USF Libraries
USF Digital Collections

Clear speech effects for vowels produced by monolingual and bilingual talkers

MISSING IMAGE

Material Information

Title:
Clear speech effects for vowels produced by monolingual and bilingual talkers
Physical Description:
Book
Language:
English
Creator:
DeMasi, Teresa
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Intelligibility
Perception
Production
Second language
Communication
Dissertations, Academic -- Speech Language Pathology -- Masters -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: 'Clear speech' is a speaking style that talkers often employ when they know they may have trouble being understood, as when speaking in noise or to a person with a hearing loss. When 'clear speech' produced by native talkers is presented in noise to native listeners, it has been shown to be about 10-15 percentage points more intelligible, on average, than normally produced speech. Recent research has shown that bilingual listeners may experience a smaller intelligibility benefit than monolingual listeners from 'clear speech' produced by monolingual talkers. The present study compares the ability of monolingual and bilingual talkers to produce this clear speech intelligibility benefit. The present study investigates the hypothesis that bilinguals may produce a smaller intelligibility benefit than monolinguals when asked to speak clearly.Three groups of talkers were recorded: 13 monolingual native English speakers, 22 'early' Spanish-English bilinguals, with an age of onset of learning English (AOL) of 12 or earlier, and 14 later Spanish-English bilinguals, with an AOL of 15 or later. Talkers produced the target words "bead, bid, bayed, bed, bad" and "bod" in both clear and conversational speech styles. Two repetitions of each word were mixed with noise and presented to monolingual English-speaking listeners across two days of testing. Both monolingual and early bilingual talkers showed a similar degree of clear speech benefit in noise (about 5.5%). Later bilinguals were the least intelligible overall and showed a smaller overall clear speech benefit in noise. Surprisingly, early bilinguals were significantly more intelligible than monolinguals in both speaking conditions (by about 6.5%).For the later bilinguals only, performance was significantly worse for one target word ("bid") in the clear speech condition than in the normal speech condition. These data suggest that later bilinguals, but not early bilinguals, may experience a relative disadvantage when speaking in noise, due to a reduced ability to improve intelligibility by speaking more clearly. Therefore, these persons may benefit from communication strategies or accent reduction programs designed to increase their ability to make themselves understood in difficult speaking environments.
Thesis:
Thesis (M.S.)--University of South Florida, 2007.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Teresa DeMasi.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 87 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001920234
oclc - 187983924
usfldc doi - E14-SFE0002110
usfldc handle - e14.2110
System ID:
SFS0026428:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001920234
003 fts
005 20080107135823.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 080107s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002110
035
(OCoLC)187983924
040
FHM
c FHM
049
FHMM
090
RC423 (ONLINE)
1 100
DeMasi, Teresa.
0 245
Clear speech effects for vowels produced by monolingual and bilingual talkers
h [electronic resource] /
by Teresa DeMasi.
260
[Tampa, Fla.] :
b University of South Florida,
2007.
520
ABSTRACT: 'Clear speech' is a speaking style that talkers often employ when they know they may have trouble being understood, as when speaking in noise or to a person with a hearing loss. When 'clear speech' produced by native talkers is presented in noise to native listeners, it has been shown to be about 10-15 percentage points more intelligible, on average, than normally produced speech. Recent research has shown that bilingual listeners may experience a smaller intelligibility benefit than monolingual listeners from 'clear speech' produced by monolingual talkers. The present study compares the ability of monolingual and bilingual talkers to produce this clear speech intelligibility benefit. The present study investigates the hypothesis that bilinguals may produce a smaller intelligibility benefit than monolinguals when asked to speak clearly.Three groups of talkers were recorded: 13 monolingual native English speakers, 22 'early' Spanish-English bilinguals, with an age of onset of learning English (AOL) of 12 or earlier, and 14 later Spanish-English bilinguals, with an AOL of 15 or later. Talkers produced the target words "bead, bid, bayed, bed, bad" and "bod" in both clear and conversational speech styles. Two repetitions of each word were mixed with noise and presented to monolingual English-speaking listeners across two days of testing. Both monolingual and early bilingual talkers showed a similar degree of clear speech benefit in noise (about 5.5%). Later bilinguals were the least intelligible overall and showed a smaller overall clear speech benefit in noise. Surprisingly, early bilinguals were significantly more intelligible than monolinguals in both speaking conditions (by about 6.5%).For the later bilinguals only, performance was significantly worse for one target word ("bid") in the clear speech condition than in the normal speech condition. These data suggest that later bilinguals, but not early bilinguals, may experience a relative disadvantage when speaking in noise, due to a reduced ability to improve intelligibility by speaking more clearly. Therefore, these persons may benefit from communication strategies or accent reduction programs designed to increase their ability to make themselves understood in difficult speaking environments.
502
Thesis (M.S.)--University of South Florida, 2007.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 87 pages.
590
Advisor: Catherine L. Rogers, Ph.D.
653
Intelligibility.
Perception.
Production.
Second language.
Communication.
690
Dissertations, Academic
z USF
x Speech Language Pathology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2110



PAGE 1

Clear Speech Effects for Vowels Produ ced by Monolingual and Bilingual Talkers by Teresa DeMasi A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science Department of Communicati on Sciences and Disorders College of Arts and Sciences University of South Florida Major Professor: Cath erine L. Rogers, Ph.D. Jean Krause, Ph.D. Stefan Frisch, Ph.D. Date of Approval: July 19, 2007 Keywords: intelligibility, perception, production, second language, communication Copyright 2007, Teresa DeMasi

PAGE 2

Dedication I dedicate this thesis to a nyone who ever thought they could not achieve the impossible. If you want something badly enough, you truly can accomplish anything you put your mind to.

PAGE 3

Acknowledgements The research in this thesis was supported by NIH-NIDCD grant #5R03 DC005561 to Catherine L. Rogers, P h.D. and by a CAS travel award. This thesis would not have b een possible if it weren’t for a number of very special people in my life. First and foremost, I would like to thank Dr. Catherine Rogers. Besides being my advisor and professor, you have also been my mentor and my fr iend. You’ve taught me to have confidence in myself and you’ve made me realize that it is possible to accomplish a difficult task, despite the many ch allenges that come along the way. I truly admire you for your patience, your understand ing, and your dedication to your work. You’ve inspired me to consider pursuing rese arch in the future. You’ve always been there for me, especially through hard times and I would just like to say THANK YOU!!! I would also like to than k my committee members, Dr. Jean Krause and Dr. Stefan Frisch. Thank you for your support in wr iting this thesis and for taking the time to read my revisions at the last minute! The many helpful comments and suggesti ons of Dr. Diane Kewley-Port are gratefully acknowledged, as well. Next, I would like to thank my lab partners, Merete M ller Glassbrenner and Michelle Bianchi. Merete…you were there from the first day I stepped foot in the lab and you have always been there to ad dress my questions and concerns. Michelle…you’ve been my partner in crime and you were there when I thought I couldn’t make it on my own. Thank you both for bei ng my friends and helping me get through life when being far away from home. I coul d not have gotten thr ough those endless hours of editing and acoustic analys is without you both…it’s over!

PAGE 4

Next, I would like to thank my family. Mom and Dad…thank you for your unconditional love and support despite my decision to pursue my graduate degree 1200 miles away from home. Philip…thank you for being there as my little brother. Grandma…thank you for always thinking of me and sending me letters even though I didn’t always write back. In case I haven’t sa id this already, I love each and every one of you. Next, I would like to thank my two best friends. Diana…thank you for always listening to me complain about how much homework I have and for yelling at me when procrastination takes over. Yasmin…I know you’ve been through di fficult times while I was away at school. Thank you for being so strong through it all…I truly admire you for that. Thank you both for being there for me even when I don’t return phone calls! Thank you to all of my professors and clini cal supervisors at USF. You have provided me with knowledge and skills that I can take with me throughout my future endeavors. Thank you to all of my clients…you are the reason why I chose this profession. Last, but not least, I would like to extend a huge THANK YOU to the class of 2007. It’s been a long road, but…WE DID IT!!!

