabic speaking population). However, Alusi et al. did attempt to meet the criterion of word familiarity by choosing words from childrens books and newspapers in order to include educated and un-educated populations. He did not describe a specific comparison between participants from different countries to support his argument. Dialectal Differences in Arabic One of the goals in the present study is to determine whether there is a difference in word recognition scores for normal hearing and hearing impaired children listening to Jordanian and Saudi dialects. The issue of Arabic language diversity and its dialects is important in the selection of speech material and speaker. Although a single standard Arabic language is used in all Arab countries, dialectal differences do appear. The

PAGE 34

22 difference between standard and dialect with the Arabic language, similar to other languages, such as Greek, Swiss German, and Haitian Creole, is referred to as diglossia. Furgeson (1959) defined diglossia as the presence of a stable situation of a language in which there is a dialect of a language primarily used in daily communication and a very different superimposed variety of the language that is part of a highly respected large body of written literature or a previous period. The written literature is learned by formal education and is used in formal speech (such as news and political speeches), but not in every day conversation. All written materials (e.g., school books, news paper, commercial material, official documents, and instructional materials) are written in standard Arabic and in social settings people use the colloquial dialect. Thus children learn the colloquial dialect first and the standard language is taught formally in educational settings (Al-Kahtani, 1997; Altoma, 1969; Fatihi, 2001; Ferguson, 1956). With the increase in mass media and early education, children nowadays are more exposed to standard Arabic through television programs, radio and early reading experience. The differences between the two versions of Arabic are grammatical, phonetic, and lexical in nature. Standard Arabic is considered syntactically more complex and richer in lexicon. In spite of these differences, there are many similarities. Altoma (1969) found that 83.5% of words in different colloquial dialects are shared with standard Arabic, as well as the syllabic structures of words. In general standard Arabic is considered a more prestigious language but there is no competition between the two versions of language since each serves a different purpose and they are not totally interchangeable in use by

PAGE 35

23 situation. In other words there are situations where standard Arabic is inappropriate and visa versa (Al-Kahtani, 1997; Ferguson, 1959; Ibrahim, 2000). Abd-el-Jawad (1987) and Al-Kahtani (1997) reported that educated Arabic speakers frequently switch between standard and colloquial Arabic in a very natural manner. Differences are to some extent between social groups, but these differences have no effect on communication. Linguists have determined the status of a dialect based on the characteristic of mutual intelligibility among dialects. Mutual intelligibility is based on a scale of the physical proximity of the regions, that is, the closer the region the more mutually intelligible, and the further apart the regions, the less mutually intelligible (Fatihi, 2001). The differences and similarities between dialects and between standard Arabic and dialects must be considered when developing speech perception test materials and when choosing speakers. Several groups of researchers (Beattie, et al. 1975; Beattie, et al. 1977; Carhart, 1965; Doyne & Steer, 1951; Hirsh, et al. 1954; Hood & Poole 1980; Kruel, et al. 1969; Palmer, 1955; Wilson & Carter, 2001; Wilson & Oyler, 1997) demonstrated significant differences in speech perception scores with different speakers, regardless of gender or age, assuming the speech perception test material were recorded by native speakers of English in standard dialects. In the case of Arabic, it can be argued that there is no difference among the different Arab countries when using standard Arabic. The question whether standard Arabic is a representative sample of the spoken dialects remains unanswered. Special Considerations for the Present Study Participants Speech perception differs significantly between age groups. Elliot (1979) found that scores of children age 9 were poorer than older children on speech perception in

PAGE 36

24 noise test (SPIN), while the older group (15 to 17 year olds) scores were comparable to adult scores. Elliot, Connors, Kille, Levin, Ball & Katz (1979) found no significant difference in scores for children age 5 to 8 years while 10 year olds performed at adult levels (no 9 year old participants were included in this study). Schwartz and Goldman (1974) assessed the performance of young children in nursery, kindergarten and first grade and found significant differences between the three groups. In another study Sanderson-Leepa & Rintelmann (1976) compared the speech performance of children ages 3.5 to 11.5 on different speech perception tests and found no significant difference in the 7.5 and 9.5 age groups on the WIPI and PBK-50 tests compared to the younger and older groups. Ashoor & Prochazka (1985) found similar age differences for the Saudi Arabic test when comparing scores for 4 to 5 year old children to scores for 6 to 9 year old children. Based on these results, participants aged 6 to 9 years were recruited in the present study. As mentioned above, studies by Elliot (1979), Elliot et al. (1979), Sanderson-Leepa & Rintelmann (1976), Schwartz & Goldman (1974), and Goldman et al. (1970), show that significant differences in word recognition scores are found for children younger than 6 years and older than 9 years in comparison to children age 6 to 9 years. These others reported no significant difference in word recognition scores between children aged 6 through 9 years. Speakers Individual differences between speakers can affect speech perception scores (Hood & Poole 1980). Different speakers produce different articulation curves when using the same words. Most words maintain their order of difficulty across speakers. Palmer (1955) investigated the effect of gender on speech perception scores. He based his question on

PAGE 37

25 the notion that hard of hearing individuals have an easier time hearing mens voice than womens voice. In his study he used nine speakers; three male adults, three female adults, and three female children. When he compared scores from each group for hard of hearing and normal hearing participants at a fixed intensity level, no significant difference was found across speakers. One goal in the present study is to investigate the possibility of a difference in scores with different dialects. Following Palmers methodology, in the present study three Jordanian and three Saudi male speakers were selected to complete the word recording. Stimuli for Dialectal Differences In the present study the researchers goal was to investigate the possibility of difference in scores with different dialects since. In this study, a +9 dB signal to noise ratio was used to avoid ceiling effect in normal hearing childrens performance and to reduce the variability in scores. The choice of +9 dB signal to noise difference was based on the Goldman et al. (1970) study of speech perception of children in quiet and in noise. Goldman et al. (1970) observed a reduction in scores compared to the quiet condition that started at -9 dB noise level. Resulting word recognition scores are expected to be less than 100% correct. Schwartz & Goldman (1974) used the same level of signal to noise ratio (+9 dB) to investigate the effect of different contexts and listening environments (quiet and noise). They observed a significant increase in number of errors for all contexts when noise was introduced. The effect of the smallest amount of noise was clearly demonstrated in a study by Larson, Petersen, & Jacquot (1974) when they tested the use of NU-6 word lists with children age 5.5 to 6.5 years of age. The presence of noise at +20 dB S/N ratio had a significant effect on the childrens performance compared to adult performance under the same conditions. Keep in mind that for their

PAGE 38

26 study Larson et al. have used adult material to test very young children. Based on the above listed studies, a signal to noise level of + 9 dB will be used for the present study.

PAGE 39

27 Table 2-1 Childrens speech reception tests in English. Test Investigator Stimulus Response format Response task Target population Published PBK-50 Haskins 1949 Monosyll. Words Open set Verbal 6-9 years No GFW Goldman, Fristoe, & Woodcock 1970 Monosyll. Words Closed set Picture pointing > 4 years Yes Spondee recognition test Erber 1974 Spondee words Closed set Writing 8-16 years No WIPI Ross & Lerman 1970 Monosyll. Words Closed set Picture pointing 3-6 years Yes BKB Bench, Koval, & Bamford 1979 Sentences Open set Verbal 8-15 years No PSI Jerger & Jerger 1980 Monosyll words and sentences Closed set Picture ointing & verbal 3-10 years Yes NU-CHIPS Elliott & Katz 1980 Monosyll. Words Closed set Picture pointing > 2.5 years Yes ANT Erber 1980 Numbers Closed set Picture pointing 3-8 years No Table 2-2 Speech recognition word lists mean dB HL levels at 50% and slopes. Investigator Language Target population Test dB @ 50% Slope %/dB Wilson & Oyler 1997 English Adults CID W-22 15.6 4.8 Wilson & Oyler 1997 English Adults NU-6 13.4 4.4 Harris et al. 2003a Korean Adults Male speaker 11.4 5 Harris et al. 2003a Korean Adults Female speaker 10.7 5.1 Nissen et al. 2005a Chinese Mandarin Adults Male speaker 5.4 7.3 Nissen et al. 2005a Chinese Mandarin Adults Female speaker 2.3 8.2 Alusi et al. 1974 Arabic Adults 22.5 5 Haskins 1949 English Children PBK-50 NA 4

PAGE 40

28 Table 2-3 Speech reception threshold wordlists mean threshold dB HL levels and slopes between 20 and 80%. Investigator Language Target population Test dB @ 50% Slope %/dB Hirsh et al. 1952 English Adults CIDW-1 0.5 8 Yourng et al. 1982 English Adults CIDW-2 -0.3 10 Ashoor & Prochazka 1982 Arabic Adults Mono-syllabic 2.2 5 Harris et al. 2003b Korean Adults Male speaker 6.8 10.3 Harris et al. 2003b Korean Adults Female speaker 5.6 9 Nissen et al. 2005b Chinese Manderin Adults Male speaker -0.2 9.7 Nissen et al. 2005b Chinese Manderin Adults Female speaker -0.7 10.5 Siegenthaler et al. 1954 English Children Mono-syllabic NA 8.6 Ashoor & Prochazka 1985 Arabic Children Mono-syllabic 0 5

PAGE 41

CHAPTER 3 METHODOLOGY This dissertation consists of two studies. In the first study, monosyllabic word intelligibility was investigated as a function of presentation level in Jordanian dialect. The following questions were addressed in the first study: Is there a significant difference in word recognition ability of Jordanian children, given increases in intensity presentation levels? Can four parallel word lists be developed (i.e. their psychometric qualities do not differ)? The results of this study were used in the development of four parallel lists of fifty monosyllabic words to be used as a speech perception test in Jordan. The second study compared word recognition scores of Jordanian children listening to words spoken in Jordanian and Saudi dialects. Specifically, the study was designed to answer the following questions: Do word recognition abilities of Jordanian children differ when listening to words presented in a Jordanian dialect compared to words presented in a Saudi dialect? Do word recognition abilities of Jordanian children who display normal hearing and hearing disabilities differ when hearing words presented in a Jordanian dialect compared to words presented in a Saudi dialect? Does word recognition ability differ when hearing speakers using the same dialect? Participants Inclusion criteria included age range and hearing sensitivity. Age range was limited to children age six to nine years. Children displaying normal hearing sensitivity and mild to moderately severe hearing loss were recruited. Hearing status was determined 29

PAGE 42

30 using pure tone average thresholds of 500, 1000 and 2000 Hz. A pure tone average threshold of 15 dB HL or less were considered normal hearing and thresholds greater than 26 dB HL and less than 69 dB HL were considered within the range of mild to moderately severe hearing loss. Tympanometry and pure tone audiometry were used to determine hearing sensitivity fitting inclusion criteria. If the tympanogram showed normal middle-ear function, the researcher proceeded to pure tone audiometry using supra aural headphones (TDH-39). The participant was excluded if middle-ear function was abnormal. If the pure tone average did not meet criteria, the testing was terminated. In both cases results were explained to the parent if present. In case of hearing loss detection or abnormal middle-ear function, the caregiver or adult accompanying the participant was informed and briefly counseled on intervention. Participants were recruited by word of mouth from the larger Amman area. Participating children came to the clinic in Amman accompanied by their caregivers or an adult with the parents permission. Twenty normal hearing participants, ten males and ten females, were included in the first study. A total of ten normal hearing, four males and eight female, and ten hard of hearing, six male and four female, participated in the second study. For detailed description of participant recruitment see Appendix A. Speakers Speech materials were recorded by native Arabic speakers who were recruited from the area of Provo, Utah, the location of Brigham Young University where the recording took place. Potential speakers were initially personally contacted by phone or by word of mouth. The purpose of the study and the speakers role in the study were explained. Five adult Jordanian males agreed to participate in the recording. Preliminary 5-minute recordings of continuous speech were made for the purpose of judgment of

PAGE 43

31 dialect and clarity of speech. Five different adult Jordanian natives (three females and two males) were asked to judge the speakers using two criterion, i.e., the dialect was a general Jordanian dialect, and the ease of understanding the speakers speech as rated on a 10-point scale. The top ranked speaker completed the recording of the 304 monosyllabic words for study one. For the completion of recordings for study two the 3 top ranked Jordanian speakers were selected and agreed to participate in the recording. Only three Saudi speakers agreed to participate in the recording for study two. For the purpose of creating the 8 multi-speaker babble noise, four additional female Arabic speakers were recruited. The female speakers included the primary investigator and three females recruited form the Provo area by word of mouth. The recordings from the 2 male Jordanian speakers who were not included in the final recording and 2 of the speakers used for word recording randomly selected as well as the 4 female speakers were used to compose the noise track. All speakers have acquired Arabic in their home country. They were all students or spouses of students who have come to the USA to peruse higher education degrees. All speakers have been in the USA for no longer than 5 years. Recording and Editing The recording took place in an anechoic chamber at Brigham Young University campus in Provo, Utah, USA. The chamber has a 0 dB SPL noise floor allowing for a recording of 60 dB signal to noise ratio (S/N). The equipment used in the recording was the same as reported by Harris et al. (2003a) and Nissen et al. (2005a). The recording microphone used was a Larson-Davis model 2541, which has a flat frequency response up to 20 kHz. The microphone was positioned approximately 6 inches from the speaker at a 0 azimuth and covered by a 3 inch windscreen. The microphone signal was

PAGE 44

32 amplified by a Larson-Davis model 900B microphone preamp, coupled to a Larson-Davis model 2200C preamp power supply. The signal was digitized at a 44.1 KHz sampling rate with 24-bit quantization by an Apogee AD-8000 analog-to-digital converter, with preamp gain set to utilize the full range of the 24-bit analog-to-digital converter. The digitized signal was stored on a hard drive for later editing. The speech material was printed in large font on A4 paper and clipped to a clipboard that was suspended in front of the speaker at a comfortable height to allow him to read the words facing the microphone at 0 azimuth. The speaker was asked to say each word a minimum of four times. A native judge (the primary investigator) rated each word for clarity and the best production was selected for inclusion in the word recognition test. The intensity of each selected word was edited using Sadie Disk Editor software (Studio Audio & Video Ltd. 2004) to produce a final recording with the same average root mean square (RMS) power as the 1 kHz calibration tone in an attempt to equate the test word threshold audibility (Harris et al. 2003a; Nissen et al. 2005a; and Wilson and Strouse, 1999). The recording was converted from 24to 16-bit quantization using the NS high dither option in the Sadie Disk Editor software. The final word recording was digitally saved as wav files. Each word was saved as a separate file. A total of 250 words were selected to be included in the word recognition test; study one. For study two 33 monosyllabic words were selected from Ashoor and Prochazkas (1985) speech reception threshold list for children. The words were edited digitally to ensure same level of RMS as the 1 kHz tone. The multi speaker babble material was completed using 4 of the male Jordanian speakers and 4 Jordanian female speakers. The speakers were asked to read an article taken form a Jordanian daily newspaper. The recording was completed following the

PAGE 45

33 same procedure as for the word list. Three minutes of continuous speech was selected for each of the 8 speakers. All silence between words and sentences was removed for each speaker. Then the RMS level for each speaker was adjusted to have all 8 speakers at the same level. Next the recordings were digitally mixed and run through a compression/limiter to limit the range of peak variability. The resulting recording average RMS was measured and adjusted to equal the level of the 1 kHz calibration tone. The 8 multi-speaker babble noise was saved in a separate wav file that would allow the play of noise independently from the word lists. Instrumentation for Data Collection Data collection took place at the Middle East Hearing Association clinic in Amman, Jordan in a sound treated booth designed locally for hearing testing purposes. Pure tone and speech audiometry were conducted using an Interacoustics AC40 audiometer, connected to TDH-39 headphones. The audiometer was calibrated using a Larson Davis System 824 sound level meter and a 6cc coupler. Calibration was based on ANSI standards 2004. Measurements of sound levels at octave and half octave frequencies met the ANSI standard 2004 with a deviation range of -0.6 to +0.3 dB. The sound levels for speech through external input A and external input B were consistent with ANSI standard 2004 with a deviation of -0.5 to +0.1 dB. Repeated measurements of sound pressure level produced by the audiometer were within permissible ANSI tolerance level of + 3dB for frequencies of 500 to 4000Hz and + 5dB for 6000 to 8000 Hz. Special software was developed by Dr. Richard Harris at Brigham Young University (2005) to control the playback of 1 kHz tone, noise and word lists from wav files. The software also provides the documentation of data in an excel file spread sheet with the following details: the date and time of presentation, participant assigned number,

PAGE 46

34 participant gender, test ear, speaker gender, intensity level, signal to noise ratio, list name, time of recording per list, wav file, word (in this case in Arabic), and the score. Prior to data collection, the VU meter was adjusted to 0 VU using 1 KHz tone. For the purpose of consistency in speech audiometry data collection, the left headphone was randomly selected to always be used over the test ear. This step was taken to reduce variability in the sound level presented. Study One Speech Material Speech material was selected following Egans (1948) criterion of monosyllabic words, representative of spoken Arabic and commonly used words. A total of 304 monosyllabic words were selected for initial recording. Words that have similar forms in colloquial and standard Arabic were selected. Children and teachers of first through third grade rated these words as familiar. Rating took place at four elementary schools in Amman Jordan. The words were read aloud to the children, they were asked to raise their hand if they knew the meaning of the word and were able to use it in a meaningful sentence. To ensure the accuracy of their response, the primary investigator (Nadia Abdulhaq) started with three trial words that were familiar to children (window, door, and table), randomly asked children to put words in a sentence, in addition unfamiliar words (from old classical text) and nonsense words were included. For more detailed description of word selection see Appendix B. The primary investigator judged the recorded 304 words for clarity of recording and in conjunction with familiarity ratings selected 250 words to be used in data collection. The 250 words were divided randomly into two sets of ten lists of 25 words. To allow the presentation using the software developed by Dr. Harris, a text file was

PAGE 47

35 created for each list A1 to A10 and B1 to B10 for each set of randomization respectively. Lists were presented in counterbalanced order. Lists A1 through A10 were presented to participants 1 through 10 and lists B1 through B10 were presented to participants 11 through 20. See Table 3-1 for order of presentation and level of presentation. The signal was routed to the subjects through the Interacoustics AC40 audiometer, via the TDH-39 headphones. Speech material was presented at 10 different intensity levels starting at 0 dB HL and increasing by 5 dB increments up to 45 dB HL; to include a range of low to high word recognition scores. Procedure for Data Collection Each participant was given the following general instructions: You will hear some words through the headphones, your will repeat the words you hear. For example, if you hear the word /kitab/ (book), you would say? The researcher would wait for the participant to repeat the word. That is good. And if you hear the word /daftar/ (copy book) you would say? and the word /madrasa/ (school) you would say? Once the participant has followed instructions further details were provided: The words will be spoken very softly; you might or might not hear them at first. That is OK. If you hear a word, repeat it. If you dont, just wait to hear the next one. The words will get louder and louder. The earphone was placed over the test ear. Two lapel microphones were attached to the participants shirt, one for talk back and the second for audio input of the audio-visual recording. A video camera was set in front of the participant to obtain video recording. The video camera was adjusted in height and angle to provide the best view of the participants face.

PAGE 48

36 After the participant repeated a word, the investigator entered 1 for a correct repetition and 0 for an incorrect repetition. If the participants did not repeat the word, the investigator periodically used the talk forward microphone to ask the participant if he/she had heard the word, and to encourage him/her to repeat whatever they heard. The investigator waited 3to 5 seconds before recording an incorrect score in the case of no repetition. After each list presentation, the presentation level was increased by 5dB. Participants were encouraged between lists using verbal praise (good job, or you are doing really well). Most participants did not need a break or rest period throughout the data collection procedure. All data for each participant, including presentation order and scoring details, were saved in excel spreadsheets. Statistical Analysis The dependent variable, defined as correct word recognition, was saved in binary format (correct versus incorrect). The independent variables were different presentation intensity levels and different word lists. The raw data were used in logistic regression analysis. The logistic regression analysis provides derived variables that are key parameters for the regression line that may be tested and compared between conditions (in this study, the word lists). The raw scores of each word in the four lists were used in the logistic regression analysis to calculate the logistic parameters for each list and half list. See Appendix C for details on logistic regression analysis. Recall that this study was designed to test the following hypotheses: 1. Increased presentation sound intensity level will improve word recognition ability. 2. It is possible to create four word lists that are parallel in function and can be used interchangeably to test Jordanian childrens word recognition ability.

PAGE 49

37 Study Two Speech Material The speech material consists of 33 monosyllabic words from Ashoors (1985) childrens list. Ashoors list was used as source of material because it was developed for the purpose of Arabic speech perception specifically for children. The selection of the 33 words was based on two criteria: monosyllabic words and familiar to Jordanian children. Recall that Ashoors list consisted of a mixture of monoand di-syllabic words, familiar to Jordanian Arabic speaking children. The familiarity of words to Jordanian Arabic speaking children was established through ratings provided by children in 1 st through 3 rd grades (Abdulhaq, unpublished). See Appendix D for the list of 33 words and their meaning in English. A total of six recordings, each recorded by a different speaker (three Jordanian and three Saudi speakers), were presented to all participants in counterbalanced order at constant intensity levels (see table 3-2 for details). The speech material was presented by playing wav files from a standard desktop PC connected to external inputs of the audiometer using the same software as described in study one. The words were presented to the hard of hearing participants at an audible level of 40 dB SL (re: the participants pure tone average). For normal hearing participants, the multi-speaker babble track was presented then the word list was presented. Words were presented at 50 dB HL and the multi-speaker babble noise at 41 dB HL, i.e., with a signal to noise ratio (SNR) of +9 dB. Statistical Analysis For the purpose of statistical analysis of the data in these studies, the dependent variable was defined as the correct word recognition score, and the independent variables

PAGE 50

38 were defined as dialect, speakers, and hearing status. Raw scores were used in mixed ANOVA to determine difference between speakers and dialects. To determine the effect of dialect on normal hearing and hard of hearing childrens scores a paired t-test of dialects based on scores from each group separately was used. To determine the significance of differences between speakers within a dialect paired t-tests of speakers were used based on scores from all participants.

PAGE 51

39 Table 3-1 Order of list presentation by participant and level of presentation Participant Presentation dB HL level 0 5 10 15 20 25 30 35 40 45 1 A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 2 A2 A3 A4 A5 A6 A7 A8 A9 A10 A1 3 A3 A4 A5 A6 A7 A8 A9 A10 A1 A2 4 A4 A5 A6 A7 A8 A9 A10 A1 A2 A3 5 A5 A6 A7 A8 A9 A10 A1 A2 A3 A4 6 A6 A7 A8 A9 A10 A1 A2 A3 A4 A5 7 A7 A8 A9 A10 A1 A2 A3 A4 A5 A6 8 A8 A9 A10 A1 A2 A3 A4 A5 A6 A7 9 A9 A10 A1 A2 A3 A4 A5 A6 A7 A8 10 A10 A1 A2 A3 A4 A5 A6 A7 A8 A9 11 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 12 B2 B3 B4 B5 B6 B7 B8 B9 B10 B1 13 B3 B4 B5 B6 B7 B8 B9 B10 B1 B2 14 B4 B5 B6 B7 B8 B9 B10 B1 B2 B3 15 B5 B6 B7 B8 B9 B10 B1 B2 B3 B4 16 B6 B7 B8 B9 B10 B1 B2 B3 B4 B5 17 B7 B8 B9 B10 B1 B2 B3 B4 B5 B6 18 B8 B9 B10 B1 B2 B3 B4 B5 B6 B7 19 B9 B10 B1 B2 B3 B4 B5 B6 B7 B8 20 B10 B1 B2 B3 B4 B5 B6 B7 B8 B9

PAGE 52

40 Table 3-2 Word list and dialect presentation order. Participants Present. order 1st 2nd 3rd 4th 5th 6th 1 J1 S1 J2 S2 J3 S3 2 S1 J1 J3 S3 S2 J2 3 S2 J2 J1 S1 S3 J3 4 J2 S2 S3 J3 J1 S1 5 S3 J3 S1 J1 J2 S2 6 J3 S3 S2 J2 S1 J1 7 J1 S1 J2 S2 J3 S3 8 S1 J1 J3 S3 S2 J2 9 S2 J2 J1 S1 S3 J3 10 J2 S2 S3 J3 J1 S1 J = Jordanian speaker, S = Saudi speaker. The same order of presentation was used for the normal hearing and hard of hearing group.

PAGE 53

CHAPTER 4 RESULTS Study One Twenty individuals participated in this study. Pure tone thresholds average (PTA) indicated that all participants fit the criteria of a PTA of 15dB HL or better with a mean PTA of 7.8 dB HL (standard deviation = 2.8, range = 3.3 to 12.5 dB HL). The better ear was selected based on PTA or at random when both ears had the same PTA. Ten participants were tested on the right ear and 10 were tested on the left ear. Average pure tone thresholds for both ears of all participants are shown in Figure 4-1. For detailed description of participants see Table 4-1. Most participants completed the 250 words in a single session of 20 and 29 minutes without a break other than the time it took to load each new list and adjust the intensity of presentation. The pacing of the test depended on how quickly participants responded; the program required the examiner to record the response before the next word could be presented. The raw score data of all participants at the 10 intensity levels were compiled for each of the 250 words. The total of responses resulted in a score of correct recognition out of 20 possible correct score per word. Words then were reordered by difficulty from easiest to most difficult based on the total score (with lower scores indicating higher difficulty). Four lists of 50 words each were created using the top 200 words ranked by difficulty, as follows: Words that were tied in difficulty ranking were randomized prior to 41

PAGE 54

42 sorting into lists. To help ensure range of difficulty between lists, the four most difficult words were assigned randomly to four lists. The next four most difficult words again were assigned randomly to the four lists and so on until a total of 50 words were included in each list. The lists were labeled 1, 2, 3, or 4. Within each list, the words level of difficulty was used to assign it in order to create two half lists of 25 words from each original 50-item list with the goal of having an equal range of difficulty on both half lists. The half lists were labeled 1A and 1B (the two lists taken from list 1) through 4A and 4B (the two lists taken from list 4). This procedure used, to distribute words into lists, has been successful in developing difficulty equivalent word lists in studies by Harris et al. (2003a and 2003b) and Nissen et al. (2005a and 2005b). The eight 25-Arabic monosyllabic word half lists are presented in Tables 4-2. The words are written in Arabic with their phonetic transcription, please see appendix E for the English translations of the words. Word recognition scores increased with the increase of intensity. A scatter plot of the percent correct of responses, for the half lists (25 words) as a function of stimulus level, is presented in Figure 4-1. Scores at the lowest intensity level of 0 dB HL ranged from 0 to 2% while scores at the highest intensity level of 45dB HL ranged from 94 to 96%. Chi-square analysis ( 2 ) of the half lists showed a highly significant effect of intensity ( 2 = 1,134.18, p <0.0001). Statistical analysis of data confirmed, as expected, that scores at different intensity levels are not equal. Differences between lists ( 2 (3) = 1.67, p =0.64) and between half lists ( 2 (7) = 8.25, p = 0.31) were not significant. The mean slope of full lists at the 50% point was 4.44%/dB and ranged from 4.21%/dB to 4.63%/dB (range = 0.42%/dB). The mean slope

PAGE 55

43 at the 20-80% range was 3.84%/dB ranging from 3.65 to 4.01%/dB (range = 0.36%/dB). The mean threshold was 21.25 dB, and ranged from 21.21 dB to 21.32 dB (range = 0.107 dB). For the half lists, the mean slope at 50% was 4.47%/dB and ranged from 3.86 to 5.36%/dB (range = 1.5%/dB). The mean slope at the 20-80% range was 3.87%/dB with a range from 3.34 to 4.64%/dB (range =1.30%/dB). Mean threshold was 21.25 dB, and ranged from 21.04 dB to 21.38 dB (range = 0.34 dB). Table 4-3 and Table 4-4 include the detailed results of the logistic regression parameters, the slopes at the 50% and 20-80% levels, threshold intensity level, as well as threshold deviation from mean threshold, for the lists and half lists respectively. Figure 4-3 shows the psychometric function for all four lists and Figure 4-4 shows the psychometric function for the eight half lists. By visual inspection, the curves are almost identical. The intensity levels at the threshold point had a narrow range around the mean value, from -0.03 to 0.11 dB for the full lists and from -0.21 to 0.13 dB for the half lists. The difference between the calculated threshold and the mean threshold for each list and half list was used to adjust the intensity of each list and half list digitally to achieve a better equivalence between lists and half lists. The percent correct was recalculated based on these adjustments for all lists and half lists and an adjusted logistic regression curve was created. Figure 4-5 shows the difference between the unadjusted and adjusted curves. The difference between the unadjusted and adjusted curves was small and barely detectable. Study Two Ten normal hearing and 10 hard of hearing individuals participated in this study. Pure tone threshold average measures indicated that all participants fit the criteria of a PTA of 15dB HL or better with mean PTA of 6.3 dB HL (SD = 2.8, minimum = 1.7dB

PAGE 56

44 HL and maximum = 10dB HL). The better ear was selected based on PTA or at random when both ears had the same PTA; as a result 6 participants were tested on the right ear and 4 were tested on the left ear. Figure 4-6 shows the average of pure tone thresholds of normal hearing participants, for detailed description see Table 4-5. Pure tone threshold average measures indicated that all ten hard of hearing participants had a mild to moderately severe hearing loss with a mean PTA of 50.2 dB HL (SD = 11.8, minimum = 31.7dB HL and a maximum = 63.3dB HL). All but two participants had sensorineural hearing loss (SNHL). One participant (#2 in table 4-6) had chronic conductive hearing loss and one participant (#7 in table 4-6) had a mixed loss. Again the better ear was selected based on PTA or at random when both ears had the same PTA; resulting in 5 right and 5 left ears being used. Figure 4-7 shows the average of pure tone thresholds of hard of hearing participants, for detailed description see Table 4-6. Most participants completed all six lists of 33 words in a range of 13 to18 minutes, and did not require a break in data collection. The pacing of data collected depended on how quickly participants responded; the program required the examiner to record the response before the next word could be presented. After data collection was completed the score for each participant per speaker was calculated into percent correct. See Table 4-7 for detailed results. These data were analyzed to evaluate the hypotheses. Mixed analysis of variance (ANOVA) indicated a significant difference between dialects and speakers. Scores of all participants for all speakers were included in a mixed ANOVA of speaker and dialect. In summary, analysis showed a significant difference between dialects (F= 8.865, p = 0.008), significant differences among speakers (F =

PAGE 57

45 6.181; p = 0.005), and a significant difference between speakers within each dialect (F = 5.153; p = 0.011). For detailed results see Table 4-8. Paired sample t-test analysis indicated significant difference between dialects for the normal hearing group, but no significant difference for the hard of hearing group. The scores for each participant were averaged by dialect, and the averages were used in paired sample t-test for each group separately (hard of hearing and normal hearing). Difference between dialects for the normal hearing participant group were significant (t-test = -2.923 p = 0.017). Differences between dialects for the hard of hearing participant group were not significant (t-test = -1.327 p = 0.217). Thus, hearing loss seemed to outweigh dialectal difference. See Table 4-9 for detailed results. Within each dialect one speaker was significantly different than the other two speakers. A paired sample t-test was performed comparing results of speakers within each dialect for all participants. Results indicated that scores associated with speaker J1 (in the Jordanian dialect) were higher than those for J2 and J3 (t = 4.203 and 3.802, respectively; p = 0.0001 and 0.001). Paired scores associated with speakers J2 and J3 were not different (t-test -0.496, p= 0.625). For the Saudi dialect, scores associated with speaker S2 were significantly lower than those for S1 and S3 (t-test = 2.658 and -2.183 respectively and p = 0.016 and 0.042). Differences between speakers S1 and S3 were not significant (t-test = 0.047 and p = 0.962). See Table 4-10 for detailed results.

PAGE 58

46 01020304050607080901002505001000200040008000Frequency in HzIntensity in dB HL Right Left Figure 4-1 Average pure tone thresholds at all frequencies for all 20 normal hearing participants at 250, 500, 1000, 2000, 4000, and 8000 Hz. Half List Raw Scores00.10.20.30.40.50.60.70.80.9102468101214161820222426283032343638404244464850PresentationLevel (dB HL)Percentage Correcct Recognition (%) 1A 1B 2A 2B 3A 3B 4A 4B Figure 4-2 Half lists raw data scatter plot

PAGE 59

47 0102030405060708090100110135791113151719212325272931333537394143454749515355 Series1 Series2 Series3 Series4 Figure 4-3 Psychometric function of 4 lists of 50 words based on calculated percent correct. 0102030405060708090100110135791113151719212325272931333537394143454749515355 1A 1B 2A 2B 3A 3B 4A 4B Figure 4-4 Psychometric function of 8 half lists of 25 words based on calculated percent correct.

PAGE 60

48 020406080100 Lists 1-4 Adjusted Lists 1-4 UnadjustedHalf-lists 1A-4B AdjustedHalf-lists 1A-4B Unadjusted 01020304050020406080100 Percentage Correct Recognition (%)Presentation Level (dB HL)01020304050 Figure 4-5 Arabic monosyllabic psychometric functions for lists 1-4 (left) and half-lists 1A-4B (right). The top two figures depict unadjusted psychometric functions and the bottom two figures depict psychometric functions adjusted for equal performance at 50% correct recognition.

PAGE 61

49 01020304050607080901002505001000200040008000Frequency in HzIntensity in dB H L Right Left Figure 4-6 Average of pure tone thresholds of normal hearing participants at 250, 500, 1000, 2000, 4000, and 8000 Hz. 01020304050607080901002505001000200040008000Frequancy in HzIntensity in dB HL Right Left Figure 4-7 Average of pure tone thresholds of hard of hearing participants at 250, 500, 1000, 2000, 4000, and 8000 Hz.

PAGE 62

50 Table 4-1 Normal Hearing participants age, gender, test ear and PTA Participant Gender Age Test ear PTA 4 F 6 Left 6.7 8 F 6 Right 10.8 10 F 6 Right 5.0 5 F 7 Left 9.2 7 F 7 Left 9.2 12 M 7 Right 3.3 15 M 7 Right 7.5 16 M 7 Right 5.8 19 F 7 Right 5.8 20 M 7 Right 6.7 6 F 8 Right 12.5 11 M 8 Right 4.2 1 M 9 Left 6.7 2 M 9 Left 4.2 3 F 9 Right 9.2 9 F 9 Left 9.2 13 M 9 Left 10.8 14 M 9 Left 6.7 17 M 9 Left 11.7 18 F 9 Left 11.7 Mean 7.8 7.8 Standard Deviation 1.2 2.8 Range 3.0 9.2 Minimum 6.0 3.3 Maximum 9.0 12.5

PAGE 63

51 Table 4-2 Half word lists and transcription in IPA. 1A 2A 3A 4A : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

PAGE 64

52 Table 4-2 continued 1B 2B 3B 4B : : : : : : : : : : : lo: : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

PAGE 65

53 Table 4-3 Logistic regression results for the full lists, the calculated slopes at the 50% and the 20-80% levels, the threshold dB level, and difference of threshold levels from the mean threshold List a b Slope at 50% Slope from 20-80% Threshold dB dB difference 1 3.918 -0.185 4.614 3.994 21.225 -0.021 2 3.574 -0.169 4.212 3.646 21.214 -0.033 3 3.663 -0.172 4.295 3.718 21.321 0.074 4 3.930 -0.185 4.629 4.007 21.226 -0.020 M 3.771 -0.178 4.438 3.841 21.246 0.000 Minimum 3.574 -0.185 4.212 3.646 21.214 -0.033 Maximum 3.930 -0.169 4.629 4.007 21.321 0.074 Range 0.356 0.017 0.416 0.361 0.107 0.107 SD 0.180 0.009 0.215 0.186 0.050 0.050 Table 4-4 Logistic regression results for the half lists, the calculated slopes at the 50% and the 20-80% levels, the threshold dB level, and difference of threshold levels from the mean threshold Half Lists a b Slope at 50% Slope from 20-80% Threshold dB dB difference 1A 3.748 -0.177 4.426 3.832 21.170 -0.077 1B 4.104 -0.193 4.821 4.173 21.281 0.034 2A 3.969 -0.186 4.642 4.018 21.379 0.132 2B 3.249 -0.154 3.860 3.341 21.040 -0.206 3A 3.663 -0.171 4.285 3.709 21.370 0.123 3B 3.663 -0.172 4.305 3.726 21.272 0.025 4A 3.498 -0.164 4.093 3.543 21.363 0.117 4B 4.522 -0.214 5.361 4.640 21.087 -0.159 M 3.802 -0.1790 4.47 3.87 21.25 0.00 Minimum 3.249 -0.2144 3.86 3.34 21.04 -0.21 Maximum 4.522 -0.1544 5.36 4.64 21.38 0.13 Range 1.273 0.0600 1.50 1.30 0.34 0.34 SD 0.392 0.0187 0.47 0.40 0.13 0.13

PAGE 66

54 Table 4-5 Selected characteristics of normal hearing participants Participant Gender Age Test Ear PTA 14 M 6 Right 5.0 22 M 6 Left 10.0 10 F 7 Right 8.3 11 M 7 Right 8.3 3 F 8 Left 3.3 4 F 8 Left 1.7 5 F 8 Left 10.0 6 M 8 Right 5.0 20 F 8 Right 6.7 21 F 9 Right 5.0 Mean 7.5 6.3 Standard Deviation 1.0 2.8 Range 3.0 8.3 Minimum 6.0 1.7 Maximum 9.0 10.0 Table 4-6 Selected characteristics of hard of hearing participants Participant Gender Age Test Ear PTA Hearing Loss Type 8 F 6 Right 63 SNHL 16 M 6 Left 48 SNHL 18 M 6 Left 62 SNHL 2 F 8 Right 32 Conductive 7 M 8 Right 52 Mixed 1 F 9 Right 32 SNHL 15 M 9 Left 58 SNHL 17 M 9 Left 60 SNHL 19 F 9 Right 53 SNHL 23 M 9 Left 42 SNHL Mean 7.9 50.2 Standard Deviation 1.4 11.8 Range 3.0 31.7 Minimum 6.0 31.7 Maximum 9.0 63.3

PAGE 67

55 Table 4-7 Word recognition scores in percent correct for each subject per talker-talker and the average of scores by dialect. Average Group Parti-cipant J1 J2 J3 S1 S2 S3 Jordanian Saudi 1 1 0.94 0.91 0.97 0.94 1.00 0.97 0.94 0.97 1 2 0.94 0.94 0.97 0.85 0.88 0.97 0.95 0.90 1 8 0.58 0.30 0.42 0.45 0.24 0.42 0.43 0.37 1 15 0.82 0.76 0.67 0.82 0.73 0.61 0.75 0.72 1 16 0.70 0.70 0.45 0.61 0.61 0.55 0.62 0.59 1 17 0.70 0.52 0.67 0.70 0.61 0.73 0.63 0.68 1 18 0.58 0.64 0.52 0.79 0.79 0.64 0.58 0.74 1 19 0.70 0.48 0.82 0.76 0.61 0.82 0.67 0.73 1 23 0.97 0.64 0.85 0.85 0.94 0.94 0.82 0.91 1 7 0.67 0.39 0.67 0.70 0.52 0.76 0.58 0.66 2 6 0.94 0.55 0.76 0.82 0.73 0.85 0.75 0.80 2 5 0.82 0.73 0.82 0.79 0.73 0.79 0.79 0.77 2 4 0.85 0.61 0.79 0.67 0.79 0.97 0.75 0.81 2 3 0.70 0.64 0.64 0.79 0.82 0.88 0.66 0.83 2 10 0.88 0.79 0.58 0.91 0.67 0.52 0.75 0.70 2 11 0.79 0.82 0.42 0.85 0.67 0.82 0.68 0.78 2 22 0.82 0.70 0.55 0.88 0.85 0.76 0.69 0.83 2 14 0.94 0.97 0.88 0.97 0.91 0.97 0.93 0.95 2 20 0.82 0.58 0.73 0.88 0.64 0.91 0.71 0.81 2 21 0.85 0.82 0.73 0.82 0.82 0.94 0.80 0.86 Group1 = hard of hearing, Group 2= normal hearing, J1= Jordanian talker 1, J2= Jordanian talker 2, J3= Jordanian talker 3, S1= Saudi talker 1, S2= Saudi talker 2, S3 = Saudi talker3 Table 4-8 Mixed ANOVA results comparing dialects for all participants' scores df Mean Square F-test Sig. Dialect 1 0.066 8.865 0.008 Talkers 2 0.091 6.181 0.005 Talker within dialect 2 0.027 5.153 0.011 Table 4-9 Paired Sample T-tests comparing dialectal difference in the hard of hearing group and normal hearing group. Group df T-test Sig. (2-tailed) Hard of Hearing 9 -1.327 0.217 Normal Hearing 9 -2.923 0.017

PAGE 68

56 Table 4-10 Paired Sample T-test results including all participants scores Pairs df T-test Sig. (2-tailed) J1-J2 19 4.203 0.0001 J1-J3 19 3.802 0.001 J2-J3 19 -0.496 0.625 S1-S2 19 2.658 0.016 S1-S3 19 0.047 0.963 S2-S3 19 -2.183 0.042

PAGE 69

CHAPTER 5 DISCUSSION The primary goal of the present study was to create four equivalent lists of words for clinical assessment of speech perception in Jordanian Arabic speaking children. There are no published word lists developed for the specific purpose of clinical speech audiometry (word recognition) for Jordanian Arabic speaking populations. Although other speech recognition materials have been developed and published over the past 3 decades, no study has evaluated the feasibility of applying Arabic speech audiometry materials among different countries or different Arabic language dialects (e.g., the use of Egyptian Arabic word lists in Jordan). In addition, recorded materials are not widely distributed and, thus, not commonly used in audiology clinics within Arabic countries. Four lists of 50 words were developed in the present study. The lists were not significantly different in their psychometric functions. All had the same shape of curve and very similar slopes (mean 4.4 %/dB, S.D. 0.215) and threshold points (mean 21.25 dB, S.D. 0.05), properties that are appropriate for clinical use as speech recognition measures. The word lists developed in the present study were comparable in slope and curve shape to word lists published in English (Haskins, 1949; Wilson & Oyler, 1997), Korean (Harris et al. 2003a), and other words lists published in Arabic (Alusi et al. 1974). The point of threshold (50% correct word recognition) was comparable to the Arabic adult lists reported by Alusi et al. (1974). The point of threshold for the Arabic word lists in the present study differed from other languages, suggesting the likelihood of a language specific feature. The range of thresholds for the word recognition materials in 57

PAGE 70

58 the present study (from 2.3 to 22.5 dB HL), confirm the appropriateness of the materials as measures of word recognition ability. There was no attempt in the present study to develop words lists that were phonetically balanced. Carhart (1965) stated that the phonetically balanced CNC lists and W-22 lists approximated the phonetic balance of every day spoken English but cannot be true representation of everyday phonetic balance. The effect of familiarity of words on speech perception performance is greater than the phonetic balance of the word lists. Owens (1961) studied the effect of word familiarity on word recognition. He found that listeners were more likely to make errors on less familiar words and, when they made the errors, the listeners response was more likely to be a familiar word. Martin (2000) stated that phonetic balance is not the only, or the main, factor in word list equivalence. Therefore, an objective of the present study was to utilize familiar words with a balanced range of difficulty between lists. In the present study, the psychometric curves of the full lists and half lists were very similar. Statistical analyses of the full lists ( 2 (3) = 1.67, p =0.64) and half lists ( 2 (7) = 8.25, p = 0.31) showed no significant difference between word lists and half lists despite the fact that no effort was made to account for phonetic balance. In comparison to the speech materials reported previously (Alusi et al., 1974; Ashoor & Prochazka, 1982, and 1985) the recordings developed in the present study can easily be applied clinically by audiologists in Arabic-speaking patient populations with a CD player and distributed to audiologists throughout the Arab countries. However, the issue of dialectal differences requires further investigation. Specifically, the possible

PAGE 71

59 effect of dialect on word recognition performance should be assesses for children in different Arabic speaking countries. It was hypothesized that participants in this study would perform better when listening to words spoken in Jordanian versus Saudi dialect and, in addition, that the effect would be similar for both normal hearing and hard of hearing children. Dialect appeared to exert an effect on word recognition ability of normal hearing Jordanian Arabic speaking children. Results reported herein confirmed that word recognition scores were higher when normal hearing children listened to words spoken in Saudi dialect. In contrast, no significant effect for Jordanian versus Saudi dialect was found in word recognition ability of children with mild to moderately severe hearing loss In general, word recognition performance was lower and more variable among hard of hearing children than for normal hearing children, independent of dialect. However, the higher word recognition scores for the Saudi dialect were persistent in both hard of hearing and normal hearing groups. Perhaps because each list was limited to 33 words that were highly familiar to the target population (i.e., Jordanian Arabic speaking children age 6 to 9 years). In addition, the words were spoken similarly in standard Arabic and in both Jordanian and Saudi dialect, with the speaker remaining as the only potential factor affecting performance. Another consideration affecting data reported in the present study is the small sample size. Only 10 participants were included in each group, with three speakers in each dialect. Although the power analysis requirement was met by statistical criteria (power .95), a larger sample size would probably yield more representative results.