PAGE 5

i Table of Contents List of Tables iii List of Figures iv Abstract v Chapter One: Introduction 1 Factors Affecting Intelligibility 1 Talker-Related Factors 1 Listener-Related Factors 2 Environmental Factors 2 Linguistic Factors 3 Two Factors Manipulated in Present Study 4 Bilingualism Factor 4 Three aspects of language 4 Source-filter-theory 5 Phonetic differences between native and non-native speech 6 Vowels 7 Speech learning model 10 Speaking Style Factor 11 Clear speech research 12 Intelligibility in Noise 13 Clear Speech Benefit 14 Purpose of Present Study 15 Research Goals of Present Study 16 Hypotheses of Present Study 17 Chapter Two: Method 18 Participants: Talkers 18 Participants: Listeners 21 Stimuli 25 Conversational List 26 Clear Speech List 26 Recording, Instrumentation, and Procedures 26 Isolation of /bVd/ Words 29 Mixing of Noise 30 Instrumentation for Perception Data Collection 31 Perception Procedures 33

PAGE 6

ii Chapter Three: Results 37 Chapter Four: Discussion 51 Summary of Results 52 Hypotheses 56 Comparison with Previous Research 57 Implications 60 Limitations 61 Future Research 62 References 64 Appendices 68 Appendix A.1: Consent Form for Production Experiment (Monolingual) 69 Appendix A.2: Consent Form for Production Experiment (Bilingual) 72 Appendix B.1: Language Background Questionnaire for Monolinguals 75 Appendix B.2: Language Background Questionnaire for Bilinguals 77 Appendix C: Race/Ethnicity Form 80 Appendix D: Consent Form for Listening Experiment 82 Appendix E: Receipt for Production and Perception Experiments 85 Appendix F: Instructions Handout for Listening Experiment 86

PAGE 7

iii List of Tables Table 1 Demographic Data fo r Early Bilingual Talkers 22 Table 2 Demographic Data fo r Late Bilingual Talkers 24 Table 3 Results of Three-Way ANOVA of the Effects of Talker Group, Speaking Style and Target Word 38

PAGE 8

iv List of Figures Figure 1. Layout for the six-alternativ e forced-choice task completed by all listeners. 33 Figure 2. Mean percent-correct word-identification scores for the three talker groups: monolingual (MO), early bili ngual (EB) and late bilingual (LB), for both speaking styles. 39 Figure 3. Mean percent-correct word-iden tification scores for “bead” (panel A) and “bid” (panel B) for the thre e talker groups: monolingual (MO), early bilingual (EB) a nd late bilingual (LB). 42 Figure 4. Mean percent-correct word-identif ication scores for “bayed” (panel A) and “bed” (panel B) for the thre e talker groups: m onolingual (MO), early bilingual (EB) a nd late bilingual (LB). 43 Figure 5. Mean percent-correct word-iden tification scores for “bad” (panel A) and “bod” (panel B) for the thr ee talker groups: monolingual (MO), early bilingual (EB) a nd late bilingual (LB). 44 Figure 6. Mean percent-correct word-iden tification scores of each talker in the monolingual (MO) talker group. 46 Figure 7. Mean percent-correct word-identif ication scores of each talker in the early bilingual (EB) talker group. 47 Figure 8. Mean percent-correct word-identif ication scores of each talker in the late bilingual (LB) talker group. 48

PAGE 9

v Clear Speech Effects for Vowels Produ ced by Monolingual and Bilingual Talkers Teresa DeMasi ABSTRACT ‘Clear speech’ is a speaking style that ta lkers often employ when they know they may have trouble being understood, as when sp eaking in noise or to a person with a hearing loss. When ‘clear speech’ produced by native talkers is presented in noise to native listeners, it has been shown to be a bout 10-15 percentage point s more intelligible, on average, than normally produced speech. Recent research has shown that bilingual listeners may experience a smaller intelligib ility benefit than monolingual listeners from ‘clear speech’ produced by monolingual talkers. The present study co mpares the ability of monolingual and bi lingual talkers to produce this clear speech intelligibility benefit. The present study investigates the hy pothesis that bilinguals may produce a smaller intelligibility benefit than monolinguals when asked to speak clearly. Three groups of talkers were recorded: 13 mono lingual native English speakers, 22 ‘early’ Spanish-English bilinguals, with an age of onset of learning English (AOL) of 12 or earlier, and 14 later Spanish-E nglish bilinguals, with an AOL of 15 or later. Talkers produced the target words “bead, bid, baye d, bed, bad” and “bod" in both clear and conversational speech styles. Two repetitions of each word were mixed with noise and presented to monolingual English-speaking li steners across two days of testing.

PAGE 10

vi Both monolingual and early bilingual talker s showed a similar degree of clear speech benefit in noise (about 5.5%). Later b ilinguals were the least intelligible overall and showed a smaller overall clear speech bene fit in noise. Surprisingly, early bilinguals were significantly more intelligible than monolinguals in both speaking conditions (by about 6.5%). For the later bilingua ls only, performance was significantly worse for one target word (“bid”) in the clear speech c ondition than in the normal speech condition. These data suggest that later bilinguals, but not early bilinguals, may experience a relative disadvantage when speaking in noise due to a reduced ability to improve intelligibility by speaking more clearly. Therefore, these persons may benefit from communication strategies or accent reduction pr ograms designed to increase their ability to make themselves understood in difficult speaking environments.

PAGE 11

1 Chapter One Introduction The goal of oral communication is for th e talker’s ideas to be understood by the listener. There are three components to or al communication: 1) the speaker, 2) the message, and 3) the listener. The communication process begi ns with the talker having an idea. Next, the idea is encoded into a me ssage, in this case, a verbal message. The process is complete when the listener rece ives the message successfully (Smith, 2007). Sometimes the intent of the message is not understood by the listener, which may result in miscommunication. In order to determin e the success of the communication exchange, intelligibility is one measure that can be used. Intelligibility is typically measured as the number of words intended by the talker th at are correctly perceived by the listener (Miller, Heise & Lichten, 1951). Factors Affecting Intelligibility There are several factors that may affect intelligibility. These factors can be talker-related, listener-related, e nvironmental, or linguistic. Talker-Related Factors Talker-related factors can include the age, gender, and physical state of the talker. The talker may be tired and fatigued when communicating the message or the talker may have a speech or language impairment, maki ng it difficult for the listener to accurately understand his/her speech. Anothe r talker-related factor is bi lingualism. If the talker is speaking a language that is not his/her native language and if the talker does not have

PAGE 12

2 sufficient linguistic experience in th e second language, then a breakdown in communication may occur. Listener-Related Factors Listener-related factors that impact inte lligibility can include the age, hearing ability, and linguistic experien ce of the listener. Presbycus is, the decrease in hearing acuity with age (Tye-Murray, 2004), may affect the listener’s ability to accurately perceive the message. If the listener ha s a hearing impairment, he/she may require amplification or other accommoda tions in order to understand the message. If the talker is using a language that is not the listener’s native la nguage, the listener may need additional time or resources to fully process the message. Environmental Factors Environmental factors can have a nega tive impact on communication as well. These factors can include background noise and other distractions, which may make it difficult for the listener to hear the me ssage. Imagine having dinner in a crowded restaurant, where the level of background noise is quite intense. Between the extraneous conversations in the room, the clinking of silver and glassware, and your own conversation, it can be a challenging task for a listener to accurately perceive the message being encoded at the dinner table. Roge rs, Dalby and Nishi (2004) found that non-native speech embedded in background noise may creat e higher task demands for the listener than native speech presented in the same environment. In their study, native, high proficiency non-native late lear ners of English as a second language (who had a relatively mild but detectible foreign accent), and less prof icient non-native late learners of English as a second language (who had a moderate to strong foreign accent) were recorded

PAGE 13

3 reading a list of fifty sentences These sentences were later combined with noise at three different signal-to-noise ratios. The sentences were then presented in quiet and in noise to native listeners, who were instructed to write down what they heard. Intelligibility was determined based on the number of content words accurately identified in the intended order. Their results showed that speech inte lligibility decreased with increasing levels of noise for all talker groups. In the quiet li stening condition, the intelligibility of the highproficiency non-natives was similar to that of the natives; however, in the noisiest condition, the intelligibil ity of the high-proficiency non-nati ves was similar to that of the less-proficient non-natives. Thes e results suggest that even fluent bilinguals who retain even a relatively mild foreign accent may be less able to make themselves more intelligible under adverse listening conditions. Linguistic Factors In addition to talker, listene r, and environmental factor s, linguistic factors can affect intelligibility as well. These factor s include speaking style, word frequency, and sentence complexity. The speaking style that is employed by the talker may be conversational. Conversational speech tends to flow naturally. Sometimes the talker may use a rapid rate of speech, which can be difficu lt for the listener to follow. Some talkers may use clear speech in order to make themselves better understood under adverse listening conditions. Specific properties of clear speech will be discussed further below. The talker’s choice of words can also affect communication. If the talker selects words that occur frequently in the language, ther e is a better chance the message will be understood; however, if th e talker selects words that occu r infrequently in the language, there is a higher chance a communication break down may occur. Finally, short, simple

PAGE 14

4 sentences typically are easier to understa nd, whereas longer, more complex sentences may be more difficult to understand. Two Factors Manipulated in Present Study There are many factors that can negativel y impact the intelligibility of one’s speech. In the present study, two of the factors mentioned above were actively manipulated: the talker-related factor of bili ngualism and the linguistic factor of speaking style. Bilingualism Factor Bilingualism can be defined as being able to use two or more languages on an everyday basis (Grosjean, 1989). Bilingua ls can learn two languages either simultaneously or consecutively. In the pr esent study, all of the bilingual participants except one were consecutive learners. The one subject who was a simultaneous learner reported learning Spanish first and learning English between the ages of two and three years. Since this is a crucial time for language development in children, she can be considered a simultaneous learner. Wh en an individual learns two languages consecutively (e.g., learns a second language after achieving proficiency in the native language), he/she may sometimes speak th e second language with a foreign accent. According to Rogers, et al. (2004), a fore ign accent is perceived when a non-native speaker produces L2 sounds that differ from native-speaker phonetic norms. Three aspects of language. Many differences exist betw een native and non-native speech, which can be easily identified when the three main aspects of language are examined. These three aspects are the conten t, use, and form of language (Smith, 2007). Language content includes semantics, whic h refers to the vocabulary and meaning