PAGE 72

60 The results from the present study are not consistent with Alusis (1974) contention that his word lists could be used in all Arab countries since the word lists were derived from standard Arabic. There are several clear differences between the present study and the study reported by Alusi. For example, the subjects in Alusis study were educated adults, whereas subjects in the present study were children. Also, the sample size (N = 17) in Alusis study was inadequate for comparison of performance from different dialectal backgrounds. Furthermore, the speaker in Alusis study used Baghdad general dialect, whereas in the present study the speaker used general Jordanian and Saudi dialects. These differences among studies highlight the need for further investigation to compare word recognition performance with the speech materials from the present study with children from other Arabic speaking countries, particularly in relation to psychometric functions and the familiarity of words. Further clinical investigations should be conducted utilizing the word lists developed in this study. A first step would be to collect data for standardization of the word lists. Several variables must be considered in the collection of standardization data. Data should be collected from a large sample of the target population that is representative of rural as well as urban Jordan. In addition, the sample should include participants of all socioeconomic groups, as well as children who display normal and impaired hearing abilities. Another step is to establish test reliability. Factors to be considered are lists versus half list reliability, children who display normal and impaired hearing, as well as gender (males and females). Although the present study focused on word recognition, there is a need to develop other speech diagnostic materials that for evaluation of speech perception abilities of

PAGE 73

61 younger and older pediatric populations, as well as populations with different auditory problems, such as auditory processing disorder. Diagnostic speech audiometry materials that are needed clinically include: A word recognition by picture identification test for children younger than 6 years. The words used in the present study could be assessed for familiarity with younger children and on the basis of the availability of picture representations of the words. A picture-pointing word recognition measure would be very helpful in diagnosing and monitoring benefit of hearing aids and intervention in younger children. Material for central auditory processing diagnosis, such as filtered words, figure ground tests, and dichotic tests could be developed based on the present word lists. The fact that the words are digitally recorded facilitates the process of developing additional tests. Speech reception tests for estimation of threshold are lacking for Arabic-speaking children and adults. The procedures used to develop the speech materials in the present study have shown to be effective as in other studies by Harris et al. (2003b) and Nissen et al. (2005b), and can be duplicated. Results from the present study will play a significant role in the development of additional Arabic speech audiometry materials and in improving audiological services provided in Jordan. The technique for recording the speech materials and the data collection conditions in the present study render these word lists to be of high quality. The speaker was carefully chosen to have clear speech and representative of general Jordanian dialect. The words were selected based on familiarity to children who are native speakers of Arabic in Jordan. The data were collected using headphones, the most common method for presenting speech signals in this age group. And, finally, the words were digitally recorded under excellent recording sound conditions providing a very clear signal, and materials that can be easily duplicated for the production of multiple audio copies of the word lists and half lists The word recognition materials developed in the present study are likely to be of considerable value in providing diagnostic services and

PAGE 74

62 intervention to the pediatric hard of hearing population in Jordan. In addition, these word recognition materials will contribute to the rapid development of diagnostic speech audiometry materials that are currently lacking in Arabic countries.

PAGE 75

63 Table 5-1 Descriptive statistics for Hard of Hearing and Normal Hearing based on dialect Hard of Hearing Normal Hearing Jordanian Saudi Jordanian Saudi Mean 0.69 0.72 0.74 0.81 Standard Deviation 0.18 0.18 0.13 0.10 Range 0.66 0.76 0.55 0.46 Min 0.30 0.24 0.42 0.52 Max 0.97 1.00 0.97 0.97

PAGE 76

APPENDIX A PARTICIPANT RECRUITMENT The initial pool of potential participants with normal hearing included 18 children; five of them were excluded after the initial pure tone audiometric testing due to their elevated thresholds that have exceeded the inclusion criteria of 15dB HL at least at one frequency. Further testing using bone conduction audiometry revealed mild conductive hearing loss. The childrens caregivers were counseled and retest was recommended. In one case mild sensorineural hearing loss was diagnosed. This participant was included in the hard of hearing group. Data from 2 participants was excluded due to technical problems during data collection, which rendered results unreliable. Among the 20 potential participants in the hard of hearing group seven participants were excluded for having a severe hearing loss, based on the screening, and some had developmental delays that they were not able to understand the task. 64

PAGE 77

APPENDIX B SPEECH MATERIAL SELECTION The words in this study were chosen from three different sources: a word frequency lists provided by Landau (1959) which was derived from daily newspapers from Lebanon, Egypt, Iraq, and Palestine, and modern prose; lists of Arabic key words for learners of everyday Arabic as a second language (Quitregard, 1994); and from childrens story books. To achieve Egens (1948) criteria for word selection, the researcher selected words that fit one of the following possible monosyllabic word structures in Arabic: CV, CVC, CVCC, CVVC, and CVV. According to Altoma (1969) these monosyllabic word structures are shared by colloquial and standard Arabic. As mentioned above, words were selected from a range of written material. To ensure that selected words are representative of spoken Arabic, the procedure employed by Alusi (1974) and Ashoor and Prochazka (1982 and 1985) were followed. Words that have similar forms in colloquial and standard Arabic were selected. This step can be supported by Altomas (1969) finding that 83.5% of the words in a colloquial form have shared origins with words in standard Arabic. Words from Landau (1959) and Quitregard (1994) and a list of (135 words) that are familiar to first, second and third grade Jordanian Arabic speaking students (Abdulhaq, unpublished) were used. Abdulhaqs familiar word list is composed of monosyllabic words chosen from childrens story books and Jordanian daily newspaper articles, and rated by 320 first, second, and third grade students as familiar or unfamiliar. The words 65

PAGE 78

66 were read aloud to the children, they were asked to raise their hand if they knew the meaning of the word and were able to put it in a meaningful sentence. To ensure the accuracy of their response, Abdulhaq started with three trial words that were familiar to children (window, door, and table), randomly asked children to put words in a sentence, in addition unfamiliar words (from old classical text) and nonsense words were included. The initial list included 210 words. One hundred and thirty five words were rated as familiar at least by 80% of the children. Additional words from Landau and Quitregard then were added. The researcher used three different procedures to ensure the familiarity of words chosen for the word lists. Monosyllabic words were matched between the word lists from Landau (1959) and Quitregard (1994). Words that occurred in both lists were considered for further familiarity testing. Second the matching words and Abdulhaqs list were rated as familiar or unfamiliar by four Jordanian teachers of first, second and third grade. Finally teachers ratings were compared to the childrens ratings for consistency. Out of these lists, the top 304 words rated as highly familiar by students and teachers were selected for recording.

PAGE 79

APPENDIX C LOGISTIC REGRESSION ANALYSIS Logistic regression analysis calculates the probability of a response based on the independent variable, in this study, the stimulus intensity level. The logistic regression analysis provides derived variables that are key parameters for the regression line that may be tested and compared between conditions (in this study, the word lists). The results from the logistic regression analysis, a, the logistic regression intercept and b, the logistic regression slope, were used to calculate the probability of correct identification of the words at each intensity level. Based on the model described by Nissen et al. (2005a) in the description of performance on the word list in terms of logistic regression, the following modified equation was used to calculate the percent correct: Equation 1 100))exp(1)exp(1( ibaibaP Where P is the percent correct at an intensity level, a is the regression intercept, b is the regression slope, and i is the intensity level in dB HL. Using the regression intercept, slope and intensity levels in equation 1 made it possible to predict the percent correct at any given intensity level. 67

PAGE 80

APPENDIX D LIST OF WORDS FOR DIALECT COMPARISON Word Meaning Word Meaning 1 Brother 18 Class 2 Door 19 Bird 3 Egg 20 Nest 4 Dress 21 Stick 5 Bull 22 Eye 6 Grandparent 23 Uncle (paternal) 7 Pilgirim 24 Mouth 8 Thread 25 Elephant 9 Bare 26 Cottage 10 Blood 27 Thief 11 Rooster 28 Board 12 Rice 29 Almonds 13 Feather 30 Banana 14 Market 31 Fier 15 Sword 32 Sleep 16 Tea 33 Hand 17 Thornes 68

PAGE 81

APPENDIX E HALF WORD LISTS AND MEANING 1A 2A 3A 4A Good God Meadow Sun Elbow Pilgrim Horses Crown Home Button Fairies Figs Head Wrap Situation Well Market Fire Answer Brick Almonds Tight Length Under Uncle (maternal) Gas Bank Opposite Cup Bird Whale Poisonous Cold Sleeve Wind Pocket Family Free Chick Cottage Bus Dam Eye Seeds Hunt Fear Dry Dog Death Stick Roof Skewer Husband Prosperity Spirit Count Blacks Poured (adj) Shelf Tub West Glamour Person Jug Dress Tea Tower Symbol Valley Shaikh Wilderness Difficult Geese China Girl Distance Snow House weather Lesson Elephant Turn Earth Bitter Something Hair Roses Guest Zaid (name) Heart Nose Thousand Beach Feather Right Saturday Voice Religion Love Ducks 69

PAGE 82

70 1B 2B 3B 4B Lettuce Requirement Egg Grandfather Serious Garlic Bull String Palm Sell Banana Shadow Half Behind Bed Solution Touch Armey Holiday Name Same Year War Throne Mother Outfit Turn over Sword Heads Pilgrimage Increase Lima bean Trumpet Traffic Evil Neighbor Board Secret Thief Axe Width Serious Father Gauze Thorns Day Spicy Way Hot Rice Dive Belly Berry Space Line Fence Check Hunger Light Third Oil Rural Arch Mouse Rooster Money Lamp Blood Turn Youngman People Sister Party Brother Bear Limit Boot Reem (name) Core Sleep friendliness Private Plum When Neighborhood Class Worry Color Monkey Nest East Gun Treasure Wool Shower Bag Gray hair Tide Uncle (paternal) Skin

PAGE 83

LIST OF REFERENCES Abd-el-Jawad, H. (1987). Cross-dialectal variation in Arabic: competing prestigious forms. Language in Society, 16; 359-368. Al-Kahtani, A. (1997). The Problem of Diglossia in the Arab world: an attitudinal study of modern standard Arabic and the Arabic dialects. Al-Arabiyya, 30;1-30. AlMasri, M. (2003). Early Detection and Habilitation of Hearing Loss, MEHA Project 4. Amman, Jordan. Altoma, S. (1969). The Problem of Diglossia in Arabic, a comparative study of classical and Iraqi Arabic. Harvard Middle Eastern Monographs XXI. Alusi, H.A., Hinchcliffe, R., Ingham, B., Knight, J.J., & North C. (1974). Arabic speech audiometry. Audiology, 13; 212-130. American National Standards Institute (2004). Specification for Audiometers. ANSI S3.6-2004. New York: ANSI. Ashoor, A.A. & Prochazka, T. Jr. (1982). Saudi Arabic speech audiometry. Audiology, 21; 493-508. Ashoor, A.A. & Prochazka, T. Jr. (1985). Saudi Arabic speech audiometry for children. British Journal of Audiology, 19; 229-238. Arlinger, S. (2001). How to assess outcome of hearing aid fitting in children. Scandinavian Audiology, 30 Suppl. 53; 68-72. Beattie, R.C., Edgerton, B.J., & Svihovec, D.V. (1977). A comparison of the Auditec of St. Louis Cassette recordings of NU-6 and CID W-22 on a normal-hearing population. Journal of Speech and Hearing Disorders, 42(1). 60-64. Beattie, R.C., Svihovec, D.V., & Edgerton, B.J. (1975). Relative intelligibility of the CID spondees as presented via monitored live voice. Journal of Speech and Hearing Disorders, 40(1); 84-91. Beatti, R.C. & Warren, V. (1983). Slope characteristics of CID W-22 word functions in elderly hearing impaired listeners. Journal of Speech and Hearing Disorder, 4;119-127. 71

PAGE 84

72 Blamey, P., Sarant, J., Paatsch, L., Barry, J., Bow, C., Wales, R., Wright, M., Psarros, C., Rattigan, K. & Tooher, R. (2001). Relationship among speech perception, production, language, hearing loss, and age in children with impaired hearing. Journal of Speech, Language, and Hearing Research, 44(2); 264-283. Carhart, R. (1951). Basic principles of speech audiometry. Acta Oto-Laryngologica, 40; 62-71 Carhart, R. (1965). Problems in the measurement of speech discrimination. Archives of Otolaryngology, 82; 253-260. Cramer, K. & Erber, N.P. (1974). A spondee recognition test for young hearing-impaired children. Journal of speech and hearing Disorders, 39; 304-311. Curry, E.T. (1949). A study of the relationship between speech thresholds and audiometric resluts in perception deafness. Journal of Speech and Hearing Disorders, 14; 104-110. Davis, H. (1948). The Articulation area and the social adequacy index for hearing. Laryngoscope, 58; 761-778. Doyne, M. P., & Steer M.D. (1951). Studies in speech reception testing. Journal of Speech and Hearing Disorders, 16; 132-139. Dubno, J.R., Lee, F.S., Klein, A.J. & Matthews, L.J. (1995). Confidence limits for maximum word-recognition score. Journal of Speech and Hearing Research, 38; 490-502. Egan, James (1948). Articulation testing methods. Laryngoscope, 58; 955-991. Elliot, L. (1979). Performance of children age 9 to 17 years on a test of speech intelligibility in noise using sentence material with controlled word predictability. Journal of the Acoustical Society of America, 66; 651 -653. Elliot, L, Connors S., Kille, E., Levin, S., Ball, K. & Katz, D. (1979). Children's understanding of monosyllabic nouns in quiet and in noise. Journal of the Acoustical Society of America, 66; 12-21. Elliot, L. & Katz, D.R. (1980). Northwestern University Childrens Perception of Speech (NU-CHIPS). AUDITEC of St. Louis, St. Louis, Missouri. Erber, N.P. (1974). Pure-tone thresholds and word-recognition abilities of hearing-impaired children. Journal of Speech and Hearing Research, 17; 194-202 Erber, N.P. (1980). Use of the auditory numbers test to evaluate speech perception abilities of hearing-impaired children. Journal of Speech and Hearing Disorders, 45; 527-532.

PAGE 85

73 Erber, N.P. & Alencewicz,C. (1976). Audiologic evaluation of deaf children. Journal of Speech and Hearing Disorders, 41; 256-267. Fatihi, A.R. (2001). Aspects of Arabic Phonology. Kalinga Publication, Delhi. Ferguson, C. (1959). Diglossia. Word, 15; 325-340 Fletcher, H. (1958). Speech and Hearing in Communication. D. van Nostrand Company, Inc., Princeton, New Jersy. Fletcher, H. (1995). The ASA edition of speech and hearing in communication. Ed. Allen, J.B. Acoustical Society of America, Woodbury, NY. Fletcher, H. & Steinberg, J.C. (1929). Articulation testing methods. Bell System Technical Journal, 8; 806-854. Goldman,R., Fristoe, M. & Woodcock, R. (1970). Goldman-Fristoe-Woodcock Test of Auditory Discrimination. American Guidance Service, Inc. USA. Harris, R., Kim, E. & Eggett, D. (2003a). Psychometrically equivalent Korean monosyllabic speech discrimination materials spoken by male and female talkers. Korean Journal of Communication Disorders, 8 (1); 217-243. Harris, R., Kim, E. & Eggett, D. (2003b). Psychometrically equivalent Korean bisyllabic words spoken by male and female talkers. Korean Journal of Communication Disorders, 8 (1); 244-270. Haskins, H.A. (1949). A phonetically Balanced Test of Speech Discrimination for Children. Masters thesis. Northwestern University, Evanston IL. Hirsh, I.J., Davis, H., Silverman, R., Reynolds, E.G., Eldert,E., & Benson, R. (1952). Development of materials for speech audiometry. Journal of Speech and Hearing Disorder, 17;321-337. Hirsh, I. J., Reynolds, E.G., Joseph, M. (1954). Intelligibility of different speech materials. The Journal of the Acoustical Society of America, 26(4); 530-538. Hood, J.D. & Poole, J.P. (1980). Influence of the speaker and other factors affecting speech intelligibility. Audiology, 19; 434-455. Ibrahim, Z. (2000). Myths about Arabic revisited. Al-Arabiyya, 33; 13-28. The International Kindergarten Union, Child Study Committee (1928). A Study of Vocabulary of Children Before Entering the First Grade. Washington, D.C. Jauhiainen, T. (2001). Hearing impaired children in developing countries. Paper presented at the 3 rd International Congress of Pediatric Audiology, 9-12 May 2000, Copenhagen. Scandinavian Audiology, 30 Suppl 53; 83-84.

PAGE 86

74 Jerger, S., Lewis, S., Hawkins, J., & Jerger, J. (1980). Pediatric speech intelligibility test. I. Generation of test materials. International Journal of Pediatric Otorhynolaryngology, 2; 217-230. Kishon-Robin, L. & Rosenhouse, J. (2000). Speech perceptin test for Arabic-speaking children. Audiology, 39; 269-277. Kreul, E.J., Bell, D.W. & Nixon, J.C. (1969). Factors affecting speech discrimination test difficulty. Journal of Speech and Hearing Research, 12; 281-287. Landau, J. (1959). A Word Count of Modern Arabic Prose. New York: American Council of Learned Societies in cooperation with the school of education at the Hebrew University, Jerusalem. Larson, G.W., Petersen, B., & Jacquot, W.S. (1974). Use of Northwestern University Test No.6 for speech discrimination testing with children. The Journal of Auditory Research, 14; 287-292. Laukli, E. & Fjermedal, O. (1990). Reproducibility of Hearing Threshold Measurements, supplementary Data on Bone-conduction and speech audiometry. Scandinavian Audiology, 19; 187-190. Martin, F.N., Champlin, C.A., & Perez, D.D. (2000). The question of phonetic balance in word recognition testing. Journal of the American Academy of Audiology, 11; 489-493. Messouak, K (1956). Audiometrie vocale en Arabic maghrebin. Les cahiers de la CFA, No. 4. Compagnie Francaise dAudiologie, Paris. Meyer, T.A & Pisoni, D.B (1999). Some computational analyses of the PBK test: effect of frequency and lexical density on spoken word recognition. Ear and Hearing, 20 (4); 363-371. Niessen, S.L, Harris, R.W, Jennings, L.J., Eggett, D.L., & Buck, H. (2005a). Psychometrically equivalent Mandarin bisyllabic speech discrimination materials spoken by male and female talkers. International Journal of Audiology, 44(7); 379-390. Niessen, S.L, Harris, R.W, Jennings, L.J., Eggett, D.L., & Buck, H. (2005b). Psychometrically equivalent trisyllabic words for speech reception threshold testing in Mandarin. International Journal of Audiology, 44(7); 391-399. ster, A.M. (2002). The relationship between residual hearing and speech intelligibilityIs there a measure that could predict a prelingually profoundly deaf childs ability to develop intelligible speech? Speech, Music and Hearing, Quarterly Progress and Status Report (TMH-QPSR), 43; 5156.

PAGE 87

75 Owens, E. (1961). Intelligibility of words varying in familiarity. Journal of Speech and Hearing Research, 4(2); 113-129. Palmer, J. M. (1955). The effect of speaker differences on the intelligibility of phonetically balanced word lists. Journal of Speech and Hearing Disorders, 20; 192-195. Quitregard, D. (1994). Arabic Key Words, the basic 2000 word vocabulary arranged by frequency in a hundred units. The Oleander Press, New York. Ross, M. & Lerman, J. (1971). A picture identification test for hearing-impaired children. Journal of Speech and Hearing Research, 13; 44-53. Sanderson-Leepa, M. & Rintelmann, W.F. (1976). Articulation functions and test-retest performance of normal-hearing children on three speech discrimination tests: WIPI, PBK-50, and NU Auditory test No. 6. Journal of Speech and Hearing Disorders, 41(4); 503-519. Saudi Speech Pathology and Audiology Association, SSPAA (2004). http://www.sspaa.org last accessed October 2004. Schwartz, A & Goldman, R. (1974). Variables in fluencing performance on speech-sound discrimination tests. Journal of Speech and Hearing Research, 17; 25-32. Siegenthaler, B.M, Pearson, J. & Lezak, R.J. (1954). A speech reception threshold test for children. Journal of Speech and Hearing Disorder, 19; 360-366. Smith, A.W. (2001). WHO activities for prevention of deafness and hearing impairment in children. Paper presented at the 3 rd International Congress of Pediatric Audiology, 9-12 May 2000, Copenhagen. Scandinavian Audiology, 30 Suppl 53; 93-100. Soliman S (1976). Speech discrimination audiometry using Arabic phonetically balanced words. Ain Shams Medical Journal, 27; 27-30. Soliman, S., Abd El-Hady, M., Saad, M., & Kolkaila, E. (1987a). Development and study of Arabic word intelligibility by picture identification (WIPI) in normal hearing Egyptian Children. Scientific Journal of Al-Azhar University, Medical Faculty (girls), 6. Soliman, S., Fathalaa, A., & El-Mahalawi, T. (1987b). Simple speech test as a predictor of speech reception threshold (SRT) in preschool children. Proceedings of Annals of the Xth Annual Medical Congress of Aid Shams University, Cairo, Egypt. Tillman, T.W., & Carhart, R. (1966). An expanded test for speech discrimination utilizing CNC monosyllabic words (Northwestern University Auditory Test No.6). Technical Report No. SAM-TR-66-55, USAF School of Aerospace Medicine, Brooks Air Force Base, Texas.

PAGE 88

76 Thornton, A. & Raffin, M. (1978). Speech discrimination scores modeled as a binominal variable. Journal of Speech and Hearing Research 21; 507-518. Wilson, R.H. & Carter, A.S. (2001). Relation between slopes of word recognition psychometric functions and homogeneity of the stimulus materials. Journal of the American Academy of Audiology, 12; 7-14. Wilson, R.H. & Oyler, A.L. (1997). Psychometric function for the CID-W22 and NU Auditory Test No.6. materials spoken by the same speaker. Ear and Hearing, 18(5); 430-433. Wilson, R.H. & Strouse, A. (1999). Psychometrically equivalent spondaic words spoken by a female speaker. Journal of Speech, Language, and Hearing Research, 42; 1336-1346. Wilson, R.H., Zizz, C., Shanks, J., & Causey, G. (1990). Normative data in quiet, narrowband noise, and competing message for Northwestern university auditory test No.6 by a female speaker. Journal of Speech and Hearing Disorders, 55; 771-778. World Health Organization (1998). Report of the International Workshop on Primary Care and Hearing Care. Cape Town; South Africa March 12-14. Young, J.R., Dudley, B., & Gunter, M.B. (1982). Thresholds and psychometric functions of the individual spondaic words. Journal of Speech and Hearing Research, 25; 586-593. Zakrzewski, A, Jassem, W., Pruszewicz, A., & Obrebowski, A. (1976). Speech audiometry for children and subjective probability of Polish words. Audiology, 15; 228-231.

PAGE 89

BIOGRAPHICAL SKETCH Nadia Abdulhaq was born and raised in Nablus, Palestine. The third of six children, she graduated from high school in 1990 and enrolled at Birzeit University in Birzeit, Palestine. She obtained her Bachelor of Arts degree in English language and literature with a minor in translation, in June, 1995. She then went on to receive her Master of Arts degree in speech-language pathology from the University of Jordan in Amman, Jordan, in June, 1997. After graduating from the University of Jordan, she accepted the position of speech pathologist in the Department of Rehabilitation at the Palestine Red Crescent Society. Her job description included establishing three speech therapy units in three different cities across the West Bank of Palestine and training six rehabilitation workers as speech therapy assistants. Her work focused on children with hearing loss. After two years of work and achievements, she decided to return to the academic world and pursue further knowledge to be able to better serve the hard of hearing and deaf community of Palestine. In 2000 she received the Fulbright scholarship to start her doctoral studies at the University of Florida in Gainesville, Florida. During her five year study she was an active member of the international student body initiating projects such as the International Student Speakers Bureau and the international student welcome information table. She received numerous scholarships and recognitions, O. Ruth McQown Scholarship, Grinter Fellowship, Gibbson Dissertation Fellowship, and the Outstanding Academic 77

PAGE 90

78 Achievement award. She will graduate in December 2005 with a Ph.D. in communication sciences and disorders and Doctor of Audiology degrees. Upon her graduation, she will return to Palestine to start a speech and hearing program for the hard of hearing, new born hearing screening program, and an audiology clinic.


Permanent Link: http://ufdc.ufl.edu/UFE0013122/00001

Material Information

Title: Speech Perception Test for Jordanian Arabic-Speaking Children
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013122:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013122/00001

Material Information

Title: Speech Perception Test for Jordanian Arabic-Speaking Children
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013122:00001


This item has the following downloads:


Full Text












SPEECH PERCEPTION TEST FOR JORDANIAN
ARABIC SPEAKING CHILDREN















By

NADIA MOHAMED ABBAS ABDULHAQ


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006

































Copyright 2006

by

NADIA MOHAMED ABBAS ABDULHAQ

































This dissertation is dedicated to my parents Abbas and Ursula.