PAGE 15

5 differences between the two languages. Langua ge use includes pragmatics, which refers to the social aspects of language. Language form includes syntax, phonology, and morphology. Syntax includes word order, grammar, and sentence structure. For example, in English an adjective precedes a noun, while in Spanish the opposite is true (e.g. “white house” in English; “casa blanca” in Spanish). Phonology includes the sounds that make up the language. This would be the individual vowels and consonants. For example, Spanish only has five monophthonga l vowel sounds in its phonemic inventory (Dalbor, 1969), while Englis h has approximately 12 (Lad efoged, 1982). Both languages contain diphthongs, as well, but they will not be discussed further because they were not under investigation in the pr esent study. Morphology incl udes the smaller meaningful units of words. For example, adding –ing at the end of a verb will make it present progressive tense in English, while adding –ndo at the end of a verb in Spanish has the same effect. Differences between the nativ e and non-native languages can all affect the ways in which the non-native speaker may se lect, order and pronounce words in a second language and all of these factor s may affect intelligibility. Source-filter-theory. When producing speech, there are many articulators involved, which include the lips, teeth, tongue jaw, hard palate, velum, and pharynx. The position of the articulators in the oral cavity determines the quality of the speech sound that is produced; therefore, articula tory differences produced by non-native speakers may be perceived as a foreign accent. According to the Source-Filter-Theory, the voice is produced at the vocal folds, whic h is the source (Kent, Dembowski, & Lass, 1996). The speech signal is then filtered th rough the oral cavity, which is shaped by the articulators to create indi vidual speech sounds. These s ounds are radiated from the

PAGE 16

6 mouth, which results in the output signal of connected speech (Kent, et al., 1996). The resonances in the vocal tract are determined by the positions of the articulators; these resonances determine which source frequenc ies will be enhanced and which will be dampened. Differences in the position of the articulators between native and non-native speakers will result in differences in vocal tr act resonances. Thus, articulatory differences can cause the output speech signal to sound dist orted or accented if it does not match the sound pattern that native speakers are used to hearing. As a result, this mismatch of sound patterns may lead to reduced inte lligibility of the speech signal. Phonetic differences between native and non-native speech. Clearly, there are many differences across languages. Howe ver, the present research study will only examine intelligibility differences related to pronunciation. There are various types of phonetic differences between native and nonnative speech. Non-native talkers may delete non-native sounds, substitute native s ounds for non-native sounds, and distort nonnative sounds. These differences from native pronunciations can occur amongst consonants or vowels. Prosodi c variations are evident across languages, as well. An example of a prosodic difference would be the tonal properties that are present in Mandarin Chinese. Since this is a tone la nguage, two different words may be pronounced the same, but when they are produced with di fferent intonation patterns, the meaning of the word changes. When learning a language such as English, a na tive Mandarin Chinese speaker may transfer prosodic elements from their native language to English, thereby producing certain English words with incorr ect intonation patterns. An example of a consonant difference would be the production of the English glides, /r/ and /l/, by native Japanese speakers. Japanese does have one glide-like phoneme, whic h is often described

PAGE 17

7 as being between /r/ and /l/ (Small, 1999). Th erefore, distinct production of these two English sounds presents difficulty for Ja panese speakers (Flege, Bohn, & Jang, 1997). An example of a vowel difference would be na tive German speakers’ difficulty producing the English / -/ contrast, or native Spanish speake rs’ difficulty producing the English /i-I/ contrast (Flege, et al., 1997). Rogers and Dalby (2005) found that vowel performance on a minimal-pairs probe list produced by Mandarin-accented English (MAE) speakers was a better predictor of conn ected-speech intelligibility th an was consonant performance on the word list. This suggests that vowe ls may be of particular importance in intelligibility for non-native speakers. As a re sult of this researc h, vowel effects were selected to be studied in the present study. Vowels The physical characteristics of vowel s contribute to their intelligibility. These characteristics include temporal, spectr al, and dynamic cues. Based on analyses of American English vowels, conducted by Hille nbrand, Getty, Clark, and Wheeler (1995), these cues can be defined as the following: te mporal cues are measur ed as the duration of the vowel; spectral cues are measured based on a single, steady-stat e time slice of the vowel; and dynamic formant cues can be m easured at 20% and 80% of the vowel duration. If any of these cues are rem oved or altered, vowels may be less easily identified (Hillenbrand & Nearey, 1999). In the Hillenbrand and Nearey (1999) study, twenty listeners participated in a vowel id entification task. Vowe ls were embedded in /hVd/ words and were presented in three fo rms: natural vowel, original formant (OF) synthetic vowel, and flat formant (FF) synthe tic vowel. The natural vowel stimuli were unaltered. The flat formant synthetic vowel was created using form ant measurements of the original vowel. The flat formant vow el was synthesized using fixed formants

PAGE 18

8 measured at the steadiest segm ent of the original vowel. Th e original formant synthetic stimuli were created by modeling the full deta il of the formant traces measured from the database created by Hillenbrand et al. (1995), updated at 8 ms intervals. Results of the experiment revealed that OF synthetic vowels were identified with greater accuracy than FF vowels; and naturally produced vowels were id entified with greater accuracy than OF synthetic vowels. Even FF vowels, however were identified with relatively high accuracy (about 73.8% on average). These re sults imply that both spectral and dynamic cues both play an important role in vowel identification. These effects of these cues were also i nvestigated in a vowel identification task conducted by Ferguson and Kewley-Port ( 2002), which included young normal-hearing (YNH) and elderly hearing-impaired listene rs (EHI). Monosyllabic words that were produced in either a clear or conversationa l speaking style by a single talker were presented in 12-talker babble, and all listeners were instructed to identify the vowel in each word. All words were produced in /bVd/ context and were isolated from a carrier phrase. The YNH group found vowels to be abou t 15 percentage points more intelligible in the clear speech condition than in the conversational speech condition. The EHI group did not obtain this same clear speech benefit. It should be kept in mind that the listeners were unamplified and therefore, audibility may have been an issue. Both listener groups were found, however, use all three vowel cues studied (spectral targ et, dynamic formant movement, and duration) to id entify vowels in this task. While previous research has studied speech intelligibility of clear and conversational sent ences in noise, Ferguson and Kewley-Port (2002) was the first study to fo cus specifically on vowel identification in noise in both clear and conversational speech.

PAGE 19

9 Studies of foreign accent also have found that differences in the use of spectral and duration cues in vowel perception a nd production. For example, Bohn and Flege (1997) examined production and perception of English vowels across three groups: native English speakers and two groups of native Germ an speakers with different amounts of L2 exposure. The English vowel // was compared to / / in order to determine if the nonnatives could produce and perceive the differenc e between these two so unds. This pair of vowels was chosen because // does not ex ist in the German language but a vowel similar to / / does exist in German. Results of the production experiment found that target // and / / produced by the experienced Germans were very similar to those produced by the native English speakers. However, // and / / productions of th e inexperienced Germans differed from the other two groups in two ways: 1) the inexperienced Germans produced / / higher in the vow el space (i.e., closer to / /) and 2) the inexperienced Germans produced a smaller duratio n contrast between // and / /. Therefore, the inexperienced Germans’ production of // wa s very similar to their production of / /. Thus, it can be concluded that the inexperi enced Germans produced // differently from the other two groups in terms of both spectral and durational cues. Non-native speaker perception may also reflect perceptual cue weighting strategies that are different fr om those of native speakers. This can be seen in the German subjects mentioned above (Bohn & Flege, 1997) Results of the perception experiment showed that the inexperienced Germans relied heavily on duration cues to distinguish // from / /, while they relied less on spectral pr operties. However, the native English speakers and the experienced Germans relied on spectral properties to identify these vowels, while duration differences had very little influence on their perception. This