ACKNOWLEDGMENTS

I would like to thank my parents for their constant support and believe in me. They

have been a bottomless source of support, inspiration and encouragement. My sisters Ute

Nabila, Najat, Ulla Najwa, and brothers Mejdi and Uwe Maysarah, who where ever in the

world they were made sure to keep in touch and encourage me through the tough times

and share the joy of my success.

My friend from Palestine, Lubna Shaheen, for all the times she has listened to my

complaints and doubts and never let me go without giving me my confidence back. My

good friend, Abdelqader Abu Awad, for his support and encouragement from the time I

applied to the Fulbright program to the day of my graduation he has always been there for

me.

There is a group of people here in Gainesville without whom I would have felt very

lonely. I would like to thank all, Debra Anderson, International student coordinator, for

being a shoulder to lean on, a good listener and a friend in times of need. She has been

my inspiration in many activities on campus that have brought diversity to my experience

and made it so much richer in people, memories, and achievements. My friends in

Gainesville, Abdullhatif Qamihieh for his readiness to help at any time what ever the task

was. Maisa Haj-Tas for being a wonderful understanding roommate and her stimulating

discussions and insight into research. Marah Al-Aloul, Ashraf Al-Qdah, Suha Abu Lawi,

Shadi Krecht, Yaser Katanani, Enas Katanani and Hadil Faqih, for their support and for

being my family away from home through out these past years, and my friends from the









Fulbright program and international students for being a renewing source of hope,

inspiration and fun.

This project was made possible by the continuous support by the faculty members

of the Departments of Communication Sciences and Disorders and Communicative

Disorders at the University of Florida. I would like to thank in particular my first mentor

Dr. Patricia Kricos for her guidance, support and generosity. Dr. Griffiths for his advise

through the graduate program in general and mentoring my dissertation in particular, Dr.

James Hall III, for his guidance, support and encouragement, Dr. Richard Harris for his

hospitality, generosity and guidance through the research process, and Dr. Tom Oakland,

for his honest opinions, believes in me and support. I would also like to thank Dr. Linda

Lombardino, Dr. Alice Dyson, Dr. Keneth Gerhardt, Dr. Mary Kay Dykes, and Dr. Aida

Bamia for their support.

Funding for this project was made possible by the Gibson Dissertation Fellowship,

offered by the Collage of Liberal Arts and Sciences at the University of Florida. The data

collection and field research was made possible by the generous help and collaboration of

Samar Al-Aghbar, the Jordanian ministry of education, and elementary schools of

Amman, as well as the support of the Middle East Hearing Association and their team

members, and Phonak hearing aid distribution office in Amman.

Completion of this degree was made possible by financial support from the

department of Communication Sciences and Disorders in the form of teaching

assistantship and Grinter Fellowship award for four years, as well as the College of

Liberal Arts and Sciences O. Ruth McQuown Scholarship for women, in addition to the









Fulbright scholarship that was the seed funding for this degree, and finally and most

importantly, my parents financial support.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

L IST O F TA B L E S ................. .......................................... .... .... ...... ...... ....... ix

LIST OF FIGURES ............................... ... ...... ... ................. .x

ABSTRACT .............. .................. .......... .............. xi

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

B background of the Study .................................................... .............................. 1
Rationale and Purpose ........................ ......... ... ..... ...... ....................
Hearing Loss in Developing Countries ...................................... ............... 3
H hearing Loss in Jordan ............................................................ ............4
Present Condition of Arabic Speech Audiometry ..............................................5
R e search Q u e stio n s.......................................................................................... .. 7
Experim ent O ne .................................................. ........ .. ........ .. 7
Experim ent Tw o ........................ .. .............. .......................... ....
H y p o th e sis ........................................................... ................ 7
Experim ent O ne .................................................. ........ .. ........ .. 7
Experim ent Tw o ........................ .. .............. .......................... ....

2 REVIEW OF THE LITERATURE .................................... .......................... .......... 8

Importance of Speech Audiometry ......... ............ ......... ................. 8
Speech A udiom etric Tools ............... ................. ................................................
Pediatric Speech Perception M materials .............................. ..... ............. 13
Adult Speech Perception M materials ........................................ ............... 14
Speech R exception Threshold M material ............................ .......... .................... 15
Full List and Half List U se ................................. .................................. 16
Arabic Speech Audiometry................. ................................. 18
Comparison of Arabic Speech Reception Tests ..................... ............... 18
Use of Arabic Speech Reception Tests .................... ........ ............... 20
D ialectal D differences in A rabic ............................................... ........ ............... 21
Special Considerations for the Present Study .................................23
P articip ants ........................................................................2 3









S p eak ers .................. ...................................................................................... 2 4
Stim uli for D ialectal D ifferences..................................... ......... ..............25

3 M E T H O D O L O G Y ............................................................................ ................... 29

P a rtic ip a n ts ........................................................................................................... 2 9
S p e a k e rs .................................................................................................... 3 0
Recording and Editing .................. ..................................... ...............3 1
Instrumentation for Data Collection ............................... .....................33
S tu d y O n e ............................................................................3 4
Speech M material ................. .... ...................... ......... 34
Procedure for Data Collection ................. .......... ....................35
Statistical A analysis .......................... .......... ................ ....... 36
Sp eech M material .............................. .......................... .... ........ .... ..... ...... 37
Statistical A analysis .......................... .......... ................ ....... 37

4 R E S U L T S .............................................................................4 1

S tu d y O n e .............................................................................4 1
S tu d y T w o ............................................................................................................. 4 3

5 DISCUSSION .......... .. .............. ...................57

APPENDIX

A PARTICIPANT RECRUITMENT ....... ..................................... ........... 64

B SPEECH MATERIAL SELECTION ........................................................... 65

C LOGISTIC REGRESSION ANALYSIS ......................................67

D LIST OF WORDS FOR DIALECT COMPARISON .................... ..................68

E HALF WORD LISTS AND MEANING ...........................................69

LIST OF REFEREN CES ................................................................................... 71

B IO G R A PH IC A L SK E T C H ....................................................................................... 77
















LIST OF TABLES


Table page

2-1 Children's speech reception tests in English ........................................................27

2-2 Speech recognition word lists' mean dB HL levels at 50% and slopes .................27

2-3 Speech reception threshold wordlists' mean threshold dB HL levels and slopes
betw een 20 and 80% ......... .................................................. ....... ........28

3-1 Order of list presentation by participant and level of presentation ........................39

3-2 W ord list and dialect presentation order. ..................................... ............... 40

4-1 Normal Hearing participants' age, gender, test ear and PTA.................................50

4-2 Half word lists and transcription in IPA....................................... ...................... 51

4-3 Logistic regression results for the full lists, the calculated slopes at the 50% and
the 20-80% levels, the threshold dB level, and difference of threshold levels
from the m ean threshold................................................. .............................. 53

4-4 Logistic regression results for the half lists, the calculated slopes at the 50% and
the 20-80% levels, the threshold dB level, and difference of threshold levels
from the m ean threshold................................................. .............................. 53

4-5 Selected characteristics of normal hearing participants .......................................54

4-6 Selected characteristics of hard of hearing participants................................54

4-7 Word recognition scores in percent correct for each subject per talker-talker and
the average of scores by dialect.................................................................... 55

4-8 Mixed ANOVA results comparing dialects for all participants' scores ...................55

4-9 Paired Sample T-tests comparing dialectal difference in the hard of hearing
group and norm al hearing group. ........................................ ........................ 55

4-10 Paired Sample T-test results including all participants' scores.............................56

5-1 Descriptive statistics for Hard of Hearing and Normal Hearing based on dialect ...63
















LIST OF FIGURES


Figure page

4-1 Average pure tone thresholds at all frequencies for all 20 normal hearing
participants at 250, 500, 1000, 2000, 4000, and 8000 Hz ......................................46

4-2 H alf lists raw data scatter plot ............................................................................... 46

4-3 Psychometric function of 4 lists of 50 words based on calculated percent correct..46

4-4 Psychometric function of 8 half lists of 25 words based on calculated percent
c o rre ct......... ........... .. ........... .. .................................................... 4 7

4-5 Arabic monosyllabic psychometric functions for lists 1-4 (left) and half-lists 1A-
4B (right). .............................................................................48

4-6 Average of pure tone thresholds of normal hearing participants at 250, 500,
1000, 2000, 4000, and 8000 H z. ........................................ .......................... 49

4-7 Average of pure tone thresholds of hard of hearing participants at 250, 500,
1000, 2000, 4000, and 8000 H z. ........................................ .......................... 49















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

SPEECH PERCEPTION TEST FOR JORDANIAN
ARABIC SPEAKING CHILDREN

By

Nadia Mohamed Abbas Abdulhaq

May 2006

Chair: Scott Griffiths
Major Department: Communication Sciences and Disorders

Comprehensive audiologic evaluation includes a variety of tests that provide a

determination of the type of hearing loss. Among these tests are tests of word recognition.

Many speech perception tests have been developed over the past half century to assess

different aspects of speech. Management of hearing loss in developing countries and the

USA differs significantly. The prevalence and demographics of the hard of hearing

population is also different, and thus imposes different needs. According to the World

Health Organization the number of deaf and hard of hearing individuals in developing

countries is twice as much as in developed countries. There is a need for early

identification and intervention for hearing loss in developing countries. At the same time

there is a severe lack of equipment and highly trained professionals to provide such

services. The focus of this study is speech audiometric measure for Jordanian Arabic

speaking children. The first goal of this study is to develop four Jordanian Arabic 50-

word lists appropriate to use for word recognition measure for Jordanian children age 6 to









9 years. The second goal is to investigate the effect of using words recorded in Saudi

dialect on the word recognition abilities of Jordanian normal hearing and hard of hearing

children.

Twenty individuals age 6 through 9 years participated in this study. A Jordanian

speaker recorded 250 Arabic words selected for familiarity to elementary aged Jordanian

children. The raw score data of all participants at the 10 intensity levels were compiled

for each of the 250 words. Four equally difficult lists of fifty words each were derived

from this experiment. In the second experiment, 3 Jordanian and 3 Saudi male speakers

were selected to record 33 words. These were played back at a constant comfortable level

for 10 children with normal hearing and 10 children with hearing impairment who were

asked to identify the recorded words. Dialect produced a significant difference in

performance for children with normal hearing, but not for children with hearing

impairment. These results will lead to the development of improved techniques for

assessing auditory performance in Arabic-speaking children.














CHAPTER 1
INTRODUCTION


Background of the Study

An audiological evaluation typically includes measures of tympanometry,

acoustic reflexes, otoacoustic emission, and pure-tone (air and bone conduction)

threshold and speech audiometry. Results of these tests are used to diagnose hearing loss

and determine the type of hearing loss. Hearing testing has come a long way from

estimating hearing from the distance a person can hear a voice. Since the early 1900s,

pure-tone audiometry has been instrumental in measuring hearing sensitivity. The work

of Harvey Fletcher on the perception of speech in relation to sound level and noise effect

laid the foundation for decades of speech perception research (Fletcher, 1995). Speech

perception measures in the present format in the United States of America (USA) have

been used since the late 1940s and early 1950s (Egan, 1948; Haskins, 1949; Hirsh, Davis,

Silverman, Reynolds, Eldert, & Benson, 1952). Many speech perception tests have been

developed and improved over the past half century to assess different aspects of speech;

e.g., speech reception thresholds, speech pattern identification, and speech reception in

noise (Elliot & Katz, 1980; Erber, 1974; Goldman, Fristoe, & Woodcock, 1970; Jerger,

Lewis, Hawkins, & Jerger, 1980; Ross & Lerman, 1970; Tillman & Carhart, 1966).

Speech reception tests that mostly are used clinically include the Central Institute

for the Deaf W-22 (CID W-22) by Hirsh et al. (1952), Northwestern University Auditory

Test No. 6 (NU-6) by Tillman & Carhart (1966), Phonetically Balanced Kindergarten









Test (PBK-50) by Haskins (1949), and Northwestern University Children's Perception of

Speech (NU-CHIPS) by Elliot & Katz (1980). Common characteristics among these tests

are that they are based on monosyllabic frequently used words which are familiar to the

target population (at least at the time of test development) and were developed based on

measures of correct responses as a function of intensity of presentation. The articulation

function is a common representation of speech audibility result, which forms an ogive or

S-shaped curve and indicates the degree a person's hearing ability improves given

increases in the intensity of the material presented (Carhart, 1951). The shape of the

ogive and the 50% point (threshold point) are affected by the speech material presented

as well as the speaker. That is, different speech materials produce different ogives. If

different speakers recorded the same speech material, the result would yield different

ogives (Beattie, Edgerton, & Svihovec, 1977; Beattie, Svihovec, &Edgerton, 1975;

Carhart, 1965; Doyne & Steer, 1951; Hirsh, Reynolds, & Joseph, 1954; Kruel, Bell, &

Nixon, 1969; Wilson & Carter, 2001; Wilson & Oyler, 1997). Results from adult speech

recognition test (NU-6) and children's tests (PBK-50 and WIPI) were compared

(Sanderson-Leepa & Rintelman, 1976). The adult test was found to be inappropriate for

use with young children age 3.5 to 9.5 years while WIPI was most appropriate for young

children ages 3.5 to 5.5, and NU-6 was appropriate but more difficult than the PBK-50

for children age 7.5 to 11.5 the. That is, speech material needs to be age appropriate, and

words need to be familiar to the target group. Carhart (1965), and Kreul et al. (1969)

recommend the use of a limited set of recorded materials to establish normative measures

because different recordings by the same speaker can result in different articulation

functions.









Studies of speech perception in other languages (e.g., Arabic, Polish, Korean, and

Chinese) have followed Egen's (1948) and Hirsh et al. (1952) lead in developing speech

reception material (Alusi, Hinchcliffe, Ingham, Knight, & North, 1974; Zakrzewski,

Jassem, Pruszewicz, & Obrebowski, 1976; Ashoor & Prochazka, 1982; Ashoor &

Prochazka 1985; Harris, Kim, & Eggett, 2003a; Harris, Kim, & Eggett, 2003b; Nissen,

Harris, Jennings, Eggett, & Buck, 2005a; Nissen, Harris, Jennings, Eggett, & Buck,

2005b). Special attention was paid to the familiarity of words, the equivalence of word

lists, and in some cases the phonetic balance of word lists. The resulting lists have had

similar articulation functions to the English word lists in shape and slope. Thresholds of

speech reception and word recognition were more varied across languages.

Rationale and Purpose

Hearing Loss in Developing Countries

The management of hearing loss in developing countries and the USA differs

significantly. The prevalence and demographics of the hard of hearing population also

differs and thus imposes different needs for services. According to the World Health

Organization (WHO), the number of deaf and hard of hearing individuals in developing

countries is twice that in developed countries (Smith, 2001). According to the American

Academy of Audiology an average of 3 in every 1,000 healthy newborns in the United

States has severe sensorineural hearing loss. In Jordan 6 in every 1,000 healthy newborns

have hearing loss (Al-Masri, 2003). For comparison purposes, the WHO reported that 4

to 5 children under the age of 18 in every 1,000 children have sensorineural hearing loss

in the South-East-Asian Region (Smith, 2001). This number is reported to be inaccurate

due to the lack of epidemiology surveys in developing countries; the actual numbers are

projected to be double what has been reported (Smith, 2001). Globally, the majority of









children are living in developing counties, which indicates that the majority of children

with hearing loss live in those countries as well (Jauhiainen, 2001).

The WHO identified the major causes of hearing loss to be chronic otitis media,

genetics, maternal and perinatal problems, and ototoxicity (Smith, 2001). The rates of

complications due to otitis media are 1/1000 in developed countries and 60/1000 in

developing countries. The incidence of deaths due to ear infections complications are

1/100,000 in developed countries and 1 /100 in developing countries (WHO, 1998). In

spite of the higher prevalence of hearing loss in developing countries, services and

technology generally are limited or lacking, especially in rural areas (WHO, 1998;

Jauhiainen, 2001). The prevalence of Ear, Nose and Throat doctors ranges from 1/

30,000 to 1/150,000 in developed countries and 1/2,000,000 in the less developed

countries in Africa (excluding South Africa and Egypt) (WHO, 1998).

Early identification and intervention for hearing loss in developing countries are

need. At the same time there is a severe lack of equipment and highly trained

professionals to provide such services. In spite of this general statement, services in some

developing counties are more advanced than others, and some individuals have the means

to afford world class services (Jauhiainen, 2001). The focus of this study is on audiology

in Jordan, specifically speech audiometric measure for Jordanian Arabic speaking

children.

Hearing Loss in Jordan

Pilot data from screenings conducted by the Middle East Hearing Association

(MEHA) suggest estimates of the hard of hearing and deaf population in Jordan to be

64,000, with 2,200 infants with severe hearing loss born yearly (Al-Masri, 2003). Some

institutions provide limited services, including the Speech and Hearing Clinic at the









University of Jordan, Al-Ahliyya Amman University, Holyland Institute for the Deaf,

King Hussein Medical Center, and some private otolaryngology and audiology clinics

and hearing aid providers. However, diagnostic and rehabilitative services and

professional training are seriously lacking. There are initiatives by the Jordanian

government, the royal family in particular, to develop such services. Under the patronage

of His Royal Highness, Prince Firas Raad, MEHA was established in 1998 in cooperation

with the Canadian International Scientific Exchange Program, and a new center was

opened recently to provide services to the hearing impaired and deaf population. Projects

being implemented through this organization include newborn hearing screening, genetic

hearing loss research, audiologic evaluation, hearing aid fitting, audiologic rehabilitation

and follow up for children. Equipment needed for such services is available and some

basic audiometric measures are being performed. An important part of evaluation, speech

perception audiometry, is missing from the diagnostic battery. Speech perception

materials suitable for testing Jordanian Arabic speaking children are unavailable. The

purpose of this study is to develop a speech perception test for Jordanian Arabic speaking

children.

Present Condition of Arabic Speech Audiometry

Carhart (1951) emphasized the importance of using familiar words that are in the

listener's native language. Tests developed in Arabic have been recorded using Maroccan

(Messouak, 1956), Iraqi (Alusi, 1974), Egyptian (Soliman, 1976; Soliman, Abd El-Hady,

Saad, & Kolkaila, 1987; and Soliman, Fathallah, & El-Mahalawi, 1987), and Saudi

(Ashoor et al., 1982; and Ashoor et al. 1985) dialects. The question arises as to whether

the different dialects have an effect on speech reception. Although all Arabic countries

share one standard Arabic language that is taught formally in school in the form of the









written language (Altoma, 1969; Al-Kahtani, 1997; Fatihi, 2001; Ferguson, 1956)

dialectal differences exist. Studies conducted to measure the dialectal effect on speech

reception in Arabic could not be located. Thus, one may question the effect of the dialect

related to the words used in that dialect as well as the articulation and voice

characteristics of the speaker.

Recordings by different speakers result in different word recognition scores, and

the use of different speech material has the same effect (Beattie et al., 1977; Beattie et al.,

1975; Carhart, 1965; Doyne & Steer, 1951; Hirsh et al., 1954; Kruel et al., 1969; Wilson

& Carter, 2001; Wilson & Oyler, 1997). This suggests a need to develop speech

audiometric material based on one recording to ensure reliable results. Recordings

reported in other studies (e.g. Alusi et al., 1974; Ashoor & Prochazka, 1985) have not

been available for wide use. One intended outcome of this study is to make available a

high quality digital recording of speech material appropriate for use in testing word

recognition loss of Jordanian children. Audiologic testing in Jordan currently is based on

non-speech related audiometry, including pure-tone audiometry, otoacoustic emissions,

and auditory evoked potentials.

This study has two main goals. One is to develop four Jordanian Arabic 50-word

lists appropriate to use in word recognition measures for Jordanian children ages 6

through 9. Another goal is to investigate the effect of using words recorded in Saudi

dialect on word recognition abilities of Jordanian children who display normal hearing

and hearing loss.









Research Questions

Experiment One

1. Is there a significant difference in word recognition abilities of Jordanian Arabic
speaking children, given increases in intensity of presentation?

2. Can four parallel word lists be developed in Jordanian Arabic language (e.g., their
psychometric qualities of the word lists do not differ significantly)?


Experiment Two

1. Do word recognition abilities of Jordanian Arabic speaking children differ when
listening to words presented in Jordanian versus Saudi dialects?

2. Do word recognition abilities of Jordanian Arabic speaking children who display
normal and hearing disabilities differ when hearing words presented in a Jordanian
versus Saudi dialect?

3. Does word recognition ability differ when hearing speakers using the same dialect?


Hypothesis


Experiment One

1. An increase in presentation level will increase word recognition scores.

2. Four lists of 50-words with similar characteristics can be created to be used in
measures of word recognition abilities.


Experiment Two

1. Jordanian children will produce higher word recognition scores when listening to
words presented in Jordanian dialect than when listening to words presented in Saudi
dialect.

2. Jordanian and Saudi dialects will have the same effect on word recognition scores of
Jordanian children with normal hearing and with hearing loss.

3. Word recognition will differ for two talkers with the same Arabic dialect.














CHAPTER 2
REVIEW OF THE LITERATURE


Importance of Speech Audiometry

Non-speech audiometric procedures provide valuable information needed for

description of hearing loss and estimation of amplification benefit. These tests, however,

cannot measure the effect of hearing loss on speech (Carhart, 1951; Cramer & Erber

1974; Davis, 1948; Doyne & Steer, 1951; Erber, 1974; Hirsh, et al. 1952).

Speech audiometry requires a language related test material, and thus may be

influenced by the phonetic, melodic, and intonational differences between languages

(Carhart, 1951). Different languages require speech tests that consider the features unique

to the language. Distinctive features of languages can result in different auditory

requirements and affect the auditory capacities evaluated. To provide comprehensive

audiologic services, it is necessary to use speech audiometry for measurement of hearing

and the outcome of management.

Speech perception is important in facilitating cognitive development and normal

language acquisition (Cramer & Erber 1974). Hearing loss is an obstacle of language and

speech acquisition. Pure-tone thresholds provide information about detection of sound at

specific frequencies, yet provide little information about the perception of complex

signals such as speech. Knowledge of a person's ability to perceive speech can provide

information as to the extent hard of hearing children are able to communicate effectively

and, in addition, how they are likely to learn language. Speech audiometry is a clinical









approach in which well-defined speech samples are presented using a calibrated system

to measure an important aspect of hearing ability. Measurement of speech perception

different intensities results in an articulation function taking the shape of an ogive or S-

shaped curve, that indicates the degree a person's hearing ability improves with an

increase of the intensity of the material presented (Carhart, 1951).

Over 50 years ago, Carhart (1951) reported on the importance of speech

audiometry in hearing assessment. He considered it to be the most useful contribution to

hearing testing after the introduction of pure-tone audiometry. Carhart emphasized the

value of well-defined speech audiometry to provide finer classification of hearing loss,

and its importance in providing measures for educational and rehabilitation purposes.

Another early hearing researcher, Hallowell Davis (1948), acknowledged that there is no

simple interpretation of the pure-tone audiogram to express the patient's ability to hear

speech. It is important to measure the patient's ability to discriminate speech just as it is

important to measure the hearing loss in decibels.