PAGE 20

10 suggests that the inexperienced Germans were not attending to the appropriate cues when identifying non-native vowels. The differen ce from native speaker norms in the use of perceptual cues may have influenced thei r productions of the E nglish vowels // and / /. This perception study suggests that the amount of experience a speak er has with the L2 may influence his/her ability to use native-sp eaker appropriate per ceptual cue weighting strategies to identify vowel s in a native-like manner. In another bilingual study, Kewley-Por t, Akahane-Yamada, and Aikawa (1996) conducted an experiment to determine which spectral-temporal properties influenced the intelligibility of Japanese-accented English vowels. Japanese speakers were recorded producing /bVt/ words. These words were pr esented to native English listeners via a minimal-pairs forced-choice perception task. There were three major results discovered in this experiment. First, Japanese speakers were not able to produ ce the spectral property of English vowels adequately to differentiate vowels that are located close together in the vowel space. Second, Japanese speakers were sometimes able to produce the dynamic contrasts among English vowels, but not always Third, Japanese sp eakers were able to produce the correct durational cues for the Engl ish vowels. In order to determine how the acoustic properties contributed to intelligib ility, regression analys es were performed. Results of these analyses showed that sp ectral and dynamic properties of vowels were more important than duration cues in the intelligibility of Japanese-accented English vowels to native English-speaking listeners. Speech learning model. Inaccurate production of American English (AE) vowels by bilingual talkers may cont ribute to a foreign accent. As mentioned previously, American English has about 12 monophthongal vowels, whereas Spanish only has five

PAGE 21

11 monophthongal vowels. According to Fleg e’s (1995) Speech Learning Model, as children learn their native langua ge, vowel categories are create d in their brains, for each of the vowel sounds that occur in their nati ve language. Based on this theory, native English speakers have at least 12 vowel cate gories in their vowel space, while native Spanish speakers have only five categories in the corresponding space. This difference in vowel inventory between the two languages su ggests that Spanish speakers may have difficulty producing some English vowels accu rately. The native vowel categories are considered “old” categories. As bilin guals learn a second language and develop proficiency with their L2, they develop “new” categories, which contain the vowel sounds of the L2 that do not occur in their L 1. The more distinction there is between the “old” and “new” categories, the more native-li ke the second language learner will sound. If an “old” category is heard as similar to a new L2 sound, the old category may be used in its place; if the differe nces between the non-native’s “old” productions and the L2native pronunciation of the new L2 sound are noticeable to native speakers of the L2, these differences will be perc eived as a foreign accent. This increased accentedness may in turn contribute to redu ced speech intelligibility. Speaking Style Factor While the talker-related factor of bilingualism contributes to reduced intelligibility, the linguistic factor of speaking style also affects intelligibility. Normal everyday speech patterns are known as convers ational speech. Another type of speaking style that can be employed by talkers is clear speech. Clear speech is a style of speech that talkers may use when they know they may have trouble being understood such as when talking in a noisy listening environment or when talking to a listener who has a

PAGE 22

12 hearing impairment. Picheny, Durlach, and Br aida (1986) conducted acoustic analyses of conversational and clear speech tokens pr oduced by native speakers recorded from a previous study (Picheny, Durlach, & Braida 1985). Several differences between clear and conversational speech were found. Thes e differences include the following: 1) speaking rate is slower in clear speech; 2) in conversational speech, unstressed vowels are modified (reduced) and final stops are usua lly not released, while in clear speech, unstressed vowels are not modified as much and more final consona nts are released; and 3) the root mean square (RMS) intensities for stop consonants are greater in clear speech than in conversational speech. According to Bradlow and Bent (2002), clear speech consists of the following: slow er rate of speech, more frequent pauses between words, greater intensity, more varied pitch range, greater emphasis on the release of plosives, and less vowel reduction. Clear speech research Research has shown that th e use of clear speech does improve intelligibility of the sp eech signal. Picheny, et al. (1985) recorded three native English speakers reading 50 nonsense sent ences in both clear and conversational speaking styles. These sentences were pres ented monaurally via headphones to five listeners with sensorineural hearing loss. A ll listeners were instructed to either write down the sentence that they hear d or to repeat the sentence. Results indicated that for native speakers, clear speech is about 17% mo re intelligible than conversational speech for sentences in quiet. It is important to k eep in mind that these results are based on the perception of listeners with hearing impairm ent, while the present study used listeners with normal hearing. Also in the Pichney, et al. (1985) study, the researchers examined sentences, while the present study investigates vowels.

PAGE 23

13 Previous research has shown that the cl ear speech benefit is about 8% for vowels alone (Ferguson, 2004). Ferguson (2004) reco rded 41 talkers (21 female, 20 male), reading sentences in both clear and conve rsational speech conditions. From these sentences, 1640 /bVd/ words were extrac ted and embedded in 12-talker babble for presentation to seven young adult listeners in a vowel identification task. Results indicated that vowels were more intelligible in the clear speech condition than in the conversational speech condition. A gender effect was also found. Females were significantly more intelli gible than males in the clear sp eech condition, while this same effect was not present in th e conversational speech conditi on. The Ferguson (2004) study concluded that the ability to produce a clear speech benefit implies that the talker has linguistic knowledge of the cues that listeners use to identify vowels; it also implies that the talker has the ability to modify those cues in order to make himself more intelligible. Intelligibility in Noise Non-native speakers with a foreign a ccent may not always be well understood under normal listening conditions, and they may be even more difficult to understand under adverse listening conditions. Rogers, et al. (2004) found th at even relatively mildly accented speakers may experience disproportionate decreases in intelligibility in noise compared to native speakers. Reduced intelligibility may cause breakdowns in communication, especially in noisy environments. This may lead to heightened frustration levels for both communicative part ners. Additionally, this may even cause non-native speakers to withdraw from social in teractions, or it may cause native speakers to exclude non-natives from social interactions.

PAGE 24

14 In the Rogers, et al. (2004) study cite d earlier, additional results were found related to the issues of b ilingualism and clear speech. Na tive speakers were found to be more intelligible than non-nati ve speakers overall. Intere stingly, it was found that highly proficient non-native speakers performed ve ry similarly to the native speakers under quiet listening conditions; how ever, when even a relatively small amount of noise was added to the speech signal, the high profic ient non-natives experienced a greater intelligibility deficit than did the native speakers. This implies even relatively proficient non-native speakers may benefit from instru ction in how to make themselves more intelligible in noisy listening conditions. Clear Speech Benefit While it has been documente d that native speakers are able to produce a clear speech benefit (Picheny, et al. 1985), it is unknown whether non-native speakers are able to produce this same benefit or the extent to which it may differ amongst bilinguals with varying degrees of L2 proficiency. Bradlow and Bent (2002) suggested that clear speech is native-listener oriented, and will only be nefit listeners who are experienced with the sound structures of the target language. In their study, two native English speakers were recorded reading 64 simple sentences in two speech conditions, clear and conversational. These sentences were originally designed for children and, thus, contained a limited vocabulary. The sentences were later embe dded in white noise and presented to 32 native and 32 non-native listeners in a senten ce recognition task. All non-native listeners came from various language backgrounds and had limited exposure in an Englishspeaking environment. Results indicated th at non-native listeners received a much smaller clear speech benefit in noise than did na tive listeners. It is important to keep in

PAGE 25

15 mind that the non-native listeners in th e Bradlow and Bent (2002) study were not separated by age of acquisition of English. Since early learners te nd to be much more native-like than late-learners (Rogers, Lister, Febo, Besing & Abrams, 2006), the question of whether a clear speech benef it differs amongst bilingual listeners from various L2 proficiency levels was not a ddressed by Bradlow and Bent (2002). Additionally, there has not been any research found to date that ha s compared native and non-native talkers’ ability to produce a clear sp eech benefit in noise. Previous research has used native talkers. We already know th at some non-native listene rs receive less of a clear speech benefit in noise than native li steners for some speech materials, but we do not yet know whether non-native talkers are ab le to produce a clear speech benefit in noise when asked to speak clearly. Purpose of Present Study Understanding the degree to which bili nguals can improve intelligibility when asked to speak clearly, compared to monoli nguals, can be useful in accent modification therapy for those who seek it. For example, if a speaker is not able to produce and perceive L2 sounds accurately, then there is a possibility that the speaker will not be able to improve clarity when he/she needs to do s o. The results of the present study can help to better understand the effect that the in struction to speak clearly will have on the intelligibility of non-native speech. The present study will also add to ou r understanding of the problem because a reduced clear speech benefit for non-natives ta lkers may suggest that bilinguals lack the ability to enhance the appropriate cues wh en asked to speak more clearly, perhaps reflecting differences in linguist ic knowledge. If this is th e case, acoustic analyses might

PAGE 26

16 help to determine which cues native and non-native speakers enhance or do not enhance when asked to speak more clearly. The present study compared the intelligibility of six American English vowels produced by three groups of speakers: mono lingual native speakers of English, early Spanish-English bilinguals, and late Span ish-English bilinguals. Spanish-English bilinguals were chosen because they constitu te a large portion of the bilingual population in Tampa, Florida, which is where the st udy was conducted. According to the United States Census Bureau (2000), there are 28 million Spanish-speaking bilinguals in the United States. This rapid growth of SpanishEnglish bilinguals has gi ven rise to the need for research in the area of speech production in this population. Six vowels were targeted in the present study. They are /i, I, e, / and /a/. These spec ific vowels were selected because in American English, they represent a variety of high, mid, and low vowels, and a good combination of tense and lax vowels, as well. Also, all of these vowels occur frequently in the English language. Three of these vowels occur in the Spanish language (/i, e, a/) and are therefor e often considered “old” vowel s in the bilinguals’ L2 vowel repertoire (Flege, 1995). Three of these vow els do not occur in the Spanish language (/I, /) and are therefore often considered “n ew” vowels in the b ilinguals’ L2 vowel repertoire (Flege, 1995). Research Goals of Present Study Results of this study provide insight on how native Spanish speakers’ productions of AE vowels are perceived by native English listeners and what the differences are in intelligibility across talker groups, speaking st yles, and individual vowels. Additionally, results of this study provide information as to how vowels, specifically, may contribute to

PAGE 27

17 non-native talker intelligibil ity. The following three resear ch goals were addressed in this experiment. First, a comparison across a ll three talker groups was made to determine the differences in native and non-native talkers’ ability to produce a clear speech benefit in noise. Second, a comparison was made of the differences in the effects of clear speech across individual target vowels for each talker group. Third, the performance of individual talkers was compared within each group. Hypotheses of Present Study The investigators formulated hypotheses base d on their expected findings for this study. First, it was hypothesi zed that there would be no change in vowel production across conversational and clear speech condition s for the late bilinguals for some target vowels. Second, it was postulated that the early bilinguals would produce a clear speech benefit in noise. Lastly, it was hypothesized that the native speakers would also produce a clear speech benefit in noise.