Fletcher noted, "The process of speaking and hearing are very intimately related, so

much so that I have often said that, we speak with our ears. We can listen without

speaking but we can not speak without listening" (p.Al Fletcher, 1995). Speech

perception and language acquisition are two closely related processes. Hearing loss

usually disrupts the process of speech perception and thus delays language acquisition.

Therefore, Cramer & Erber (1974) emphasized the need for an accurate and valid

measure of speech perception for hearing impaired children. Arlinger (2001) emphasized

the importance of using speech recognition to measure hearing aid benefit in children,

and considered periodic evaluation of hearing aid benefit of great importance for children









to monitor their language and speech development. Blamey, Sarant, Paatsch, Barry, Bow,

Wales, Wright, Psarros, Rattigan, & Tooher (2001) used a measure of speech perception

to evaluate benefit of amplification, and the relationship between speech perception and

speech production, language, hearing loss and age in 87 children age 4 to 12 years.

According to this study, the authors expected the language delay to be 4 to 5 years by the

time children are 12 years old. Speech perception scores are expected to improve

significantly with the improvement of language; specifically they expect children to score

90% on the open set Bench-Kowal-Bamford (BKB) sentence test when they reach the

level of language proficiency of a 7 year old. Oster (2002) examined the relation between

audiological measures and speech intelligibility for eleven profoundly deaf Swedish

teenagers (age 15 to 17 years). Their pure-tone averages (PTA) ranged from 90 to 108 dB

HL. Correlation analysis was assessed for the intelligibility of the children's speech and

their pure-tone average, the shape of the audiogram, and residual hearing use. Results

showed that there was great variation in their speech intelligibility in spite of the narrow

range of PTA, indicating that the speech intelligibility cannot be estimated based on PTA.

The correlation between speech recognition scores and speech intelligibility scores

resulted in a positive correlation of 0.73, confirming a high correlation between residual

hearing use and speech intelligibility. Oster (2002) concluded that a simple speech test

can be used as a predictor of prelingually deaf children's ability of developing intelligible

speech. Laukil & Fjermedal (1990) researched the reproducibility of bone conduction

thresholds and the speech recognition thresholds. The results show low variability and no

significant difference between the two measures making the speech recognition threshold









measure a reliable one for Norwegian Spondees. These studies indicate that speech

perception measures are very important and reliable clinical tools.

Speech Audiometric Tools

Speech audiometry is an important part of an audiologic evaluation battery

(Arlinger 2001; Carhart 1951; Cramer & Erber, 1974; Curry 1949; Davis 1948; Erber,

1974). There are two essential measures of speech audiometry: the measure of speech

recognition threshold (SRT) and the measure of speech recognition at suprathreshold

levels (Carhart 1951; Davis, 1948). Davis (1948) has identified these measures as two

dimensions of hearing loss; one is hearing loss (dB level) and the second is

discrimination loss (word recognition score). Discrimination loss relates to the loss of the

ability to recognize words even when they are made audible. Several tests have been

developed to measure these two dimensions of hearing loss. The most frequently used

ones for speech perception are speech recognition threshold (SRT) test and word

recognition scores (WRS). An example of the SRT test is the Central Institute for the

Deaf spondee word test (CID W-2) by Hirsh et al. (1952). Many word recognition

materials lists have been developed over the years including the Phonetically Balanced 50

word lists (PB-50) described by Egan (1948), the CID W-22 (Hirsh et al., 1952), the

Northwestern University Test No.6 (NU-6) by Wilson and Oyler (1997), the PB

Kindergarten word test (PBK) by Haskins (1949), the Word Intelligibility for Picture

Identification Test (WIPI) by Ross & Lerman (1970), and the Northwestern University

Children's Perception of Speech (NU-CHIPS) by Elliot & Katz (1980). Egan (1948)

specified the criteria of selecting word lists for the word recognition tests as follows:

1. Monosyllabic structured words
2. Equal average of difficulty between lists
3. Equal range of difficulty within lists









4. Equal phonetic composition between lists
5. A composition representative of spoken English
6. Commonly used words


The above listed speech perception tests were developed following Egan's

guidelines. Speech audiometric tests have been developed in many different languages,

each to fit the requirements of measuring speech perception based on the specific features

of that language.

Measurement of speech perception tests included the articulation function that

defines word recognition scores at different intensity levels. The measurement starts at a

very low intensity level where the material is unintelligible. As the signal intensity level

is increased, the listener is able to identify correctly more of the stimuli up to a point at

which the intensity is high enough for the listener listening to his native language to

identify all the material without error. Fletcher (1929) demonstrated that the shape of the

curve changes from one material to another and, using the same material, measurement of

speech recognition with different speakers results in different shaped curves. With speech

discrimination loss, the articulation curve not only shifts at the dB axis but it also changes

in shape. The curve reaches a plateau at percentage correct levels below 100% (Davis,

1948). Davis concludes that this effect results from the loss of sensitivity, especially in

the high frequencies, which is important for consonant recognition and clarity of speech.

The shape of the curve is again different for each hearing impaired individual. It might be

of normal shape and shifted to the right indicating higher intensity levels, or it might have

a different shape and not reach the 100% correct identification.









Pediatric Speech Perception Materials

Children's speech perception tests have been developed with different materials

and different tasks appropriate for different age and ability levels. Some tests are "open

set" (i.e. the listener has no knowledge of the category of word or any contextual cues)

and require verbal response (e.g. Phonetically Balanced Kindergarten 50 Word Test,

Haskins 1949), whereas others are "closed set" (i.e. the listener is provided with a set of

3-6 options to chose from or the category of words is specified to provide some cues)

requiring picture pointing (e.g., Word Intelligibility by Picture Identification, Ross &

Lerman, 1970). Following is a detailed description of the most commonly used children

speech perception tests (see table 2-1 for list).

Haskins (1949) developed The Phonetically Balanced Kindergarten 50 Word Test.

Though the Haskins lists appeared on in the author's masters' thesis at Northwestern

University, they have been widely used.The four lists were developed based on the

phonetically balanced word lists (PB-50) used by Egan (1948) for assessment of speech

perception in adults Haskins selected words that were among the 2500 words most used

by kindergarten children (The International Kindergarten Union, 1928). Measurement of

the psychometric function and the equivalence of wordlists were completed with adult

normal hearing participants with one randomization of lists presented at 5 intensity

levels. As a result, lists 1, 3, and 4 were found to be equivalent while list 2 was easier.

The slope of the psychometric function between 20 and 80% word correct was 4%/dB, as

reported in Mayer & Pisoni (1999). In spite of the wide use of these lists, no formal data

collection and analysis was completed with pediatric populations.

Other speech perception audiometric tests for children were mostly developed for

closed sets of words (Elliot & Katz, 1980; Erber, 1974; Erber, 1980; Goldman et al.,









1970; Jerger et al., 1980; Ross & Lerman, 1970), and sometimes with groups of hard of

hearing children as the only participants (Erber & Alencewicz, 1976; Ross & Lerman,

1970). Several closed set picture presentation tests are used regularly in audiology clinics.

These tests include the Word Intelligibility by Picture Identification (WIPI) by Ross and

Lerman (1970), the Northwestern University Children's Perception of Speech (NU-

CHIPS) by Elliott and Katz (1980), Goldman-Fristoe-Woodcock Test of Auditory

Discrimination (Goldman et al., 1970), and The Pediatric Speech Intelligibility test (PSI)

by Jerger et al. (1980). Erber & Alencewicz (1976) suggested a picture pointing closed

test to evaluate the word recognition ability of children with hearing loss that provides a

distinction of word recognition and word pattern recognition.

Adult Speech Perception Materials

Beattie & Warren (1983) described for adult word recognition tests in English, an

increase in intelligibility with the increase of intensity equivalent to 4.5%/dB in the range

of 20 to 80% scores, with an approximation of maximum intelligibility at level s of 25 dB

SL. Wilson, Zizz, Shanks, & Causey (1990) reported the NU-6 word recognition

threshold when spoken by a female speaker to be 4.5%, similar to other studies, while the

intensity level of 50% correct recognition was shifted 5dB to the right (higher than

previous studies). Wilson & Oyler (1997) compared the psychometric function of the

CID W-22 word lists and the Northwestern University No.6 (NU-6) as spoken by the

same talker and found the 50% score level to be at 15.6dB HL for the W-22 and 13.4 dB

HL for the NU-6. The slopes between 20% and 80% points were 4.8%/dB for the W-22

and 4.4%/dB for the NU-6. These results are comparable to speech audiometry data for

other languages For example, Harris et al. (2003a) studied the psychometric function of

wordlists spoken by males and females and found the mean 50% level in Korean to be at









11.4 dB HL for male speakers and 10.7dB HL for female speakers with mean slopes

between 20% and 80% points of 4.4%/dB for male and female speakers. Niessen et al.

(2005a) found a threshold level for Chinese Mandarin speech materials of 5.4 dB HL for

male speakers and 2.3dB HL for female speakers, and mean slopes between 20% and

80% points were 6.3%/dB for male speakers and 7.1% for female speakers. The

difference in the Chinese word lists might have been because disyllabic words were used

in the composition of the lists. Alusi et al. (1974) developed equivalent lists for speech

recognition in Arabic with a threshold level of 22.5 dB HL and a slope of 5%/dB.

In summary, studies of monosyllabic lists in three languages (English, Korean, and

Arabic) show similarity in psychometric function slopes (ranging from 4.4% to 5.1%)

whereas the study of disyllabic Chinese word lists showed steeper slopes (6.3% and

7.1%). Table 2-2 lists the thresholds and slopes of word lists in the different languages

listed above. Review of data displayed in Table 2-2 shows that there is a difference in

threshold among languages, with the Chinese disyllabic words having the lowest

threshold (2.3 dB HL) and the highest for Arabic monosyllabic word lists (22.5 dB HL).

Speech Reception Threshold Material

Studies of speech recognition thresholds (SRT) lists in different languages revealed

steeper slopes than those reported for word recognition lists. SRT measures in English are

composed of disyllabic spondees. Hirsh et al. (1952) reported a psychometric function

slope between 20% and 80% of 8%/dB. Young, Dudley, & Gunter (1982) reported a

slope of 10%/dB, and Wilson & Strouse (1999) reported a slope of 7.4%/dB. In studies of

languages other than English, such as the study of trisyllabic Chinese Mandarin materials,

Nissen et al. (2005b) found a slope between 20 and 80% of 9.7%/dB for a male speaker

and 10.5%/dB for a female speaker. According to Nissen (2005b), slopes of SRT tests in









other languages are at similar levels, i.e., Polish = 10.1%, Spanish = 11.1%, and Italian =

7.3%. Harris et al. (2003b) in their study of Korean disyllabic words found a slope of

10.3% for male speakers and 9% for female speakers. Ashoor & Prochazka (1982, 1985)

reported a slope of SRT word lists of 5% for both adults and children's wordlists. Slopes

of Ashoor's lists are less steep than those reported in other languages, but similar to

slopes of word lists used in Alusi's word recognition scores for adults. Siegenthaler,

Pearson, & Lezak (1954) investigated the speech reception threshold for children using

monosyllabic words and found the slope to be 8.6%/dB between 20% and 80% correct

word recognition. Ashoor & Prochazka found the threshold of word recognition to be at

2.2dB HL for adults (1982) and at 0 dB HL for children (1985). Table 2-3 lists speech

reception threshold wordlists in different languages.

Full List and Half List Use

The use of full lists or half lists depends on the patients' performance on the test.

Studies by Beattie and Warren (1983), Dubno, Lee, Klein, Matthews and Lam (1995),

and Thornton and Raffin (1978) investigating the confidence intervals of using full lists

of 50 words and half lists of 25 words in initial testing and retesting of patients'

performance on speech reception. Thornton and Raffin (1978) described the variability

in speech discrimination scores based on the CID W-22 test, and highlighted the

differences in variability between using full lists and half lists. Their results show the fact

that the closer the scores are to either end of the spectrum of scores (0 or 100) and the

more words are included in the list the less variability there is in scores and the smaller

the confidence interval. For example if a patient scored 96% on a 50 word list the

confidence interval is between 86-100. In other words if the patient scored 96% the first

time and was retested, a score between 86-100 will be considered not different from the









first score. While if the patient scored 96% on a half list of 25 words, the confidence

interval is between 80 and 100. With lower scores, closer to 50%, the confidence interval

grows larger more so for half lists than for full lists. This can be used as an indication for

the need to use a full list versus a half list. That is if the score is closer to 50%, the use of

50 word lists would provide a more accurate measure. In comparing the test retest

results from hearing impaired participants, Beattie and Warren (1983) found the standard

deviation in test retest results using 25 word lists to be 10%, which was reduced to 8%

using 50 word lists and to 6% when using 100 word lists. In Beattie et al.'s judgment, this

difference was not significant to increase the size of test material and was satisfied with

the 25-word list size.

Dubno et al. (1995) studied the correlation between the degree of hearing loss and

word recognition scores. They studied the word recognition scores from 407 ears with

normal hearing and mild to severe hearing loss, with the goal of providing data for

confidence limits of scores on 25 and 50 NU-6 word lists in relation to the PTA. The

authors provided tables of scores for 25 and 50 word lists corresponding to the 95%

confidence limit of best performance (PBmax). They found a correlation between word

recognition score and PTA, where a lower PTA resulted in higher scores. These findings

are intended to help in diagnosis decisions on whether the score is considered within

expected range for the degree of loss or whether it is poorer than expected and thus

requiring additional testing. Dubno et al. (1995) cautioned about the use of these tables to

generalize to other lists since different material would have different results. Still this

gives an indication for clinicians and researchers to be cautious when using word lists at

one presentation level and to keep this data in mind when making clinical decisions.









Arabic Speech Audiometry

Several speech perception tests are available in the Arab countries. These include

speech recognition tests for adults in Moroccan (Messouak, 1956), Iraqi (Alusi et al.,

1974), Egyptian (Soliman, 1976) dialects, and an SRT test (Ashoor & Prochazka, 1982)

in Saudi dialect. Pediatric speech audiometry tests have been developed including an

SRT test in Saudi (Ashoor & Prochazka, 1985) and Egyptian (Soliman et al., 1987b)

dialect, an Arabic word intelligibility (recognition) by picture identification in Egyptian

dialect (Soliman et al., 1987a), and an Arabic speech pattern contrast (ArSPAC) test

developed in Israel (Kishon-Rabin & Rosenhouse, 2000). Some of these publications are

inaccessible (Messouak, 1956; Soliman, 1976; Soliman et al., 1987a; Soliman et al.,

1987b). Insufficient details are found in the literature to allow for in depth discussions of

these tests. Recordings of the test materials are not widely available, even though the

word lists are printed in the publications (Alusi et al., 1974; Ashoor & Prochazka, 1985;

Kishon-Rabin & Rosenhouse, 2000). Allusi et al.'s (1974) and Ashoor & Prochazka's

(1982 and 1985) studies address word recognition comparable to the present study;

therefore these studies are discussed in detail.

Comparison of Arabic Speech Reception Tests

Alusi et al. (1974) and Ashoor & Prochazka (1982) used monosyllabic words in

their word lists. The structure of monosyllabic words was CVC, CVCC, CVVC, and

CVVCC, with all 28 consonants and 6 vowels of standard Arabic represented. Alusi et al.

(1974) divided the 150 monosyllabic words into 6 phonetically balanced lists of 25

words. Ashoor & Prochazka's adult lists (1982) contained 120 words divided into 6

phonetically balanced lists of 20 words, and their children's lists (1985) included 80

words divided into 8 lists of 10 words. Both Alusi and Ashoor based their phonetic









balance of lists on the frequency of consonant and vowel occurrence compared to

continuous text, based on counts conducted at the time of study. Ashoor & Prochazka

(1985) in his children's lists focused on equal distribution of syllable structure more than

the phonetic representation, while he kept the overall phonetic balance in all 80 words.

To ensure word familiarity, both researchers used standard Arabic, the main

teaching language in schools and universities as well as the language of mass media (i.e.,

newspaper, radio and TV broadcast). The word sources of choice were elementary school

books, children's stories, and daily newspapers. Absurd words and technical vocabulary

were excluded. In addition both researchers chose words that are similar in standard

Arabic and colloquial. Ashoor & Prochazka (1982 and 1985) ensured the word

familiarity by collecting ratings from a large number of participants coming from 14

different regions of Saudi Arabia.

Lists were recorded at voice intensity of 70 to 75dB SPL, fluctuation was limited to

+5 dB on volume meter in sound treated booths, and ambient noise did not exceed 30

dBA SPL. Neither researcher used carrier phrases. Alusi et al.,'s recording (1974) was in

standard Baghdad dialect while Ashoor and Prochazka's recordings (1982 and 1985)

were in standard Saudi dialect. The rates of recording varied. Alusi et al. (1974) recorded

8 words per minute, whereas Ashoor and Prochazka recorded 12 words per minute for

adults and 6 words per minute for children (Ashoor & Prochazka, 1982, and 1985).

The intelligibility tests were performed with somewhat different criterion. Alusi et

al. (1974) included 17 participants age 20 to 38 years representing several Arab countries,

since his goal was to develop a test that could be used in different Arab countries. Ashoor

& Prochazka (1982) enrolled 74 adult participants age 20 to 35 years representing 14









rural and urban areas in Saudi Arabia, and 100 children age 4 to 9 years representing

most Saudi dialect areas. Both studies (Alusi et al., 1974; and Ashoor & Prochazka,

1982) presented the stimulus words via headphones to their adult participants, while

Ashoor and Prochazka (1985) presented the stimulus words via sound field to their

pediatric participants. In all studies the authors attempted to measure the difference in

inter-list difficulty and also the difference associated with presentation method. The word

lists were presented at different intensity levels ranging from speech detection thresholds,

increasing by 5dB steps, up to the point where 100% of the words were identified

correctly. There was no difference between lists in terms of difficulty or the order of

stimuli presentation in ascending or descending dB level, within each study. Resulting

articulation function curves from all three studies were similar to those of other

languages, including English, in shape and slope. The findings suggest the lists are

suitable as speech recognition measure.

Ashoor & Prochazka (1982) found a slight difference between the adult groups of

students and non-students, i.e., a curve shift further to the right for the latter group

indicating higher threshold levels (in dB). In addition, he found a difference between two

age groups (4 to 5 years and 5 to 9 years). The younger group needed higher intensity

levels than the older group to reach threshold level of 50% word recognition, a finding

that may be related to maturation and knowledge of the language.

Use of Arabic Speech Reception Tests

Though Alusi et al. (1974) and Ashoor & Prochazka (1982 and 1985) recorded

their material, these recordings were not marketed for wide distribution. The limited

distribution may have several explanations. One is the small number of audiologists and

audiological services in Arab countries. For example, Saudi Arabia has one of the most









advanced services for communication disorders among the Arab countries. With a

population of 25 million, there are 14 registered audiologists at the Saudi Speech

Pathology and Audiology Association and five facilities that provide audiological

services (SSPAA, 2004). The second possible explanation for the limited distribution of

speech materials is the difference in dialects between Arab countries. Although Arab

countries share the standard written Arabic language, there is a wide range of dialects

(Fatihi, 2001). Published speech recognition tests are in Moroccan, Baghdadi, Egyptian

or Saudi standard dialect. The possibility of using one test across the Arab countries has

not been investigated. Alusi et al. (1974) has suggested the possibility of using the word

lists he developed in all Arab dialects since the words were taken from standard Arabic.

However, Alusi's speech materials were recorded in a Baghdad standard dialect. In

developing the speech test, Alusi had a limited number of participants (17) representing

"several" Arab countries (the author did not specify which countries), who were young

educated adults. The sample did not necessarily represent the large Arabic speaking

population). However, Alusi et al. did attempt to meet the criterion of word familiarity by

choosing words from children's books and newspapers in order to include educated and

un-educated populations. He did not describe a specific comparison between participants

from different countries to support his argument.

Dialectal Differences in Arabic

One of the goals in the present study is to determine whether there is a difference in

word recognition scores for normal hearing and hearing impaired children listening to

Jordanian and Saudi dialects. The issue of Arabic language diversity and its dialects is

important in the selection of speech material and speaker. Although a single standard

Arabic language is used in all Arab countries, dialectal differences do appear. The









difference between standard and dialect with the Arabic language, similar to other

languages, such as Greek, Swiss German, and Haitian Creole, is referred to as diglossia.

Furgeson (1959) defined diglossia as the presence of a stable situation of a language in

which there is a dialect of a language primarily used in daily communication and a very

different superimposed variety of the language that is part of a highly respected large

body of written literature or a previous period. The written literature is learned by formal

education and is used in formal speech (such as news and political speeches), but not in

every day conversation. All written materials (e.g., school books, news paper,

commercial material, official documents, and instructional materials) are written in

standard Arabic and in social settings people use the colloquial dialect. Thus children

learn the colloquial dialect first and the standard language is taught formally in

educational settings (Al-Kahtani, 1997; Altoma, 1969; Fatihi, 2001; Ferguson, 1956).

With the increase in mass media and early education, children nowadays are more

exposed to standard Arabic through television programs, radio and early reading

experience.

The differences between the two versions of Arabic are grammatical, phonetic, and

lexical in nature. Standard Arabic is considered syntactically more complex and richer in

lexicon. In spite of these differences, there are many similarities. Altoma (1969) found

that 83.5% of words in different colloquial dialects are shared with standard Arabic, as

well as the syllabic structures of words. In general standard Arabic is considered a more

prestigious language but there is no competition between the two versions of language

since each serves a different purpose and they are not totally interchangeable in use by









situation. In other words there are situations where standard Arabic is inappropriate and

visa versa (Al-Kahtani, 1997; Ferguson, 1959; Ibrahim, 2000).

Abd-el-Jawad (1987) and Al-Kahtani (1997) reported that educated Arabic

speakers frequently switch between standard and colloquial Arabic in a very natural

manner. Differences are to some extent between social groups, but these differences have

no effect on communication. Linguists have determined the status of a dialect based on

the characteristic of mutual intelligibility among dialects. Mutual intelligibility is based

on a scale of the physical proximity of the regions, that is, the closer the region the more

mutually intelligible, and the further apart the regions, the less mutually intelligible

(Fatihi, 2001). The differences and similarities between dialects and between standard

Arabic and dialects must be considered when developing speech perception test materials

and when choosing speakers. Several groups of researchers (Beattie, et al. 1975; Beattie,

et al. 1977; Carhart, 1965; Doyne & Steer, 1951; Hirsh, et al. 1954; Hood & Poole 1980;

Kruel, et al. 1969; Palmer, 1955; Wilson & Carter, 2001; Wilson & Oyler, 1997)

demonstrated significant differences in speech perception scores with different speakers,

regardless of gender or age, assuming the speech perception test material were recorded

by native speakers of English in standard dialects. In the case of Arabic, it can be argued

that there is no difference among the different Arab countries when using standard

Arabic. The question whether standard Arabic is a representative sample of the spoken

dialects remains unanswered.

Special Considerations for the Present Study

Participants

Speech perception differs significantly between age groups. Elliot (1979) found

that scores of children age 9 were poorer than older children on speech perception in









noise test (SPIN), while the older group (15 to 17 year olds) scores were comparable to

adult scores. Elliot, Connors, Kille, Levin, Ball & Katz (1979) found no significant

difference in scores for children age 5 to 8 years while 10 year olds performed at adult

levels (no 9 year old participants were included in this study). Schwartz and Goldman

(1974) assessed the performance of young children in nursery, kindergarten and first

grade and found significant differences between the three groups. In another study

Sanderson-Leepa & Rintelmann (1976) compared the speech performance of children

ages 3.5 to 11.5 on different speech perception tests and found no significant difference

in the 7.5 and 9.5 age groups on the WIPI and PBK-50 tests compared to the younger and

older groups. Ashoor & Prochazka (1985) found similar age differences for the Saudi

Arabic test when comparing scores for 4 to 5 year old children to scores for 6 to 9 year

old children'. Based on these results, participants aged 6 to 9 years were recruited in the

present study.

As mentioned above, studies by Elliot (1979), Elliot et al. (1979), Sanderson-Leepa

& Rintelmann (1976), Schwartz & Goldman (1974), and Goldman et al. (1970), show

that significant differences in word recognition scores are found for children younger

than 6 years and older than 9 years in comparison to children age 6 to 9 years. These

others reported no significant difference in word recognition scores between children

aged 6 through 9 years.

Speakers

Individual differences between speakers can affect speech perception scores (Hood

& Poole 1980). Different speakers produce different articulation curves when using the

same words. Most words maintain their order of difficulty across speakers. Palmer (1955)

investigated the effect of gender on speech perception scores. He based his question on









the notion that hard of hearing individuals have an easier time hearing men's voice than

women's voice. In his study he used nine speakers; three male adults, three female adults,

and three female children. When he compared scores from each group for hard of hearing

and normal hearing participants at a fixed intensity level, no significant difference was

found across speakers. One goal in the present study is to investigate the possibility of a

difference in scores with different dialects. Following Palmer's methodology, in the

present study three Jordanian and three Saudi male speakers were selected to complete

the word recording.

Stimuli for Dialectal Differences

In the present study the researcher's goal was to investigate the possibility of

difference in scores with different dialects since. In this study, a +9 dB signal to noise

ratio was used to avoid ceiling effect in normal hearing children's performance and to

reduce the variability in scores. The choice of +9 dB signal to noise difference was based

on the Goldman et al. (1970) study of speech perception of children in quiet and in noise.

Goldman et al. (1970) observed a reduction in scores compared to the quiet condition

that started at -9 dB noise level. Resulting word recognition scores are expected to be less

than 100% correct. Schwartz & Goldman (1974) used the same level of signal to noise

ratio (+9 dB) to investigate the effect of different contexts and listening environments

(quiet and noise). They observed a significant increase in number of errors for all

contexts when noise was introduced. The effect of the smallest amount of noise was

clearly demonstrated in a study by Larson, Petersen, & Jacquot (1974) when they tested

the use of NU-6 word lists with children age 5.5 to 6.5 years of age. The presence of

noise at +20 dB S/N ratio had a significant effect on the children's performance

compared to adult performance under the same conditions. Keep in mind that for their






26


study Larson et al. have used adult material to test very young children. Based on the

above listed studies, a signal to noise level of+ 9 dB will be used for the present study.











Table 2-1 Children's speech reception tests in English.
Test Investigator Stimulus Respons Respons Target Published
e format e task population
PBK-50 Haskins 1949 Monosyll. Open set Verbal 6-9 years No
Words
GFW Goldman, Monosyll. Closed Picture > 4 years Yes
Fristoe, & Words set pointing
Woodcock
1970
Spondee Erber 1974 Spondee Closed Writing 8-16 years No
recognition words set
n test
WIPI Ross & Monosyll. Closed Picture 3-6 years Yes
Lerman 1970 Words set pointing
BKB Bench, Koval, Sentences Open set Verbal 8-15 years No
& Bamford
1979
PSI Jerger & Monosyll Closed Picture 3-10 years Yes
Jerger 1980 words and set pointing
sentences & verbal
NU- Elliott & Katz Monosyll. Closed Picture >2.5 years Yes
CHIPS 1980 Words set pointing
ANT Erber 1980 Numbers Closed Picture 3-8 years No
set pointing


Table 2-2 Speech recognition word lists' mean dB HL levels at 50% and slopes.
Investigator Language Target Test dB @ 50% Slope
population %/dB
Wilson & Oyler English Adults CID 15.6 4.8
1997 W-22
Wilson & Oyler English Adults NU-6 13.4 4.4
1997
Harris et al. Korean Adults Male 11.4 5
2003a speaker
Harris et al. Korean Adults Female 10.7 5.1
2003a speaker
Nissen et al. Chinese Adults Male 5.4 7.3
2005a Mandarin speaker
Nissen et al. Chinese Adults Female 2.3 8.2
2005a Mandarin speaker
Alusi et al. 1974 Arabic Adults 22.5 5
Haskins 1949 English Children PBK-50 NA 4










Table 2-3 Speech reception threshold wordlists' mean threshold
between 20 and 80%.


dB HL levels and slopes


Investigator Language Target Test dB @ 50% Slope
population %/dB


Hirsh et al. 1952
Yourng et al.
1982
Ashoor &
Prochazka 1982
Harris et al.
2003b
Harris et al.
2003b
Nissen et al.
2005b
Nissen et al.
2005b
Siegenthaler et
al. 1954
Ashoor &
Prochazka 1985


English
English

Arabic

Korean

Korean

Chinese
Manderin
Chinese
Manderin
English

Arabic


Adults
Adults

Adults

Adults

Adults

Adults

Adults

Children

Children


CIDW-1 0.5
CIDW-2 -0.3


Mono-
syllabic
Male
speaker
Female
speaker
Male
speaker
Female
speaker
Mono-
syllabic
Mono-
syllabic


2.2

6.8

5.6

-0.2

-0.7

NA

0


8
10


10.3

9

9.7

10.5

8.6

5














CHAPTER 3
METHODOLOGY


This dissertation consists of two studies. In the first study, monosyllabic word

intelligibility was investigated as a function of presentation level in Jordanian dialect.

The following questions were addressed in the first study: Is there a significant difference

in word recognition ability of Jordanian children, given increases in intensity presentation

levels? Can four parallel word lists be developed (i.e. their psychometric qualities do not

differ)? The results of this study were used in the development of four parallel lists of

fifty monosyllabic words to be used as a speech perception test in Jordan.

The second study compared word recognition scores of Jordanian children listening

to words spoken in Jordanian and Saudi dialects. Specifically, the study was designed to

answer the following questions: Do word recognition abilities of Jordanian children differ

when listening to words presented in a Jordanian dialect compared to words presented in

a Saudi dialect? Do word recognition abilities of Jordanian children who display normal

hearing and hearing disabilities differ when hearing words presented in a Jordanian

dialect compared to words presented in a Saudi dialect? Does word recognition ability

differ when hearing speakers using the same dialect?

Participants

Inclusion criteria included age range and hearing sensitivity. Age range was

limited to children age six to nine years. Children displaying normal hearing sensitivity

and mild to moderately severe hearing loss were recruited. Hearing status was determined









using pure tone average thresholds of 500, 1000 and 2000 Hz. A pure tone average

threshold of 15 dB HL or less were considered normal hearing and thresholds greater

than 26 dB HL and less than 69 dB HL were considered within the range of mild to

moderately severe hearing loss. Tympanometry and pure tone audiometry were used to

determine hearing sensitivity fitting inclusion criteria. If the tympanogram showed

normal middle-ear function, the researcher proceeded to pure tone audiometry using

supra aural headphones (TDH-39). The participant was excluded if middle-ear function

was abnormal. If the pure tone average did not meet criteria, the testing was terminated.

In both cases results were explained to the parent if present. In case of hearing loss

detection or abnormal middle-ear function, the caregiver or adult accompanying the

participant was informed and briefly counseled on intervention.

Participants were recruited by word of mouth from the larger Amman area.

Participating children came to the clinic in Amman accompanied by their caregivers or an

adult with the parent's permission. Twenty normal hearing participants, ten males and

ten females, were included in the first study. A total of ten normal hearing, four males

and eight female, and ten hard of hearing, six male and four female, participated in the

second study. For detailed description of participant recruitment see Appendix A.

Speakers

Speech materials were recorded by native Arabic speakers who were recruited

from the area of Provo, Utah, the location of Brigham Young University where the

recording took place. Potential speakers were initially personally contacted by phone or

by word of mouth. The purpose of the study and the speaker's role in the study were

explained. Five adult Jordanian males agreed to participate in the recording. Preliminary

5-minute recordings of continuous speech were made for the purpose of judgment of









dialect and clarity of speech. Five different adult Jordanian natives (three females and

two males) were asked to judge the speakers using two criterion, i.e., the dialect was a

general Jordanian dialect, and the ease of understanding the speaker's speech as rated on

a 10-point scale. The top ranked speaker completed the recording of the 304

monosyllabic words for study one. For the completion of recordings for study two the 3

top ranked Jordanian speakers were selected and agreed to participate in the recording.

Only three Saudi speakers agreed to participate in the recording for study two.

For the purpose of creating the 8 multi-speaker babble noise, four additional female

Arabic speakers were recruited. The female speakers included the primary investigator

and three females recruited form the Provo area by word of mouth. The recordings from

the 2 male Jordanian speakers who were not included in the final recording and 2 of the

speakers used for word recording randomly selected as well as the 4 female speakers

were used to compose the noise track.

All speakers have acquired Arabic in their home country. They were all students or

spouses of students who have come to the USA to peruse higher education degrees. All

speakers have been in the USA for no longer than 5 years.

Recording and Editing

The recording took place in an anechoic chamber at Brigham Young University

campus in Provo, Utah, USA. The chamber has a 0 dB SPL noise floor allowing for a

recording of 60 dB signal to noise ratio (S/N). The equipment used in the recording was

the same as reported by Harris et al. (2003a) and Nissen et al. (2005a). The recording

microphone used was a Larson-Davis model 2541, which has a flat frequency response

up to 20 kHz. The microphone was positioned approximately 6 inches from the speaker

at a 00 azimuth and covered by a 3 inch windscreen. The microphone signal was









amplified by a Larson-Davis model 900B microphone preamp, coupled to a Larson-Davis

model 2200C preamp power supply. The signal was digitized at a 44.1 KHz sampling

rate with 24-bit quantization by an Apogee AD-8000 analog-to-digital converter, with

preamp gain set to utilize the full range of the 24-bit analog-to-digital converter. The

digitized signal was stored on a hard drive for later editing. The speech material was

printed in large font on A4 paper and clipped to a clipboard that was suspended in front

of the speaker at a comfortable height to allow him to read the words facing the

microphone at 00 azimuth. The speaker was asked to say each word a minimum of four

times. A native judge (the primary investigator) rated each word for clarity and the best

production was selected for inclusion in the word recognition test. The intensity of each

selected word was edited using Sadie Disk Editor software (Studio Audio & Video Ltd.

2004) to produce a final recording with the same average root mean square (RMS) power

as the 1 kHz calibration tone in an attempt to equate the test word threshold audibility

(Harris et al. 2003a; Nissen et al. 2005a; and Wilson and Strouse, 1999). The recording

was converted from 24- to 16-bit quantization using the NS high dither option in the

Sadie Disk Editor software. The final word recording was digitally saved as wav files.

Each word was saved as a separate file. A total of 250 words were selected to be included

in the word recognition test; study one. For study two 33 monosyllabic words were

selected from Ashoor and Prochazka's (1985) speech reception threshold list for children.

The words were edited digitally to ensure same level of RMS as the 1 kHz tone.

The multi speaker babble material was completed using 4 of the male Jordanian

speakers and 4 Jordanian female speakers. The speakers were asked to read an article

taken form a Jordanian daily newspaper. The recording was completed following the









same procedure as for the word list. Three minutes of continuous speech was selected for

each of the 8 speakers. All silence between words and sentences was removed for each

speaker. Then the RMS level for each speaker was adjusted to have all 8 speakers at the

same level. Next the recordings were digitally mixed and run through a

compression/limiter to limit the range of peak variability. The resulting recording average

RMS was measured and adjusted to equal the level of the 1 kHz calibration tone. The 8

multi-speaker babble noise was saved in a separate wav file that would allow the play of

noise independently from the word lists.

Instrumentation for Data Collection

Data collection took place at the Middle East Hearing Association clinic in

Amman, Jordan in a sound treated booth designed locally for hearing testing purposes.

Pure tone and speech audiometry were conducted using an Interacoustics AC40

audiometer, connected to TDH-39 headphones. The audiometer was calibrated using a

Larson Davis System 824 sound level meter and a 6cc coupler. Calibration was based on

ANSI standards 2004. Measurements of sound levels at octave and half octave

frequencies met the ANSI standard 2004 with a deviation range of -0.6 to +0.3 dB. The

sound levels for speech through external input A and external input B were consistent

with ANSI standard 2004 with a deviation of-0.5 to +0.1 dB. Repeated measurements of

sound pressure level produced by the audiometer were within permissible ANSI tolerance

level of+ 3dB for frequencies of 500 to 4000Hz and +5dB for 6000 to 8000 Hz.

Special software was developed by Dr. Richard Harris at Brigham Young

University (2005) to control the playback of 1 kHz tone, noise and word lists from wav

files. The software also provides the documentation of data in an excel file spread sheet

with the following details: the date and time of presentation, participant assigned number,









participant gender, test ear, speaker gender, intensity level, signal to noise ratio, list

name, time of recording per list, wav file, word (in this case in Arabic), and the score.

Prior to data collection, the VU meter was adjusted to 0 VU using 1 KHz tone. For

the purpose of consistency in speech audiometry data collection, the left headphone was

randomly selected to always be used over the test ear. This step was taken to reduce

variability in the sound level presented.

Study One

Speech Material

Speech material was selected following Egan's (1948) criterion of monosyllabic

words, representative of spoken Arabic and commonly used words. A total of 304

monosyllabic words were selected for initial recording. Words that have similar forms in

colloquial and standard Arabic were selected. Children and teachers of first through third

grade rated these words as familiar. Rating took place at four elementary schools in

Amman Jordan. The words were read aloud to the children, they were asked to raise their

hand if they knew the meaning of the word and were able to use it in a meaningful

sentence. To ensure the accuracy of their response, the primary investigator (Nadia

Abdulhaq) started with three trial words that were familiar to children (window, door,

and table), randomly asked children to put words in a sentence, in addition unfamiliar

words (from old classical text) and nonsense words were included. For more detailed

description of word selection see Appendix B.

The primary investigator judged the recorded 304 words for clarity of recording

and in conjunction with familiarity ratings selected 250 words to be used in data

collection. The 250 words were divided randomly into two sets of ten lists of 25 words.

To allow the presentation using the software developed by Dr. Harris, a text file was









created for each list Al to A10 and B to B10 for each set of randomization respectively.

Lists were presented in counterbalanced order. Lists Al through A10 were presented to

participants 1 through 10 and lists BI through B10 were presented to participants 11

through 20. See Table 3-1 for order of presentation and level of presentation. The signal

was routed to the subjects through the Interacoustics AC40 audiometer, via the TDH-39

headphones. Speech material was presented at 10 different intensity levels starting at 0

dB HL and increasing by 5 dB increments up to 45 dB HL; to include a range of low to

high word recognition scores.

Procedure for Data Collection

Each participant was given the following general instructions:

"You will hear some words through the headphones, your will repeat the words you

hear. For example, if you hear the word /kitab/ (book), you would say?" The researcher

would wait for the participant to repeat the word. "That is good. And if you hear the word

/daftar/ (copy book) you would say? and the word /madrasa/ (school) you would say?"

Once the participant has followed instructions further details were provided:

"The words will be spoken very softly; you might or might not hear them at first.

That is OK. If you hear a word, repeat it. If you don't, just wait to hear the next one. The

words will get louder and louder."

The earphone was placed over the test ear. Two lapel microphones were attached to

the participant's shirt, one for talk back and the second for audio input of the audio-visual

recording. A video camera was set in front of the participant to obtain video recording.

The video camera was adjusted in height and angle to provide the best view of the

participant's face.









After the participant repeated a word, the investigator entered 1 for a correct

repetition and 0 for an incorrect repetition. If the participants did not repeat the word, the

investigator periodically used the talk forward microphone to ask the participant if he/she

had heard the word, and to encourage him/her to repeat whatever they heard. The

investigator waited 3to 5 seconds before recording an incorrect score in the case of no

repetition. After each list presentation, the presentation level was increased by 5dB.

Participants were encouraged between lists using verbal praise ("good job", or "you are

doing really well"). Most participants did not need a break or rest period throughout the

data collection procedure. All data for each participant, including presentation order and

scoring details, were saved in excel spreadsheets.

Statistical Analysis

The dependent variable, defined as correct word recognition, was saved in binary

format (correct versus incorrect). The independent variables were different presentation

intensity levels and different word lists. The raw data were used in logistic regression

analysis. The logistic regression analysis provides derived variables that are key

parameters for the regression line that may be tested and compared between conditions

(in this study, the word lists). The raw scores of each word in the four lists were used in

the logistic regression analysis to calculate the logistic parameters for each list and half

list. See Appendix C for details on logistic regression analysis.

Recall that this study was designed to test the following hypotheses: 1. Increased

presentation sound intensity level will improve word recognition ability. 2. It is possible

to create four word lists that are parallel in function and can be used interchangeably to

test Jordanian children's word recognition ability.









Study Two

Speech Material

The speech material consists of 33 monosyllabic words from Ashoor's (1985)

children's list. Ashoor's list was used as source of material because it was developed for

the purpose of Arabic speech perception specifically for children. The selection of the 33

words was based on two criteria: monosyllabic words and familiar to Jordanian children.

Recall that Ashoor's list consisted of a mixture of mono- and di-syllabic words, familiar

to Jordanian Arabic speaking children. The familiarity of words to Jordanian Arabic

speaking children was established through ratings provided by children in 1st through 3rd

grades (Abdulhaq, unpublished). See Appendix D for the list of 33 words and their

meaning in English.

A total of six recordings, each recorded by a different speaker (three Jordanian and

three Saudi speakers), were presented to all participants in counterbalanced order at

constant intensity levels (see table 3-2 for details). The speech material was presented by

playing wav files from a standard desktop PC connected to external inputs of the

audiometer using the same software as described in study one.

The words were presented to the hard of hearing participants at an audible level of

40 dB SL (re: the participant's pure tone average). For normal hearing participants, the

multi-speaker babble track was presented then the word list was presented. Words were

presented at 50 dB HL and the multi-speaker babble noise at 41 dB HL, i.e., with a signal

to noise ratio (SNR) of +9 dB.

Statistical Analysis

For the purpose of statistical analysis of the data in these studies, the dependent

variable was defined as the correct word recognition score, and the independent variables






38


were defined as dialect, speakers, and hearing status. Raw scores were used in mixed

ANOVA to determine difference between speakers and dialects. To determine the effect

of dialect on normal hearing and hard of hearing children's scores a paired t-test of

dialects based on scores from each group separately was used. To determine the

significance of differences between speakers within a dialect paired t-tests of speakers

were used based on scores from all participants.













Table 3-1 Order of list presentation by participant and level of presentation
Participant Presentation dB HL level
0 5 10 15 20 25 30 35 40 45
1 Al A2 A3 A4 A5 A6 A7 A8 A9 A10
2 A2 A3 A4 A5 A6 A7 A8 A9 A10 Al
3 A3 A4 A5 A6 A7 A8 A9 A10 Al A2
4 A4 A5 A6 A7 A8 A9 A10 Al A2 A3
5 A5 A6 A7 A8 A9 A10 Al A2 A3 A4
6 A6 A7 A8 A9 A10 Al A2 A3 A4 A5
7 A7 A8 A9 A10 Al A2 A3 A4 A5 A6
8 A8 A9 A10 Al A2 A3 A4 A5 A6 A7
9 A9 A10 Al A2 A3 A4 A5 A6 A7 A8
10 A10 Al A2 A3 A4 A5 A6 A7 A8 A9
11 Bl B2 B3 B4 B5 B6 B7 B8 B9 B10
12 B2 B3 B4 B5 B6 B7 B8 B9 B10 B1
13 B3 B4 B5 B6 B7 B8 B9 B10 Bl B2
14 B4 B5 B6 B7 B8 B9 B10 Bl B2 B3
15 B5 B6 B7 B8 B9 B10 Bl B2 B3 B4
16 B6 B7 B8 B9 B10 Bl B2 B3 B4 B5
17 B7 B8 B9 B10 Bl B2 B3 B4 B5 B6
18 B8 B9 B10 Bl B2 B3 B4 B5 B6 B7
19 B9 B10 Bl B2 B3 B4 B5 B6 B7 B8
20 B10 Bl B2 B3 B4 B5 B6 B7 B8 B9










Table 3-2 Word list and dialect presentation order.
Participants Present. order
1st 2nd 3rd 4th 5th 6th
1 J1 S1 J2 S2 J3 S3
2 Si J1 J3 S3 S2 J2
3 S2 J2 J1 S1 S3 J3
4 J2 S2 S3 3 J1 S1
5 S3 J3 S1 J1 J2 S2
6 J3 S3 S2 J2 S1 J1
7 J1 S1 J2 S2 J3 S3
8 S1 J1 J3 S3 S2 J2
9 S2 J2 J1 S1 S3 J3
10 J2 S2 S3 J3 J1 S1
J = Jordanian speaker, S = Saudi speaker. The same order of presentation was
used for the normal hearing and hard of hearing group.














CHAPTER 4
RESULTS


Study One

Twenty individuals participated in this study. Pure tone thresholds average (PTA)

indicated that all participants fit the criteria of a PTA of 15dB HL or better with a mean

PTA of 7.8 dB HL (standard deviation = 2.8, range = 3.3 to 12.5 dB HL). The better ear

was selected based on PTA or at random when both ears had the same PTA. Ten

participants were tested on the right ear and 10 were tested on the left ear. Average pure

tone thresholds for both ears of all participants are shown in Figure 4-1. For detailed

description of participants see Table 4-1.

Most participants completed the 250 words in a single session of 20 and 29 minutes

without a break other than the time it took to load each new list and adjust the intensity of

presentation. The pacing of the test depended on how quickly participants responded; the

program required the examiner to record the response before the next word could be

presented.

The raw score data of all participants at the 10 intensity levels were compiled for

each of the 250 words. The total of responses resulted in a score of correct recognition

out of 20 possible correct score per word. Words then were reordered by difficulty from

easiest to most difficult based on the total score (with lower scores indicating higher

difficulty). Four lists of 50 words each were created using the top 200 words ranked by

difficulty, as follows: Words that were tied in difficulty ranking were randomized prior to









sorting into lists. To help ensure range of difficulty between lists, the four most difficult

words were assigned randomly to four lists. The next four most difficult words again

were assigned randomly to the four lists and so on until a total of 50 words were included

in each list. The lists were labeled 1, 2, 3, or 4.

Within each list, the word's level of difficulty was used to assign it in order to

create two half lists of 25 words from each original 50-item list with the goal of having

an equal range of difficulty on both half lists. The half lists were labeled 1A and 1B (the

two lists taken from list 1) through 4A and 4B (the two lists taken from list 4). This

procedure used, to distribute words into lists, has been successful in developing difficulty

equivalent word lists in studies by Harris et al. (2003a and 2003b) and Nissen et al.

(2005a and 2005b). The eight 25-Arabic monosyllabic word half lists are presented in

Tables 4-2. The words are written in Arabic with their phonetic transcription, please see

appendix E for the English translations of the words.