PAGE 28

18 Chapter Two Method Participants: Talkers The population that was targeted in the first part of this experiment included native monolingual English talkers, highly-expe rienced early SpanishEnglish bilinguals, and less-experienced la te Spanish-English bilinguals. Pa rticipants were recruited through campus advertisements including flyers, the st udent newspaper, and e-mail. Groups were selected based on language experience. Na tive English-speaking monolinguals were placed in the monolingual group (MO); nativ e Spanish talkers who began learning English intensively prior to the age of 12 y ears were placed in the early bilingual (EB) group; and native Spanish talkers who began learning English intensively after age 15 years were placed in the late bilingual (LB) group. Additionally, in order to qualify for the EB group, participants had to rate themselves as English dominant or balanced in at least two of four modalities (listening, speaking, reading, or writing) and one of these modalities had to be non-print (listening or speaking). The ages of 12-15 years were chosen as a separation range between the tw o bilingual groups because previous research has found that most people who acquire a se cond language after age 15 years will speak the L2 with a perceptible fore ign accent (Flege, et al., 1997). In addition, those who arrive in the U.S. before age 12 will typically have at least 6 years of schooling in English and will therefore have a relatively long period of immersion by age 18.

PAGE 29

19 The inclusion/exclusion criteria for a ll subjects included th e following: either gender could participate; no subject with a pr ior history of speech or hearing impairment was included; all subjects had to fall within a certain age range (native talkers had to be between the ages of 18 and 45 years, while bilin gual talkers had to be between the ages of 18 and 60 years); and all subjects had to pa ss a pure-tone hearing screening at 25dB HL at 500, 1000, 2000, and 4000 Hz. The audiometer was calibrated to ANSI 1989 standards. Additionally, it wa s preferred that all native ta lkers be from the Tampa Bay area or have lived in the state of Florida for at least five ye ars. This last criterion was chosen in order for the native talker group to have a relatively cons istent and standard regional accent. Some native talkers were al lowed to participate ev en if they did not meet this criterion, as long as they did not possess a strong regional accent. Given the demographics of Spanish talkers in the Ta mpa Bay area, it was not practical to find a large subset of native Spanish talker subjects who spoke the same dialect; therefore, all New World varieties of Spanish were deem ed acceptable for participation in this experiment. For the native English talkers, no native/fl uent talkers of any language other than English were allowed to participate; they c ould not have exposure to a foreign language other than studying a foreign language in hi gh school or college to meet graduation requirements. Additional criteria for the bilingual talkers included the following: Spanish had to be their native language; they could not speak the Cas tilian dialect which is spoken in Spain, since this is very different from Sp anish dialects of the Americas; they should have learned English either as a second la nguage or simultaneously with Spanish; they could not speak any languages other than E nglish and Spanish; they could not have

PAGE 30

20 exposure to any other language other than a foreign language studied in high school or college to meet graduation requirements. A ll talkers were paid $10 per hour for their time, with the exception of one early bilingu al participant who opted to volunteer her time for this experiment. A receipt was pr ovided to those participants who received monetary compensation. This receipt can be found in Appendix E. Original participants included 24 monoli nguals (3 male, 21 female) with an age range of 18-38 years (mean 22.6 years); 33 early bilinguals (6 male, 27 female) with an age range of 18-35 years (mean 21.5 years); a nd 21 late bilinguals (5 male, 16 female) with an age range of 19-57 years (mean 29.4 years). Some subjects were dropped for various reasons. Although both genders were eligible to participat e, all male subjects were eventually dropped due to a dispropor tionate number of male volunteers. In previous research, Ferguson (2004) found a gende r effect on intelligib ility between clear and conversational speech. In order not to sk ew our data, we decide d it was best to drop all male participants. Other participants were dropped if they failed to meet the inclusion and exclusion criteria listed above. In some instances, participants seemed to fit all criteria, but later on, after car eful review of their language background questionnaires, it was clear that some of them, in fact, did not meet all criteria for inclusion, and these participants were dropped. Of the 24 monolingual participants 11 of them were dropped. Of the 33 early bilingual pa rticipants, 11 were dropped. Of the 21 late bilingual participants, seven were dropped. Participan ts were dropped for the following reasons: including the following: prior history of speech or hearing impairment; poor vocal quality (e.g., hoarseness) as observed by investigat ors; presence of a regional dialect of American English or noticeable pronunciation patterns of African American Vernacular

PAGE 31

21 English (AAVE), as subjectively judged by i nvestigators; unclear age of acquisition of English, as reported on language background ques tionnaire. Some bilingual participants were also dropped because they did not meet criteria for placement in either bilingual group, as per their report on language background questionnaire (e.g. for inclusion in EB group, participants must report being Eng lish dominant for at least two of four modalities: speaking, listening, reading, and wr iting; one of which must be either speaking or listening). After subjects were dropped, there were 49 participants remaining. These 49 participants included 13 native English talker s with an age range of 18-34 years (mean 22.38 years), 22 early bilinguals with an ag e range of 18-35 years (mean 21.73 years), and 14 late bilinguals with an age range of 19-57 years (mean 29.57 years). Additional demographic information for the two groups of bilingual talkers is provided in Tables 1 and 2 on the following pages. Participants: Listeners The population that was targeted in the second part of this experiment included monolingual, native English-speaking listeners Inclusion/exclusion criteria included the following: either gender could pa rticipate; no subject with a hi story of speech or hearing impairment could participate; all participan ts had to be between the ages of 18 and 45 years; participants could not speak any language other than English fluently; they could not have a strong regional accent; they could not rate themselves as having less than native-like proficiency in E nglish; and they had to pass a pure-tone hearing screening presented binaurally at 25dB at 500, 1000, 2000, and 4000 Hz.

PAGE 32

22 Table 1 Demographic Data for Early Bilingual Talkers Language background information Language most co mfortable for: Code Age Born/ Raised in US? Country AOI Speak Listen Read Write EB02 18 N Dom.Rep. 7 E E E E EB04 18 N Mexico 6 E E E E EB05 19 Y Cuba 4.5 E E E E EB06 19 N Mexico 5 B B E E EB08 19 N Nicaragua 8 E E E E EB10 19 Y Nicaragua 6 B B B B EB11 20 Y Cuba 6 E E E E EB12 24 N Puerto Rico 10 E E E E EB14 20 Y Cuba 4 E E E E EB16 19 Y Mexico 6 S E E E EB17 19 Y Cuba 4 E E E E EB18 35 N Venezuela 9 E S S E EB19 18 Y Cuba 4 E E E E EB20 27 N Venezuela 4 B E E B EB23 29 N Puerto Rico 9 E E E E EB24 26 Y Colombia 5 E E E E EB25 21 N Colombia 11 E E E E

PAGE 33

23 EB26 26 N Venezuela 12 B B E E EB29 19 Y Cuba 2 B B E E EB30 19 N Venezuela 8 B B B E EB31 22 Y Mexico 6 E E E E EB33 22 N Colombia 6 S E E S Avg./ Sum. 21.7 10 Y;12 N 3 Colom.4 Venez. 6 Cuba 4 Mexico 5 Other 6.5 14 E; 6 B; 2 S 16 E; 5 B; 1 S 19 E; 2 B; 1 S 19 E; 2 B; 1 S Note. The columns in the above table contain the subject code (EB: early bilingual), age of subject in years, whether or not the subject was born in the U.S. (Y: yes, N: no), and the country where either the subject or his/he r family originates from. AOI indicates the age of onset of immersion in an English-spea king environment. The last four columns indicate the language that th e subject reported being more comfortable in for speaking, listening, reading, and writing (E: English, S: Spanish, B: both languages). These data were retrieved from the language background questionnaire completed by all bilingual subjects.