Word recognition scores increased with the increase of intensity. A scatter plot of

the percent correct of responses, for the half lists (25 words) as a function of stimulus

level, is presented in Figure 4-1. Scores at the lowest intensity level of 0 dB HL ranged

from 0 to 2% while scores at the highest intensity level of 45dB HL ranged from 94 to

96%. Chi-square analysis (X2) of the half lists showed a highly significant effect of

intensity (X2 = 1,134.18, p <0.0001). Statistical analysis of data confirmed, as expected,

that scores at different intensity levels are not equal.

Differences between lists (X2 (3) = 1.67, p =0.64) and between half lists (X 2 (7) =

8.25, p = 0.31) were not significant. The mean slope of full lists at the 50% point was

4.44%/dB and ranged from 4.21%/dB to 4.63%/dB (range = 0.42%/dB). The mean slope









at the 20-80% range was 3.84%/dB ranging from 3.65 to 4.01%/dB (range = 0.36%/dB).

The mean threshold was 21.25 dB, and ranged from 21.21 dB to 21.32 dB (range = 0.107

dB). For the half lists, the mean slope at 50% was 4.47%/dB and ranged from 3.86 to

5.36%/dB (range = 1.5%/dB). The mean slope at the 20-80% range was 3.87%/dB with a

range from 3.34 to 4.64%/dB (range =1.30%/dB). Mean threshold was 21.25 dB, and

ranged from 21.04 dB to 21.38 dB (range = 0.34 dB). Table 4-3 and Table 4-4 include the

detailed results of the logistic regression parameters, the slopes at the 50% and 20-80%

levels, threshold intensity level, as well as threshold deviation from mean threshold, for

the lists and half lists respectively. Figure 4-3 shows the psychometric function for all

four lists and Figure 4-4 shows the psychometric function for the eight half lists. By

visual inspection, the curves are almost identical.

The intensity levels at the threshold point had a narrow range around the mean

value, from -0.03 to 0.11 dB for the full lists and from -0.21 to 0.13 dB for the half lists.

The difference between the calculated threshold and the mean threshold for each list and

half list was used to adjust the intensity of each list and half list digitally to achieve a

better equivalence between lists and half lists. The percent correct was recalculated based

on these adjustments for all lists and half lists and an adjusted logistic regression curve

was created. Figure 4-5 shows the difference between the unadjusted and adjusted curves.

The difference between the unadjusted and adjusted curves was small and barely

detectable.

Study Two

Ten normal hearing and 10 hard of hearing individuals participated in this study.

Pure tone threshold average measures indicated that all participants fit the criteria of a

PTA of 15dB HL or better with mean PTA of 6.3 dB HL (SD = 2.8, minimum = 1.7dB









HL and maximum = 10dB HL). The better ear was selected based on PTA or at random

when both ears had the same PTA; as a result 6 participants were tested on the right ear

and 4 were tested on the left ear. Figure 4-6 shows the average of pure tone thresholds of

normal hearing participants, for detailed description see Table 4-5.

Pure tone threshold average measures indicated that all ten hard of hearing

participants had a mild to moderately severe hearing loss with a mean PTA of 50.2 dB

HL (SD = 11.8, minimum = 31.7dB HL and a maximum = 63.3dB HL). All but two

participants had sensorineural hearing loss (SNHL). One participant (#2 in table 4-6) had

chronic conductive hearing loss and one participant (#7 in table 4-6) had a mixed loss.

Again the better ear was selected based on PTA or at random when both ears had the

same PTA; resulting in 5 right and 5 left ears being used. Figure 4-7 shows the average of

pure tone thresholds of hard of hearing participants, for detailed description see Table 4-

6.

Most participants completed all six lists of 33 words in a range of 13 tol8 minutes,

and did not require a break in data collection. The pacing of data collected depended on

how quickly participants responded; the program required the examiner to record the

response before the next word could be presented. After data collection was completed

the score for each participant per speaker was calculated into percent correct. See Table

4-7 for detailed results. These data were analyzed to evaluate the hypotheses.

Mixed analysis of variance (ANOVA) indicated a significant difference between

dialects and speakers. Scores of all participants for all speakers were included in a mixed

ANOVA of speaker and dialect. In summary, analysis showed a significant difference

between dialects (F= 8.865, p = 0.008), significant differences among speakers (F =









6.181; p = 0.005), and a significant difference between speakers within each dialect (F =

5.153; p = 0.011). For detailed results see Table 4-8.

Paired sample t-test analysis indicated significant difference between dialects for

the normal hearing group, but no significant difference for the hard of hearing group. The

scores for each participant were averaged by dialect, and the averages were used in paired

sample t-test for each group separately (hard of hearing and normal hearing). Difference

between dialects for the normal hearing participant group were significant (t-test = -2.923

p = 0.017). Differences between dialects for the hard of hearing participant group were

not significant (t-test = -1.327 p = 0.217). Thus, hearing loss seemed to outweigh

dialectal difference. See Table 4-9 for detailed results.

Within each dialect one speaker was significantly different than the other two

speakers. A paired sample t-test was performed comparing results of speakers within

each dialect for all participants. Results indicated that scores associated with speaker J1

(in the Jordanian dialect) were higher than those for J2 and J3 (t = 4.203 and 3.802,

respectively; p = 0.0001 and 0.001). Paired scores associated with speakers J2 and J3

were not different (t-test -0.496, p= 0.625). For the Saudi dialect, scores associated with

speaker S2 were significantly lower than those for S1 and S3 (t-test = 2.658 and -2.183

respectively and p = 0.016 and 0.042). Differences between speakers S1 and S3 were not

significant (t-test = 0.047 and p = 0.962). See Table 4-10 for detailed results.

















Frequency in Hz

250 500 1000 2000 4000 800(
0

10 -

20

30

B 40

* 50

60


- Right
--Left


Figure 4-1 Average pure tone thresholds at all frequencies for all 20 normal hearing
participants at 250, 500, 1000, 2000, 4000, and 8000 Hz.


Half List Raw Scores


Ar


KX


9
V


2 4 6 8 10 12


14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

PresentationLevel (dB HL)


Figure 4-2 Half lists raw data scatter plot


1

0.9

0.8

" 0.7

0.6

S0.5

6 0.4

S0.3

S0.2

0.1

0


0



























1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55

Figure 4-3 Psychometric function of 4 lists of 50 words based on calculated percent
correct.


//


- 1A
-- 1B
--2A
--2B
--3A
--3B
--4A
--4B


7/


x


1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55


Figure 4-4 Psychometric function of 8 half lists of 25 words based on calculated percent
correct.


I



























20


S100

Lists 1-4 Adjusted Half-lists 1A-4B Adjusted
80

S60


40


20


0
0 10 20 30 40 50 0 10 20 30 40 50
Presentation Level (dB HL)





Figure 4-5 Arabic monosyllabic psychometric functions for lists 1-4 (left) and half-lists
1A-4B (right). The top two figures depict unadjusted psychometric functions
and the bottom two figures depict psychometric functions adjusted for equal
performance at 50% correct recognition.











































Figure 4-6 Average of pure tone thresholds of normal hearing participants at 250, 500,
1000, 2000, 4000, and 8000 Hz.


Figure 4-7 Average of pure tone thresholds of hard of hearing participants at 250, 500,
1000, 2000, 4000, and 8000 Hz.


Frequency in Hz
250 500 1000 2000 4000 8000


10

20

30

540
. 50 Right

6 60-

70

80

90

100


Frequancy in Hz
250 500 1000 2000 4000 8000
0

10 -

20

30

S40
-0- Right
-- Left
60 -

70 -

80

90

100












Table 4-1 Normal Hearing participants' age, gender, test ear and PTA

Participant Gender Age Test ear PTA
4 F 6 Left 6.7
8 F 6 Right 10.8
10 F 6 Right 5.0
5 F 7 Left 9.2
7 F 7 Left 9.2
12 M 7 Right 3.3
15 M 7 Right 7.5
16 M 7 Right 5.8
19 F 7 Right 5.8
20 M 7 Right 6.7
6 F 8 Right 12.5
11 M 8 Right 4.2
1 M 9 Left 6.7
2 M 9 Left 4.2
3 F 9 Right 9.2
9 F 9 Left 9.2
13 M 9 Left 10.8
14 M 9 Left 6.7
17 M 9 Left 11.7
18 F 9 Left 11.7

Mean 7.8 7.8
Standard Deviation 1.2 2.8
Range 3.0 9.2
Minimum 6.0 3.3
Maximum 9.0 12.5







51






Table 4-2 Half word lists and transcription in IPA.
1A 2A 3A 4A
khair rab mars Sams
S ku:2 a ha:3 %e xeil U ta:3
i bs:t zir L 3in L ti:n
i ra:s i laf JT ha:l jj bi:r
j su:? ,2 na:r Y rad tu:b
lo:z Sad ji 1 tu:1 taht
JL xa:l j1 Ya:z bank L aks
S ka:s ts:r hu:t L sa:m
S bard kum ri:h 1. 3eib
Wi ?ahl L hur su:s t ku:x
b ba:s sad L ein hab
S said Xo:f 3a:f kalb
S mo:t L u:d sath si:x
S zo:3 L iz Z- ru:h ad
su:d sab raf Th o:d
S varb ma3d L SaXs zi:r
eo:b S a:y bur3 ramz
|J9 wa:d S :x bar sasb
j) waz si:n bint bu~d
l eal3 jI da:r taqs dars
ia fi: 1 du:d ?i ard mur
ei ? ScaTr ward deif
zeid e qalb ii anf Li alf
Sat_ ri:S S haq sabt
S so:t y di:n e hub bat







52







Table 4-2 continued
1B 2B 3B 4B
xas fard beid 3id
3ad eo:m eo:r xeit
kaf bs:9 mo:z il
nus i xalf taxt hal
lams 3SIS i:d 2 ism
nafs T a:m harb arS
Sum zay Ja qalb self
ru:s ha3 J za:d j0 fu:l
S bu:q ssir Sar 3a:r
f: lo:h ss:r lIs fa:s
S Yard 3a:d i ab Sa:S
jS o:k jwm ha:r darb
har ruz o Yo:s karS
c tu:t 3 gaw xat -3 su:r
SSak 3u:9 T nu:r eule
ze:t U ri:h qo:s fa:r
S di:k J, ma:l daw dam
S do:r S ab u na:s 2uxt
hizb i ax dub had
bo:t ri:m lub no:m
2i uns xa:s t xo: hi:n
haj saf ham lo:n
j qird S uS Sarq fard
kanz su:f duS ki:s
U S:b mad T am 3 ild











Table 4-3 Logistic regression results for the full lists, the calculated slopes at the 50% and
the 20-80% levels, the threshold dB level, and difference of threshold levels
from the mean threshold


List
1
2
3
4

M
Minimum
Maximum
Range
SD


a
3.918
3.574
3.663
3.930

3.771
3.574
3.930
0.356
0.180


b
-0.185
-0.169
-0.172
-0.185

-0.178
-0.185
-0.169
0.017
0.009


Slope at
50%
4.614
4.212
4.295
4.629

4.438
4.212
4.629
0.416
0.215


Slope from
20-80%
3.994
3.646
3.718
4.007

3.841
3.646
4.007
0.361
0.186


Threshold
dB
21.225
21.214
21.321
21.226

21.246
21.214
21.321
0.107
0.050


dB
difference
-0.021
-0.033
0.074
-0.020

0.000
-0.033
0.074
0.107
0.050


Table 4-4 Logistic regression results for the half lists, the calculated slopes at the 50%
and the 20-80% levels, the threshold dB level, and difference of threshold
levels from the mean threshold
Slope at Slope from Threshold dB
Half Lists a b 50% 20-80% dB difference
1A 3.748 -0.177 4.426 3.832 21.170 -0.077
1B 4.104 -0.193 4.821 4.173 21.281 0.034
2A 3.969 -0.186 4.642 4.018 21.379 0.132
2B 3.249 -0.154 3.860 3.341 21.040 -0.206
3A 3.663 -0.171 4.285 3.709 21.370 0.123
3B 3.663 -0.172 4.305 3.726 21.272 0.025
4A 3.498 -0.164 4.093 3.543 21.363 0.117
4B 4.522 -0.214 5.361 4.640 21.087 -0.159

M 3.802 -0.1790 4.47 3.87 21.25 0.00
Minimum 3.249 -0.2144 3.86 3.34 21.04 -0.21
Maximum 4.522 -0.1544 5.36 4.64 21.38 0.13
Range 1.273 0.0600 1.50 1.30 0.34 0.34
SD 0.392 0.0187 0.47 0.40 0.13 0.13









Table 4-5 Selected characteristics of normal hearing participants


Participant
14
22
10
11
3
4
5
6
20
21


Mean
Standard Deviation
Range
Minimum
Maximum


Table 4-6 Selected characteristics of hard of hearing participants


Participant
8
16
18
2
7
1
15
17
19
23


Gender
F
M
M
F
M
F
M
M
F
M


Mean
Standard Deviation
Range
Minimum
Maximum


Age
6
6
6
8
8
9
9
9
9
9
7.9
1.4
3.0
6.0
9.0


Test Ear
Right
Left
Left
Right
Right
Right
Left
Left
Right
Left


PTA
63
48
62
32
52
32
58
60
53
42
50.2
11.8
31.7
31.7
63.3


Hearing
Loss Type
SNHL
SNHL
SNHL
Conductive
Mixed
SNHL
SNHL
SNHL
SNHL
SNHL


Gender
M
M
F
M
F
F
F
M
F
F


Test Ear
Right
Left
Right
Right
Left
Left
Left
Right
Right
Right


Age
6
6
7
7
8
8
8
8
8
9
7.5
1.0
3.0
6.0
9.0


PTA
5.0
10.0
8.3
8.3
3.3
1.7
10.0
5.0
6.7
5.0
6.3
2.8
8.3
1.7
10.0









Table 4-7 Word recognition scores in percent correct for each subject per talker-talker
and the average of scores by dialect.
Parti- Average
Group cipant J1 J2 J3 S1 S2 S3 Jordanian Saudi
1 1 0.94 0.91 0.97 0.94 1.00 0.97 0.94 0.97
1 2 0.94 0.94 0.97 0.85 0.88 0.97 0.95 0.90
1 8 0.58 0.30 0.42 0.45 0.24 0.42 0.43 0.37
1 15 0.82 0.76 0.67 0.82 0.73 0.61 0.75 0.72
1 16 0.70 0.70 0.45 0.61 0.61 0.55 0.62 0.59
1 17 0.70 0.52 0.67 0.70 0.61 0.73 0.63 0.68
1 18 0.58 0.64 0.52 0.79 0.79 0.64 0.58 0.74
1 19 0.70 0.48 0.82 0.76 0.61 0.82 0.67 0.73
1 23 0.97 0.64 0.85 0.85 0.94 0.94 0.82 0.91
1 7 0.67 0.39 0.67 0.70 0.52 0.76 0.58 0.66
2 6 0.94 0.55 0.76 0.82 0.73 0.85 0.75 0.80
2 5 0.82 0.73 0.82 0.79 0.73 0.79 0.79 0.77
2 4 0.85 0.61 0.79 0.67 0.79 0.97 0.75 0.81
2 3 0.70 0.64 0.64 0.79 0.82 0.88 0.66 0.83
2 10 0.88 0.79 0.58 0.91 0.67 0.52 0.75 0.70
2 11 0.79 0.82 0.42 0.85 0.67 0.82 0.68 0.78
2 22 0.82 0.70 0.55 0.88 0.85 0.76 0.69 0.83
2 14 0.94 0.97 0.88 0.97 0.91 0.97 0.93 0.95
2 20 0.82 0.58 0.73 0.88 0.64 0.91 0.71 0.81
2 21 0.85 0.82 0.73 0.82 0.82 0.94 0.80 0.86
Group 1 = hard of hearing, Group 2= normal hearing, J1= Jordanian talker 1, J2= Jordanian talker
2, J3= Jordanian talker 3, S1= Saudi talker 1, S2= Saudi talker 2, S3 = Saudi talker3


Table 4-8 Mixed ANOVA results comparing dialects for all participants' scores

df Mean Square F-test Sig.
Dialect 1 0.066 8.865 0.008
Talkers 2 0.091 6.181 0.005
Talker within dialect 2 0.027 5.153 0.011


Table 4-9 Paired Sample T-tests comparing dialectal difference in the hard of hearing
group and normal hearing group.

Group df T-test Sig. (2-tailed)
Hard of Hearing 9 -1.327 0.217
Normal Hearing 9 -2.923 0.017






56


Table 4-10 Paired Sample T-test results including all participants' scores


Pairs df T-test Sig. (2-tailed)
J1-J2 19 4.203 0.0001
J1-J3 19 3.802 0.001
J2-J3 19 -0.496 0.625
S1-S2 19 2.658 0.016
S1-S3 19 0.047 0.963
S2-S3 19 -2.183 0.042














CHAPTER 5
DISCUSSION

The primary goal of the present study was to create four equivalent lists of words

for clinical assessment of speech perception in Jordanian Arabic speaking children. There

are no published word lists developed for the specific purpose of clinical speech

audiometry (word recognition) for Jordanian Arabic speaking populations. Although

other speech recognition materials have been developed and published over the past 3

decades, no study has evaluated the feasibility of applying Arabic speech audiometry

materials among different countries or different Arabic language dialects (e.g., the use of

Egyptian Arabic word lists in Jordan). In addition, recorded materials are not widely

distributed and, thus, not commonly used in audiology clinics within Arabic countries.

Four lists of 50 words were developed in the present study. The lists were not

significantly different in their psychometric functions. All had the same shape of curve

and very similar slopes (mean 4.4 %/dB, S.D. 0.215) and threshold points (mean 21.25

dB, S.D. 0.05), properties that are appropriate for clinical use as speech recognition

measures. The word lists developed in the present study were comparable in slope and

curve shape to word lists published in English (Haskins, 1949; Wilson & Oyler, 1997),

Korean (Harris et al. 2003a), and other words lists published in Arabic (Alusi et al. 1974).

The point of threshold (50% correct word recognition) was comparable to the Arabic

adult lists reported by Alusi et al. (1974). The point of threshold for the Arabic word lists

in the present study differed from other languages, suggesting the likelihood of a

language specific feature. The range of thresholds for the word recognition materials in









the present study (from 2.3 to 22.5 dB HL), confirm the appropriateness of the materials

as measures of word recognition ability.

There was no attempt in the present study to develop words lists that were

phonetically balanced. Carhart (1965) stated that the phonetically balanced CNC lists and

W-22 lists approximated the phonetic balance of every day spoken English but cannot be

true representation of everyday phonetic balance. The effect of familiarity of words on

speech perception performance is greater than the phonetic balance of the word lists.

Owens (1961) studied the effect of word familiarity on word recognition. He found that

listeners were more likely to make errors on less familiar words and, when they made the

errors, the listener's response was more likely to be a familiar word. Martin (2000) stated

that phonetic balance is not the only, or the main, factor in word list equivalence.

Therefore, an objective of the present study was to utilize familiar words with a balanced

range of difficulty between lists. In the present study, the psychometric curves of the full

lists and half lists were very similar. Statistical analyses of the full lists (X2 (3) = 1.67, p

=0.64) and half lists (X 2 (7) = 8.25, p = 0.31) showed no significant difference between

word lists and half lists despite the fact that no effort was made to account for phonetic

balance.

In comparison to the speech materials reported previously (Alusi et al., 1974;

Ashoor & Prochazka, 1982, and 1985) the recordings developed in the present study can

easily be applied clinically by audiologists in Arabic-speaking patient populations with a

CD player and distributed to audiologists throughout the Arab countries. However, the

issue of dialectal differences requires further investigation. Specifically, the possible









effect of dialect on word recognition performance should be assesses for children in

different Arabic speaking countries.

It was hypothesized that participants in this study would perform better when

listening to words spoken in Jordanian versus Saudi dialect and, in addition, that the

effect would be similar for both normal hearing and hard of hearing children. Dialect

appeared to exert an effect on word recognition ability of normal hearing Jordanian

Arabic speaking children. Results reported herein confirmed that word recognition scores

were higher when normal hearing children listened to words spoken in Saudi dialect. In

contrast, no significant effect for Jordanian versus Saudi dialect was found in word

recognition ability of children with mild to moderately severe hearing loss. In general,

word recognition performance was lower and more variable among hard of hearing

children than for normal hearing children, independent of dialect. However, the higher

word recognition scores for the Saudi dialect were persistent in both hard of hearing and

normal hearing groups. Perhaps because each list was limited to 33 words that were

highly familiar to the target population (i.e., Jordanian Arabic speaking children age 6 to

9 years). In addition, the words were spoken similarly in standard Arabic and in both

Jordanian and Saudi dialect, with the speaker remaining as the only potential factor

affecting performance. Another consideration affecting data reported in the present study

is the small sample size. Only 10 participants were included in each group, with three

speakers in each dialect. Although the power analysis requirement was met by statistical

criteria (power .95), a larger sample size would probably yield more representative

results.









The results from the present study are not consistent with Alusi's (1974) contention

that his word lists could be used in all Arab countries since the word lists were derived

from standard Arabic. There are several clear differences between the present study and

the study reported by Alusi. For example, the subjects in Alusi's study were educated

adults, whereas subjects in the present study were children. Also, the sample size (N=

17) in Alusi's study was inadequate for comparison of performance from different

dialectal backgrounds. Furthermore, the speaker in Alusi's study used Baghdad general

dialect, whereas in the present study the speaker used general Jordanian and Saudi

dialects. These differences among studies highlight the need for further investigation to

compare word recognition performance with the speech materials from the present study

with children from other Arabic speaking countries, particularly in relation to

psychometric functions and the familiarity of words.

Further clinical investigations should be conducted utilizing the word lists

developed in this study. A first step would be to collect data for standardization of the

word lists. Several variables must be considered in the collection of standardization data.

Data should be collected from a large sample of the target population that is

representative of rural as well as urban Jordan. In addition, the sample should include

participants of all socioeconomic groups, as well as children who display normal and

impaired hearing abilities. Another step is to establish test reliability. Factors to be

considered are lists versus half list reliability, children who display normal and impaired

hearing, as well as gender (males and females).

Although the present study focused on word recognition, there is a need to develop

other speech diagnostic materials that for evaluation of speech perception abilities of









younger and older pediatric populations, as well as populations with different auditory

problems, such as auditory processing disorder. Diagnostic speech audiometry materials

that are needed clinically include:

* A "word recognition by picture identification test" for children younger than 6
years. The words used in the present study could be assessed for familiarity with
younger children and on the basis of the availability of picture representations of
the words. A picture-pointing word recognition measure would be very helpful in
diagnosing and monitoring benefit of hearing aids and intervention in younger
children.

* Material for central auditory processing diagnosis, such as filtered words, figure
ground tests, and dichotic tests could be developed based on the present word lists.
The fact that the words are digitally recorded facilitates the process of developing
additional tests.

* Speech reception tests for estimation of threshold are lacking for Arabic-speaking
children and adults. The procedures used to develop the speech materials in the
present study have shown to be effective as in other studies by Harris et al. (2003b)
and Nissen et al. (2005b), and can be duplicated.

Results from the present study will play a significant role in the development of

additional Arabic speech audiometry materials and in improving audiological services

provided in Jordan. The technique for recording the speech materials and the data

collection conditions in the present study render these word lists to be of high quality.