PAGE 34

24 Table 2 Demographic Data for Late Bilingual Talkers Language background information Language most co mfortable for: Subject Code Age Born/ Raised in US? Country AOI Speak Listen Read Write LB01 30 N Panama 21 E S B B LB03 19 N Colombia 15 S S S S LB06 19 N Colombia 16 S S S S LB07 50 N Colombia 45 S S S S LB09 21 N Colombia 20 S S E S LB10 28 N Colombia 28 S S S S LB11 22 N Colombia 22 S S S S LB12 35 N Colombia 35 S S S S LB13 19 N Puerto Rico 16 S S S S LB15 22 N Colombia 18 S S S S LB16 49 N Colombia 46 S S S S LB18 57 N Colombia 30 S S S S LB19 22 N Cuba 19 S S E E LB21 21 N Colombia 18 S S S S Avg./ Sum 29.6 14 N 11 Colom. 1 Cuba 3 Other 24.9 13 S; 1 E 14 S 11 S; 2 E; 1 B 12 S; 1 E; 1 B

PAGE 35

25 Note. The columns in the above table contain the s ubject code (LB: late bilingual), age of subject in years, whether or not the subject was born in the U.S. (Y: yes, N: no), and the country where the subject originates from. AOI indicates the age of onset of immersion in an English-speaking environment. The last four columns indicate the language that the subject reported being more comfortable in for speaking, listening, reading, and writing (E: English, S: Spanish, B: both languages). Th ese data were retrieved from the language background questionnaire completed by all bilingual subjects. Listeners were recruited through cam pus advertisements and through the Communication Sciences and Disorders (CSD) department at the University of South Florida. All subjects were given the opti on of either receiving monetary compensation for their time or receiving extra credit point s for class. Those students who chose to receive monetary compensation were paid $10 per hour for their time, with a $10 bonus at the completion of the experiment, provided th at they showed up to all sessions on time. A receipt was provided to those participants who received monetary compensation. This receipt can be found in Appendix E. Twenty listeners participated (3 male, 17 female) with an age range of 18-33 years (mean 22.45 ye ars). Range of years living in the state of Florida was reported by listeners to be between six months and 23 years. Stimuli Each participant from the three talker groups was recorded reading a series of carrier phrases, each of which contained a targ et /bVd/ word. The talkers were instructed to read the carrier phrases under two conditions: conversationa l speech and clear speech. Intelligibility was measured by presenting the recorded stimuli to the monolingual English-speaking listeners via a forced-choice perception task. The pe rcent correct on the

PAGE 36

26 forced-choice perception task was used to dete rmine the intelligibilit y of each of the three groups of talkers, as well as the differences in intelligibility, if a ny, across the two speech conditions. Conversational List For the production part of the experime nt, all three groups of talkers were presented with two lists of stimuli. The firs t list, conversational, contained 84 sentences. Forty-two of these sentences contained a carri er phrase with a /bVd/ word embedded in the middle (e.g. “Say /bid/ again.”). Each /bVd / word contained one of the following six vowels: /i, I, e, / and /a/. Seven repetitions of each carrier phrase were recorded, for a total of 42 /bVd/ words recorded. The other 42 sentences contained foil words embedded in the same carrier phrase (e.g. “Say cat again.”). Clear Speech List The second list, clear speech, contained a total of 42 sentences, each containing a target /bVd/ word embedded in the same carrier phrase that was used in the conversational list. There were also seven repe titions of each of the six vowels for a total of 42 words. Foil words were not used in this list because the investig ator did not want to distract the talker from the target form of the /bVd/ word; instead, the investigator wanted the talker to focus on the /bVd/ word for this particular task. Recording, Instrumentation, and Procedures All talkers were record ed while seated in a sound-attenuating booth with incandescent lighting. This type of lighting was used to reduce noise from the fan attached to the ceiling light during recording. All stimulus sentences were typed into a PowerPoint presentation file, and they were visually presented to each talker via a flat

PAGE 37

27 panel computer display. An Audio Technica microphone (AT4033) was placed at a 45 degree angle, approximately six inches from the corner of the talker’s mouth. All sentences were recorded onto a DSW digita l recorder (Roland, VS 880 Ex). The signal was recorded between -4 and -12dB on the VU meter of the DSW. A pre-amplifier was used, which supplied phantom power to the microphone (48V), and included a gain control for the microphone. Prior to recording, all thr ee groups of talkers were pres ented with a list of twelve words on a piece of paper, which were modeled by the investigator. This list contained the six different /bVd/ words, as well as the six foil words. The target words were typed as “bead, bid, bayed, bed, bad,” and “bod.” The foil words included: “cat, cut, cap, cup, cape,” and “coop.” These words were selected because they were different enough from the target /bVd/ form to distract the talker from the target form, but they were similar enough in that they were monosyllabic CVC word s. Voiceless stops were chosen for the first and last consonant sounds in order to incr ease variability of the types of foil words. The vowel centers were chosen by simply picking vowels from various areas of the vowel quadrilateral in order to create a real word. Some of the foils contained vowels that were not investigated in the present st udy. The idea behind the foils was to distract the talkers away from the target word enough so that they did not pick up on the target word. If the foils were too different from th e targets, then the ta lkers would have been more aware of those differences, and thus the target words would have been more salient. Each subject was asked to repeat each word after hearing the investigator’s pronunciation of each word. This was done to reduce orthographically-related errors in pronunciation.

PAGE 38

28 A practice recording was made first, wh ich contained twelve sentences. Each sentence was presented visually via a flat -panel computer display and aurally over a speaker. The visual and aural stimuli were presented simultaneously. The talker was instructed to repeat each sentence after seei ng and hearing it. This was done to familiarize each subject with the task. After completing the twelve practice items, each talker was instructed to read the sentences in the conversationa l list in a normal speaking voice. All sentences were presented one at a time and only the printed version of the sentences was presented on these trials. All talkers were given a bottle of water, and were told that they would be reading 84 sentences, with a short water break after every 21 sentences. After completing recording for the conversational list, all talkers were given new instructions for the clear speech list. They were told to speak as clearl y as possible, as if talking to someone who was hard-of-hearing or as if someone was having trouble understand ing them. This was how the two speech conditions were differentiate d. If any of the talkers produced any of the sentences with obvious dysfluencies, they were instructed by the investigator to repeat the sentence. All recording procedures took approxim ately one hour per subject and were completed in one session. First, each talker was given paperwork to fill out, including a consent form (see Appendices A.1 and A.2) a language background questionnaire (see Appendices B.1 and B.2), and a race/ethnicity questionnai re (see Appendix C). All paperwork was approved by the Institutional Review Board (IRB). After the talker finished filling out the paperwork, the i nvestigator conducted a pure-tone hearing screening on each subject. All subjects had to pass a pure-tone hearing screening at 25dB

PAGE 39

29 HL at 500, 1000, 2000, and 4000Hz. Next, the word list was read aloud by the investigator and repeated by the subject to reduce orthographical errors. Then, the practice list of twelve sentences was recorded, followe d by the conversational list, and then the clear speech list. Wate r breaks were provided after every 21 sentences for both speech conditions. All recordings were stored on the hard drive of the DSW, and were later transferred to the hard drive of a personal computer. A back-up copy of each recording was saved onto a compact disc. Isolation of /bVd/ Words After all recordings were made, the /bVd / words were isolated in order to be presented to the monolingual native English-sp eaking listeners for the perception part of the experiment. All recorded /bVd/ words were isolated from thei r carrier phrase using the Cool Edit 2000 (2000) software program. In order to do this, first, the waveform of the carrier phrase was opened in Cool Edit 2000 (2000) on a computer screen. After playing the phrase and listening to make sure the correct phrase was selected, the beginning of the phrase was deleted by highlig hting it with the mouse and pressing the “delete” button on the computer keyboard (e.g. the word “say” in the phrase “Say bead again”). The same was done for the ending of the phrase (e.g. the word “again” in the phrase “Say bead again”). After each word was extracted from its carrier phrase, the entire word was highlighted and the RMS am plitude of the entire remaining file was equalized to 25 dB less than the maximum amp litude so all words would have a constant average intensity. Thus, the /b/ release and up to 10ms of pre-voici ng were preserved and the /d/ release and 10ms following the bur st were preserved for all talkers.

PAGE 40

30 This procedure was followed for the first five repetitions of all recorded /bVd/ words in both lists. Seven repetitions were r ecorded in case a talker had any dysfluencies while pronouncing each carrier phrase and becau se future research on this database may examine trial-to-trial variability in talk ers’ productions, but only two high-quality exemplars were used for the present study. Th ese exemplars were usually the first two repetitions of each target word, unless there was a problem such as a dysfluency or the presence of extraneous background noise. In total, the stimuli incl uded two repetitions of each of the six target words, for each of the 49 talkers, in each of the two speaking style conditions, which led to 1176 stimuli all together. Mixing of Noise The selected target words were mixed with multi-talker babble using a customized MATLab script. The reason for adding noise to the signal was because ceiling effects would be expect ed in quiet conditions for both conversational and clear speech tokens for the monolingual and early bilingual talkers (cf. Hillenbrand & Nearey, 1999), perhaps masking improvements in clear speech performance. Based on pilot testing, two signal-to-noise rati os (SNRs) were selected for the listening experiment. For the monolinguals and early bilinguals, a signa l-to-noise ratio of -8dB was chosen; whereas, for the late bilinguals, -4dB was chos en. The reason for this difference was to avoid ceiling and floor effects across all groups In pilot testing on conversational style tokens, the differences in intelligibility across talker groups was found to be so great that no single signal-to-noise ratio was found th at avoided both ceiling effects for the monolingual and early bilingual talkers and floor effects for the late bilingual talkers.

PAGE 41

31 The SNRs chosen resulted in approximately equal performance for all three groups in the pilot testing. For the mixing of the words with noise, two minutes of multi-talker babble from the SPIN sentences (Bilger, Neutzel, Ra binowitz, & Rzeczkowski, 1984) was recorded from a compact disk, amplitude equalized a nd saved to file. The customized MatLab program randomly selected a section of this two minute babble file that was equal in duration to the duration of the target wo rd file plus 1000 ms, with a new randomly selected segment for each target word. Th e program then compared the RMS amplitude of the noise section and the word file, scaled the noise to obtain the desired signal-tonoise ratio, mixed the word and noise files so that the word was centered in the noise (with 500 ms preceding and following the word), and rescaled the combined file to the original RMS of the word file (approximately -25 dB from maximum). A separate randomized mixing of words with noise was completed for each listener, so that the mixing of particularly high peak or low valley in the noise with the vowel of a particular word would not be repe ated for every listener. The 1176 resulting stimuli were then saved in a separate direct ory for each listener on the computers used for presentation of stimuli to the listeners. Instrumentation for Perception Data Collection For the perception part of the experime nt, a personal computer containing the Praat (Boersma & Weenik, 2006) software program was used. The Praat (Boersma & Weenik, 2006) experiment presentation program MFC, was used to control presentation of the stimuli.

PAGE 42

32 On presentation, the stimuli were rout ed to TDT PA5 (TDT System III, 2001) attenuators that were set to th e desired attenuation level, wh ose output was then routed to TDT HB7 headphone buffers, to which the head phones were attached for each station. All stimuli were presented at approximatel y 70 dB SPL. Equipment was calibrated as follows: a 1000 Hz calibration tone was played; this tone was saved to the same average RMS level as all the word stimuli; and the tone was then played over the right and left headphones in turn in each listening station to a Bruel & Kager model 2235 sound level meter. The reading in dB SPL of each ear was taken for each set of headphones; the reading for both ears combined was averaged (left and right always agreed to within about 2 dB); and the attenua tion level needed to get th e output to 70 dB SPL was computed. This attenuation level was se t on the programmable attenuators and the calibration tone was replayed to the sound level meter to double check that the output was indeed 70 dB SPL. Thus, the average pres entation level of the word stimuli to the listeners was equal to that of th e calibration tone (70 dB SPL). All stimuli were output to the listene rs via Sennheiser HD265 headphones in a randomized order, with one half of the stim uli (588 trials) presented on each of the two days of listening. The first repetition of each target word was used for one half of the talkers and the second repetition of the target words was used for th e other half of the talkers on the first day of testing. The rema ining stimuli were presented on the second day of testing. All listeners were seated in a quiet sound-treated r oom. Each listener was seated at a separate carrel. A ll listeners were seated at leas t one carrel apart, and dividers separated each carrel. Each listener’s carre l contained a flat screen monitor, keyboard, mouse and headphones.

PAGE 43

33 Perception Procedures For the perception part of the experi ment, monolingual native English-speaking listeners completed a six-alternative forced -choice perception task. The instructions given were: “You are going to hear a word over headphones. You have six choices on the computer screen in front of you. Use the mouse to click on the word you heard.” The six choices were: “bead, bid, bayed, bed, bad,” a nd “bod.” All listeners performed this task in the same setting. Figure 1 shows the layou t of the computer screen for this task. Figure 1. Layout for the six-alternative forced-c hoice task completed by all listeners.

PAGE 44

34 Due to the number of stimuli presented to each listener, the experiment took place over two separate sessions in order to reduce fatigue. Stimuli were also presented in quiet on two additional days of te sting, but these data will not be presented in this thesis. During the first session, the listener was gi ven paperwork to fill out, including a consent form (see Appendix D), a language background questionnaire (see Appendix B.1) and a race/ethnicity questionnaire (see Appendi x C). All forms were approved by the Institutional Review Board (IRB) of the Univer sity of South Florida. Next, a pure-tone hearing screening was conducted, which followed the same procedures that were used for the production experiment described above. Li steners were then pr esented with stimuli via headphones in one example block in order to familiarize them with the task. The example block contained 12 words presented in quiet with one break halfway through. After the example block, a level setting bl ock was presented in order to give the listener an opportunity to request a softer or louder listening level for the remainder of the experiment. The level setting block consisted of a total of 24 words presented in noise, with a break after every six words. The listene r was able to ask the investigator to adjust the listening level on each break; if adjust ments were not needed, the listener was instructed to continue to the next block. None of th e listeners required any level adjustments for this task; they all reported the listening level of 70 dB SPL to be comfortable. Following the level setting block, a set of practice trials was presented to get the listener accustomed to listening in noise. The pr actice trials consisted of four blocks of 72 words, for a total of 288 practice trials. The first pr actice block was presented in quiet. The remaining blocks were presente d with increasing levels of noise. For

PAGE 45

35 example, the second block was presented w ith a SNR of 0 dB, the third block was presented with a SNR of -4 dB, and the fourth block was presented with a SNR of -8 dB. Short breaks were permitted after each practice block of 72 words. All stimuli used for the example, level-setting, and practice blocks were different from the stimuli used for the main task (stimuli from dropped talkers were used for the example, level-setting, and practice blocks). Following the practice blocks, all listeners were presented with the test stimuli in a randomized order in individual test blocks. The test blocks consisted of six blocks of 98 words, for a total of 588 trials. Test stim uli consisted of two tokens of each of the six /bVd/ words in each speech style (clear or c onversational) recorded by all 49 talkers from the production experiment, for a total of 1176 st imuli. Half of these stimuli, 588, were presented on the first day of the experiment and the other half were presented on the second day of the experiment. All test bloc ks were presented in noise (-8 dB for MO/EB, -4 dB for LB). Short breaks were encouraged after each block of 98 words, and a longer break (about ten minutes ) was given halfway through th e experiment (after block #3). These breaks were provide d to reduce fatigue. All te st stimuli were randomized, and all listeners were presented with stimu li in a different randomized order. In subsequent sessions, each listener was given a set of practice trials at the start of the session to refresh their memory of the task. The second day of the experiment consisted of only the practice and test bl ocks. Test stimuli presented on the second day of testing included the second token of each word that was not presented on the first day of the experiment. Listeners were provided with th e same instructions as the first day. The

PAGE 46

36 number of stimuli and breaks was consistent with day one, as well. The instructions provided to all listeners on this task can be found in Appendix F. As each listener completed the perceptual task, their answers and information on the correct target word were saved into a te xt document. These data were imported into an Excel spreadsheet, and th e number of correct and incorr ect responses was computed for each condition. For example, if the target wo rd was “bid,” but the participant selected “bead,” an incorrect response was recorded in the Excel sheet, and data on the correct response and alternative chosen were saved. Listeners’ confusions (i.e., the words that were perceived rather than the target word for incorrect responses) were not analyzed for this thesis. The number-correct data were then us ed to compute percent-correct data for each talker group (summed across the talkers), target word and speaking style, for each listener. Percent-correct data were also computed for each speaking style and individual talker within a group, summed acro ss the target vowels a nd tokens. Prior to statistical analysis, all percent-correct data were c onverted to rationalized arcsine transform units (RAUs), which ar e more appropriate for the para metric statistical analysis than percent-correct data (Studebaker, 1985) Percent-correct data are used for the presentation of the figures.

PAGE 47

37 Chapter Three Results To address the first two research questi ons, a three-way within-subjects analysis of variance (ANOVA) was used. The independent variables we re the talker group (three levels: monolingual, early bilingual and late bilingual), speaking style (two levels: conversational and clear), and target word six levels: “bead, bid, bayed, bed, bad” and “bod”). The dependent variable was the R AU-transformed percent-co rrect data for each listener. To obtain the percent-correct scor es for each listener, at each level of the independent variables, the number of corr ect responses was summed across the talkers within a group, with separate scores for each target vowel and speaking style. Percent correct was then computed based on the numbe r of possible correct responses (number of possible correct responses = number of talker s in a group X 2 tokens of each target word). All of the main effects and inter actions in the three-way ANOVA were significant. The F values, degrees of freedom and p values for all of the main effects and interactions are shown in Table 3, below. Only the talker group by speaking style interaction and the three-way interaction will be discussed in detail because these are the two effects that address the fi rst two research questions.