The speaker was carefully chosen to have clear speech and representative of general

Jordanian dialect. The words were selected based on familiarity to children who are

native speakers of Arabic in Jordan. The data were collected using headphones, the most

common method for presenting speech signals in this age group. And, finally, the words

were digitally recorded under excellent recording sound conditions providing a very clear

signal, and materials that can be easily duplicated for the production of multiple audio

copies of the word lists and half lists The word recognition materials developed in the

present study are likely to be of considerable value in providing diagnostic services and






62


intervention to the pediatric hard of hearing population in Jordan. In addition, these word

recognition materials will contribute to the rapid development of diagnostic speech

audiometry materials that are currently lacking in Arabic countries.






63






Table 5-1 Descriptive statistics for Hard of Hearing and Normal Hearing based on dialect

Hard of Hearing Normal Hearing
Jordanian Saudi Jordanian Saudi
Mean 0.69 0.72 0.74 0.81
Standard Deviation 0.18 0.18 0.13 0.10
Range 0.66 0.76 0.55 0.46
Min 0.30 0.24 0.42 0.52
Max 0.97 1.00 0.97 0.97














APPENDIX A
PARTICIPANT RECRUITMENT


The initial pool of potential participants with normal hearing included 18

children; five of them were excluded after the initial pure tone audiometric testing due to

their elevated thresholds that have exceeded the inclusion criteria of 15dB HL at least at

one frequency. Further testing using bone conduction audiometry revealed mild

conductive hearing loss. The children's caregivers were counseled and retest was

recommended. In one case mild sensorineural hearing loss was diagnosed. This

participant was included in the hard of hearing group. Data from 2 participants was

excluded due to technical problems during data collection, which rendered results

unreliable. Among the 20 potential participants in the hard of hearing group seven

participants were excluded for having a severe hearing loss, based on the screening, and

some had developmental delays that they were not able to understand the task.














APPENDIX B
SPEECH MATERIAL SELECTION


The words in this study were chosen from three different sources: a word

frequency lists provided by Landau (1959) which was derived from daily newspapers

from Lebanon, Egypt, Iraq, and Palestine, and modem prose; lists of Arabic key words

for learners of everyday Arabic as a second language (Quitregard, 1994); and from

children's story books.

To achieve Egen's (1948) criteria for word selection, the researcher selected

words that fit one of the following possible monosyllabic word structures in Arabic: CV,

CVC, CVCC, CVVC, and CVV. According to Altoma (1969) these monosyllabic word

structures are shared by colloquial and standard Arabic. As mentioned above, words were

selected from a range of written material. To ensure that selected words are representative

of spoken Arabic, the procedure employed by Alusi (1974) and Ashoor and Prochazka

(1982 and 1985) were followed. Words that have similar forms in colloquial and standard

Arabic were selected. This step can be supported by Altoma's (1969) finding that 83.5%

of the words in a colloquial form have shared origins with words in standard Arabic.

Words from Landau (1959) and Quitregard (1994) and a list of (135 words) that are

familiar to first, second and third grade Jordanian Arabic speaking students (Abdulhaq,

unpublished) were used. Abdulhaq's familiar word list is composed of monosyllabic

words chosen from children's story books and Jordanian daily newspaper articles, and

rated by 320 first, second, and third grade students as familiar or unfamiliar. The words









were read aloud to the children, they were asked to raise their hand if they knew the

meaning of the word and were able to put it in a meaningful sentence. To ensure the

accuracy of their response, Abdulhaq started with three trial words that were familiar to

children (window, door, and table), randomly asked children to put words in a sentence,

in addition unfamiliar words (from old classical text) and nonsense words were included.

The initial list included 210 words. One hundred and thirty five words were rated as

familiar at least by 80% of the children. Additional words from Landau and Quitregard

then were added. The researcher used three different procedures to ensure the familiarity

of words chosen for the word lists. Monosyllabic words were matched between the word

lists from Landau (1959) and Quitregard (1994). Words that occurred in both lists were

considered for further familiarity testing. Second the matching words and Abdulhaq's list

were rated as familiar or unfamiliar by four Jordanian teachers of first, second and third

grade. Finally teachers' ratings were compared to the children's ratings for consistency.

Out of these lists, the top 304 words rated as highly familiar by students and teachers

were selected for recording.














APPENDIX C
LOGISTIC REGRESSION ANALYSIS


Logistic regression analysis calculates the probability of a response based on the

independent variable, in this study, the stimulus intensity level. The logistic regression

analysis provides derived variables that are key parameters for the regression line that

may be tested and compared between conditions (in this study, the word lists). The

results from the logistic regression analysis, a, the logistic regression intercept and b, the

logistic regression slope, were used to calculate the probability of correct identification of

the words at each intensity level. Based on the model described by Nissen et al. (2005a)

in the description of performance on the word list in terms of logistic regression, the

following modified equation was used to calculate the percent correct:



Equation (1P=(1- exp(a) x100
1 + exp(a + b x i)

Where P is the percent correct at an intensity level, a is the regression intercept, b is the

regression slope, and i is the intensity level in dB HL. Using the regression intercept,

slope and intensity levels in equation 1 made it possible to predict the percent correct at

any given intensity level.















APPENDIX D
LIST OF WORDS FOR DIALECT COMPARISON


Meaning
Brother
Door
Egg
Dress
Bull
Grandparent
Pilgirim
Thread
Bare
Blood
Rooster
Rice
Feather
Market
Sword
Tea
Thornes


Word












0&


ju


eJ
di


Meaning
Class
Bird
Nest
Stick
Eye
Uncle (paternal)
Mouth
Elephant
Cottage
Thief
Board
Almonds
Banana
Fier
Sleep
Hand


Word


Jt

61
4;


a5
c.3
40


Jj
(-AiJ~
iksi















APPENDIX E
HALF WORD LISTS AND MEANING


1A 2A 3A 4A
Good Gj God yA Meadow -- Sun
-4 Elbow c Pilgrim J# Horses c Crown
i4 Home Jj Button > Fairies j Figs
UI-j Head Wrap J.L Situation j Well
j4 Market j-1 Fire Y) Answer .-j Brick
-j3 Almonds Tight J.-. Length L Under
JL& Uncle (maternal) j.t Gas 4 Bank a Opposite
LcA Cup Bird -j Whale ^ Poisonous
ay Cold Sleeve ej Wind ? Pocket
AI Family j Free o-a Chick jS Cottage
-uA Bus D Dam Eye Seeds
Hunt i-.j Fear -4 Dry uS Dog
c4 Death Stick Roof Skewer
j3 Husband j Prosperity CjJ Spirit Count
,, Blacks Poured (adj) .ij Shelf ;uji Tub
j West Glamour u2 Person Jug
-3 Dress Tea Tower jij Symbol
-j Valley Shaikh Wilderness Difficult
j3 Geese a China c1 Girl Distance
C Snow j. House u! weather cUJ Lesson
J Elephant -j Turn u J Earth Bitter
SSomething j Hair Jj9 Roses -- Guest
0j Zaid (name) 4 Heart _i1 Nose Thousand
.L Beach c J Feather J Right c. Saturday
cj Voice Religion Love Ducks





1B 2B
oU- Lettuce u]ja Requirement
S Serious e Garlic
iS Palm Sell
Half Behind
, I Touch Armey
Same t- Year
Mother o5 Outfit
cj-J Heads Pilgrimage
L. Trumpet Traffic
S Board Secret
ije Width Serious
. Thorns Day
J; Hot Rice
j Berry y. Space
Check j-. Hunger
-.uj Oil -.J) Rural
S Rooster JU Money
Jj) Turn Youngman
S Party Brother
1. Boot Reem (name)
uL- friendliness Private
L" Neighborhood L- Class
S Monkey LuL Nest
S Treasure L-i- Wool
Gray hair Tide


3B 4B
u Egg > Grandfather
jj Bull String
jjy Banana 3A Shadow
S Bed JL Solution
Holiday Name
j,- War j Throne
- Turn over Sword
S Increase Jjs Lima bean
Evil j? Neighbor
L Thief Axe
S Father t Gauze
j> Spicy .) Way
ui- Dive jS Belly
Line j.' Fence
9j Light Third
u9~. Arch jl Mouse
3'-! Lamp 5 Blood
L~- People 1i Sister
La Bear Limit
L4 Core Sleep
-j& Plum j When
S Worry jl Color
& East aj Gun
4; Shower o4 Bag
Uncle (paternal) 4 Skin















LIST OF REFERENCES

Abd-el-Jawad, H. (1987). Cross-dialectal variation in Arabic: competing prestigious
forms. Language in Society, 16; 359-368.

Al-Kahtani, A. (1997). The "Problem" of Diglossia in the Arab world: an attitudinal
study of modern standard Arabic and the Arabic dialects. Al-Arabiyya, 30;1-30.

Al- Masri, M. (2003). Early Detection and Habilitation ofHearing Loss, MEHA Project
4. Amman, Jordan.

Altoma, S. (1969). The Problem of Diglossia in Arabic, a comparative study of classical
and Iraqi Arabic. Harvard Middle Eastern Monographs XXI.

Alusi, H.A., Hinchcliffe, R., Ingham, B., Knight, J.J., & North C. (1974). Arabic speech
audiometry. Audiology, 13; 212-130.

American National Standards Institute (2004). Specification for Audiometers. ANSI S3.6-
2004. New York: ANSI.

Ashoor, A.A. & Prochazka, T. Jr. (1982). Saudi Arabic speech audiometry. Audiology,
21; 493-508.

Ashoor, A.A. & Prochazka, T. Jr. (1985). Saudi Arabic speech audiometry for children.
British Journal ofAudiology, 19; 229-238.

Arlinger, S. (2001). How to assess outcome of hearing aid fitting in children.
Scandinavian Audiology, 30 Suppl. 53; 68-72.

Beattie, R.C., Edgerton, B.J., & Svihovec, D.V. (1977). A comparison of the Auditec of
St. Louis Cassette recordings of NU-6 and CID W-22 on a normal-hearing
population. Journal of Speech and Hearing Disorders, 42(1). 60-64.

Beattie, R.C., Svihovec, D.V., & Edgerton, B.J. (1975). Relative intelligibility of the CID
spondees as presented via monitored live voice. Journal of Speech and Hearing
Disorders, 40(1); 84-91.

Beatti, R.C. & Warren, V. (1983). Slope characteristics of CID W-22 word functions in
elderly hearing -impaired listeners. Journal of Speech and Hearing Disorder,
4;119-127.









Blamey, P., Sarant, J., Paatsch, L., Barry, J., Bow, C., Wales, R., Wright, M., Psarros, C.,
Rattigan, K. & Tooher, R. (2001). Relationship among speech perception,
production, language, hearing loss, and age in children with impaired hearing.
Journal of Speech, Language, and Hearing Research, 44(2); 264-283.

Carhart, R. (1951). Basic principles of speech audiometry. Acta Oto-Laryngologica, 40;
62-71

Carhart, R. (1965). Problems in the measurement of speech discrimination. Archives of
Otolaryngology, 82; 253-260.

Cramer, K. & Erber, N.P. (1974). A spondee recognition test for young hearing-impaired
children. Journal of speech and hearing Disorders, 39; 304-311.

Curry, E.T. (1949). A study of the relationship between speech thresholds and
audiometric resluts in perception deafness. Journal of Speech andHearing
Disorders, 14; 104-110.

Davis, H. (1948). The Articulation area and the social adequacy index for hearing.
Laryngoscope, 58; 761-778.

Doyne, M. P., & Steer M.D. (1951). Studies in speech reception testing. Journal of
Speech and Hearing Disorders, 16; 132-139.

Dubno, J.R., Lee, F.S., Klein, A.J. & Matthews, L.J. (1995). Confidence limits for
maximum word-recognition score. Journal of Speech and Hearing Research, 38;
490-502.

Egan, James (1948). Articulation testing methods. Laryngoscope, 58; 955-991.

Elliot, L. (1979). Performance of children age 9 to 17 years on a test of speech
intelligibility in noise using sentence material with controlled word predictability.
Journal of the Acoustical Society ofAmerica, 66; 651 -653.

Elliot, L, Connors S., Kille, E., Levin, S., Ball, K. & Katz, D. (1979). Children's
understanding of monosyllabic nouns in quiet and in noise. Journal of the
Acoustical Society of America, 66; 12-21.

Elliot, L. & Katz, D.R. (1980). Northwestern University Children's Perception of Speech
(NU-CHIPS). AUDITEC of St. Louis, St. Louis, Missouri.

Erber, N.P. (1974). Pure-tone thresholds and word-recognition abilities of hearing-
impaired children. Journal of Speech and Hearing Research, 17; 194-202

Erber, N.P. (1980). Use of the auditory numbers test to evaluate speech perception
abilities of hearing-impaired children. Journal of Speech and Hearing Disorders,
45; 527-532.









Erber, N.P. & Alencewicz,C. (1976). Audiologic evaluation of deaf children. Journal of
Speech and Hearing Disorders, 41; 256-267.

Fatihi, A.R. (2001). Aspects of Arabic Phonology. Kalinga Publication, Delhi.

Ferguson, C. (1959). Diglossia. Word, 15; 325-340

Fletcher, H. (1958). Speech andHearing in Communication. D. van Nostrand Company,
Inc., Princeton, New Jersy.

Fletcher, H. (1995). The ASA edition of speech and hearing in communication. Ed.
Allen, J.B. Acoustical Society of America, Woodbury, NY.

Fletcher, H. & Steinberg, J.C. (1929). Articulation testing methods. Bell System
Technical Journal, 8; 806-854.

Goldman,R., Fristoe, M. & Woodcock, R. (1970). Goldman-Fristoe-Woodcock Test of
Auditory Discrimination. American Guidance Service, Inc. USA.

Harris, R., Kim, E. & Eggett, D. (2003a). Psychometrically equivalent Korean
monosyllabic speech discrimination materials spoken by male and female talkers.
Korean Journal of Communication Disorders, 8 (1); 217-243.

Harris, R., Kim, E. & Eggett, D. (2003b). Psychometrically equivalent Korean bisyllabic
words spoken by male and female talkers. Korean Journal of Communication
Disorders, 8 (1); 244-270.

Haskins, H.A. (1949). A phonetically Balanced Test of Speech Discrimination for
Children. Master's thesis. Northwestern University, Evanston, IL.

Hirsh, I.J., Davis, H., Silverman, R., Reynolds, E.G., Eldert,E., & Benson, R. (1952).
Development of materials for speech audiometry. Journal of Speech and Hearing
Disorder, 17;321-337.

Hirsh, I. J., Reynolds, E.G., Joseph, M. (1954). Intelligibility of different speech
materials. The Journal of the Acoustical Society ofAmerica, 26(4); 530-538.

Hood, J.D. & Poole, J.P. (1980). Influence of the speaker and other factors affecting
speech intelligibility. Audiology, 19; 434-455.

Ibrahim, Z. (2000). Myths about Arabic revisited. Al-Arabiyya, 33; 13-28.

The International Kindergarten Union, Child Study Committee (1928). A Study of
Vocabulary of Children Before Entering the First Grade. Washington, D.C.

Jauhiainen, T. (2001). Hearing impaired children in developing countries. Paper
presented at the 3rd International Congress of Pediatric Audiology, 9-12 May 2000,
Copenhagen. Scandinavian Audiology, 30 Suppl 53; 83-84.









Jerger, S., Lewis, S., Hawkins, J., & Jerger, J. (1980). Pediatric speech intelligibility test.
I. Generation of test materials. International Journal of Pediatric
Otorhynolaryngology, 2; 217-230.

Kishon-Robin, L. & Rosenhouse, J. (2000). Speech perception test for Arabic-speaking
children. Audiology, 39; 269-277.

Kreul, E.J., Bell, D.W. & Nixon, J.C. (1969). Factors affecting speech discrimination test
difficulty. Journal of Speech and Hearing Research, 12; 281-287.

Landau, J. (1959). A Word Count of Modern Arabic Prose. New York: American Council
of Learned Societies in cooperation with the school of education at the Hebrew
University, Jerusalem.

Larson, G.W., Petersen, B., & Jacquot, W.S. (1974). Use of Northwestern University
Test No.6 for speech discrimination testing with children. The Journal ofAuditory
Research, 14; 287-292.

Laukli, E. & Fjermedal, 0. (1990). Reproducibility of Hearing Threshold Measurements,
supplementary Data on Bone-conduction and speech audiometry. Scandinavian
Audiology, 19; 187-190.

Martin, F.N., Champlin, C.A., & Perez, D.D. (2000). The question of phonetic balance in
word recognition testing. Journal of the American Academy ofAudiology, 11; 489-
493.

Messouak, K (1956). Audiometrie vocale en Arabic maghrebin. Les cahiers de la CFA,
No. 4. Compagnie Francaise d'Audiologie, Paris.

Meyer, T.A & Pisoni, D.B (1999). Some computational analyses of the PBK test: effect
of frequency and lexical density on spoken word recognition. Ear and Heari,,g. 20
(4); 363-371.

Niessen, S.L, Harris, R.W, Jennings, L.J., Eggett, D.L., & Buck, H. (2005a).
Psychometrically equivalent Mandarin bisyllabic speech discrimination materials
spoken by male and female talkers. International Journal ofAudiology, 44(7); 379-
390.

Niessen, S.L, Harris, R.W, Jennings, L.J., Eggett, D.L., & Buck, H. (2005b).
Psychometrically equivalent trisyllabic words for speech reception threshold testing
in Mandarin. International Journal ofAudiology, 44(7); 391-399.

Oster, A.M. (2002). The relationship between residual hearing and speech intelligibility-
Is there a measure that could predict a prelingually profoundly deaf child's ability
to develop intelligible speech? Speech, Music and Hearing. Quarterly Progress
and Status Report (TMH-QPSR), 43; 5156.









Owens, E. (1961). Intelligibility of words varying in familiarity. Journal of Speech and
Hearing Research, 4(2); 113-129.

Palmer, J. M. (1955). The effect of speaker differences on the intelligibility of
phonetically balanced word lists. Journal of Speech and Hearing Disorders, 20;
192-195.

Quitregard, D. (1994). Arabic Key Words, the basic 2000 word vocabulary arranged by
frequency in a hundred units. The Oleander Press, New York.

Ross, M. & Lerman, J. (1971). A picture identification test for hearing-impaired children.
Journal of Speech and Hearing Research, 13; 44-53.

Sanderson-Leepa, M. & Rintelmann, W.F. (1976). Articulation functions and test-retest
performance of normal-hearing children on three speech discrimination tests: WIPI,
PBK-50, and NU Auditory test No. 6. Journal of Speech and Hearing Disorders,
41(4); 503-519.

Saudi Speech Pathology and Audiology Association, SSPAA (2004).
http://www.sspaa.org, last accessed October 2004.

Schwartz, A & Goldman, R. (1974). Variables in fluencing performance on speech-sound
discrimination tests. Journal of Speech and Hearing Research, 17; 25-32.

Siegenthaler, B.M, Pearson, J. & Lezak, R.J. (1954). A speech reception threshold test for
children. Journal of Speech and Hearing Disorder, 19; 360-366.

Smith, A.W. (2001). WHO activities for prevention of deafness and hearing impairment
in children. Paper presented at the 3rd International Congress of Pediatric
Audiology, 9-12 May 2000, Copenhagen. Scandinavian Audiology, 30 Suppl 53;
93-100.

Soliman S (1976). Speech discrimination audiometry using Arabic phonetically balanced
words. Ain .,\,i %u Medical Journal, 27; 27-30.

Soliman, S., Abd El-Hady, M., Saad, M., & Kolkaila, E. (1987a). Development and study
of Arabic word intelligibility by picture identification (WIPI) in normal hearing
Egyptian Children. Scientific Journal ofAl-Azhar University, Medical Faculty
(girls), 6.

Soliman, S., Fathalaa, A., & El-Mahalawi, T. (1987b). Simple speech test as a predictor
of speech reception threshold (SRT) in preschool children. Proceedings ofAnnals
of the Xth Annual Medical Congress ofAid .\lin University, Cairo, Egypt.

Tillman, T.W., & Carhart, R. (1966). An expanded test for speech discrimination
utilizing CNC monosyllabic words (Northwestern University Auditory Test No.6).
Technical Report No. SAM-TR-66-55, USAF School ofAerospace Medicine,
Brooks Air Force Base, Texas.









Thornton, A. & Raffin, M. (1978). Speech discrimination scores modeled as a binominal
variable. Journal of Speech and Hearing Research 21; 507-518.

Wilson, R.H. & Carter, A.S. (2001). Relation between slopes of word recognition
psychometric functions and homogeneity of the stimulus materials. Journal of the
American Academy ofAudiology, 12; 7-14.

Wilson, R.H. & Oyler, A.L. (1997). Psychometric function for the CID-W22 and NU
Auditory Test No.6. materials spoken by the same speaker. Ear and Heari,,g.
18(5); 430-433.

Wilson, R.H. & Strouse, A. (1999). Psychometrically equivalent spondaic words spoken
by a female speaker. Journal of Speech, Language, and Hearing Research, 42;
1336-1346.

Wilson, R.H., Zizz, C., Shanks, J., & Causey, G. (1990). Normative data in quiet,
narrowband noise, and competing message for Northwestern university auditory
test No.6 by a female speaker. Journal of Speech and Hearing Disorders, 55; 771-
778.

World Health Organization (1998). Report of the International Workshop on Primary
Care and Hearing Care. Cape Town; South Africa March 12-14.

Young, J.R., Dudley, B., & Gunter, M.B. (1982). Thresholds and psychometric functions
of the individual spondaic words. Journal of Speech and Hearing Research, 25;
586-593.

Zakrzewski, A, Jassem, W., Pruszewicz, A., & Obrebowski, A. (1976). Speech
audiometry for children and subjective probability of Polish words. Audiology, 15;
228-231.















BIOGRAPHICAL SKETCH

Nadia Abdulhaq was born and raised in Nablus, Palestine. The third of six

children, she graduated from high school in 1990 and enrolled at Birzeit University in

Birzeit, Palestine. She obtained her Bachelor of Arts degree in English language and

literature with a minor in translation, in June, 1995. She then went on to receive her

Master of Arts degree in speech-language pathology from the University of Jordan in

Amman, Jordan, in June, 1997.

After graduating from the University of Jordan, she accepted the position of

speech pathologist in the Department of Rehabilitation at the Palestine Red Crescent

Society. Her job description included establishing three speech therapy units in three

different cities across the West Bank of Palestine and training six rehabilitation workers

as speech therapy assistants. Her work focused on children with hearing loss. After two

years of work and achievements, she decided to return to the academic world and pursue

further knowledge to be able to better serve the hard of hearing and deaf community of

Palestine.

In 2000 she received the Fulbright scholarship to start her doctoral studies at the

University of Florida in Gainesville, Florida. During her five year study she was an active

member of the international student body initiating projects such as the International

Student Speakers Bureau and the international student welcome information table. She

received numerous scholarships and recognition, O. Ruth McQown Scholarship, Grinter

Fellowship, Gibbson Dissertation Fellowship, and the Outstanding Academic






78


Achievement award. She will graduate in December 2005 with a Ph.D. in communication

sciences and disorders and Doctor of Audiology degrees. Upon her graduation, she will

return to Palestine to start a speech and hearing program for the hard of hearing, new

born hearing screening program, and an audiology clinic.