PAGE 48

38 Table 3 Results of Three-Way ANOVA of the Effects of Talker Group, Speaking Style and Target Word Effect F (df) p value Main effects Talker group 26.75 (2,38) <.001 Speaking style 50.06 (1,19) <.001 Target word 21.33 (5,95) <.001 Two-way interactions Talker group by speaking style 4.68 (2,38) .015 Talker group by target word 3.67 (10,190)<.001 Speaking style by target word 7.66 (5,95) <.001 Three-way interaction Talker group by speaking style by target word2.38 (10,190).011 Note. Data on F values, degrees of freedom (df) and levels of significance (p values) for all main effects and interactions in the three-way ANOVA of the effects of talker group, speaking style and target word on intelligibil ity of target words presented in noise to monolingual native English-speaking listeners. The significant talker group by speaking style interaction ad dresses the first research goal: to compare the three talker gr oups in terms of thei r overall clear speech

PAGE 49

39 benefit in noise. Percent-correct data s howing the performance of each talker group on each speaking style are shown in Figure 2. Figure 2. Mean percent-correct word-identificat ion scores for the th ree talker groups: monolingual (MO), early bilingual (EB) and la te bilingual (LB), for both speaking styles. Error bars indicate one standard error of the mean. The dashed line indicates conversational style performance for the m onolingual talker group. Asterisks indicate significant clear speech effects within each talker group. As shown in Figure 2, all three talker groups—monolingual (MO), early bilingual (EB) and late bilingual (LB)—were able to pr oduce some clear speech benefit. When comparing conversational speech to clear speec h, the MO group was 5% more intelligible in the clear speech condition; the EB group was 7% more intelligible in the clear speech condition; and the LB group was 3% more in telligible in the clea r speech condition. Overall, the early bilinguals performed be tter in both speech conditions, clear and 0 10 20 30 40 50 60 70 80 90 100 MOEBLBTalker groupPercent correct Conv ClearSpeaking style+5% -8 dB SNR +7% -8 dB SNR +3% -4 dB SNR * *

PAGE 50

40 conversational, than the other two groups. The MO and EB groups both had a similar degree of improvement in the clear speech condition, while the LB group had a smaller degree of improvement from conversational to clear It is important to keep in mind that the LB group was performing at a more favor able SNR of -4 dB, while the other two groups were performing at a more challenging SNR of -8 dB. As a result of this finding, one could say that the effect of being a late bilingual is e qual to about a 4 dB worse SNR for vowels when compared to the other two gr oups. Thus, the late bilinguals were less intelligible than the monolinguals and early bilinguals. Simple main effects post-hoc comparisons were used to explore the significant talker by speaking group interact ion. First, the two speaking styles were compared for each talker group. Performance in the clea r speech condition was significantly higher than performance in the conversational speech condition for each talker group ( p <.001 for the MO and EB groups and p =.018 for the LB group). Thus, each group showed a significant clear speech benefit, despite the di fferences across groups in the size of that benefit. Next, the performance of the MO and EB talker groups (the two groups for which the same SNR was used) were compared within each speaking style. Performance for the EB group was significantly high er than performance for th e MO group in both speaking styles. This difference was highly significant for both speak ing styles (p<.001), but was slightly larger for the clear speech styl e (an 8% difference between the EB and MO groups in the clear style vs. a 7% diffe rence between the EB and MO groups in conversational style).

PAGE 51

41 The significant three-way interaction addresses the second research goal: to compare the differences in the effects of cl ear speech across individual target vowels for each talker group. Percent-correct data show ing the performance of each listener group for each speaking style are shown in Figures 3-5 below. Two of the target words are shown in each figure, with each word shown as a separate panel. Simple main effects post-hoc comparisons were used to explore the significant talker group by speaking style by target word interaction and will be discusse d with the general patterns of results. All three talker groups produced a similar size clear speech bene fit for the target words “bead” (Figure 3A) and “bod” (Figure 5B). The simple main effects post-hoc comparisons showed that the clear speech be nefit was not significant for “bead” for any of the talker groups (benef it = +3 to +5%); however, fo r “bod,” the benefit was significant for all three groups (+11 to + 14%; MO: p=.001; EB/LB: p<.001). For the target word “bid” (Figure 3B), the monoli nguals and early bilingua ls each produced a 4% clear speech benefit (not significant); whereas, the late bilinguals performed worse in the clear speech condition, with a signifi cant decrement of 8% (p=.006). For the target word “bayed,” (Figure 4A) only the early bilinguals showed a significant clear speech benefit (+10%; p<.001) ; the other two groups showed a smaller, non-significant benefit. For the target wo rd “bed” (Figure 4B), only the monolingual talkers were able to produce a significant cl ear speech benefit (+9%; p=.01); the early and late bilingual talkers showed no change from conversational to clear. Finally, for the target word “bad” (Figure 5A), both early and late bilingual talkers, but not the monolingual talkers, were able to produ ce a significant clear speech benefit (+9%; p=.005 for the EB talkers; +7%; p=.041 for the LB talkers).

PAGE 52

42 Figure 3 Mean percent-correct word-identific ation scores for “bead” (panel A) and “bid” (panel B) for the three talker groups : monolingual (MO), early bilingual (EB) and late bilingual (LB). Error bars indicate one standard error of the mean. The dashed line indicates conversational style performance for the monolingua l talker group. Asterisks indicate significant clear speech e ffects within each talker group. 0 10 20 30 40 50 60 70 80 90 100 MOEBLBTalker groupPercent correct Conv ClearSpeaking style+5% -8 dB SNR +4% -8 dB SNR +3% -4 dB SNR A: "bead" 0 10 20 30 40 50 60 70 80 90 100 MOEBLBTalker groupPercent correct Conv ClearSpeaking style -8% -4 dB SNR +4% -8 dB SNR +4% -8 dB SNRB: "bid" *

PAGE 53

43 Figure 4 Mean percent-correct word-identificat ion scores for “bayed” (panel A) and “bed” (panel B) for the three talker groups : monolingual (MO), early bilingual (EB) and late bilingual (LB). Error bars indicate one standard error of the mean. The dashed line indicates conversational style performance for the monolingua l talker group. Asterisks indicate significant clear speech e ffects within each talker group. 0 10 20 30 40 50 60 70 80 90 100 MOEBLBTalker groupPercent correct Conv ClearSpeaking style +3% -8 dB SNR +10% -8 dB SNR +4% -4 dB SNRA: "bayed" 0 10 20 30 40 50 60 70 80 90 100 MOEBLBTalker groupPercent correct Conv ClearSpeaking style+9% -8 dB SNR 0% change -8 dB SNR 0% change -4 dB SNR B: "bed" *

PAGE 54

44 Figure 5 Mean percent-correct word-identificati on scores for “bad” (panel A) and “bod” (panel B) for the three talker groups: mono lingual (MO), early bilingual (EB) and late bilingual (LB). Error bars indicate one sta ndard error of the mea n. The dashed line indicates conversational style performance for the monolingua l talker group. Asterisks indicate significant clear speech e ffects within each talker group. 0 10 20 30 40 50 60 70 80 90 100 MOEBLBTalker groupPercent correct Conv ClearSpeaking style +7% -4 dB SNR +9% -8 dB SNR +1% -8 dB SNRA: "bad" * 0 10 20 30 40 50 60 70 80 90 100 MOEBLBTalker groupPercent correct Conv ClearSpeaking style+12% -4 dB SNR +14% -8 dB SNR +11% -8 dB SNR B: "bod" ** *

PAGE 55

45 Next, simple main effects post-hoc comparisons were used to compare performance of the MO and EB groups (the two groups for which stimuli were presented at the same SNR) for each target word, within each speaking style. In the conversational speech style, performance of the EB group wa s significantly higher than that of the MO group for the target words “bead, bid” a nd “bed.” The two groups did not differ significantly in performance for the targ et words “bayed, bad” and “bod.” In the clear speech style, performance of the EB group was significantly higher than that of the MO group for the target words “bead, bid, bayed,” and “bad.” Performance of the EB group was also highe r for the target word “bed,” but the difference only approached significance ( p =.06). The performan ce of the two groups did not differ significantly for the target word “bod.” To address the third research question, th e performance of the individual talkers was compared within each talker group. Percent-correct scores were computed separately for each talker w ithin a group for each of the tw o speaking styles. To obtain this value, the number of correct responses for each talker was summed across the six target vowels and two tokens per vowel for each speaking style (number of possible correct responses: six vowels X two tokens per vowel = 12), with a separate score for each listener. Percent-correct data showing the performance of each talker within each listener group on each speaking style are shown in Figures 6-8. Results for each talker group are presented as a separate figure.

PAGE 56

46 Figure 6 Mean percent-correct word-identific ation scores of each talker in the monolingual (MO) talker group. Performan ce for the conversational style (Conv) is shown by solid lines and performance for the clear style (Clear) is shown by dashed lines. The dashed horizontal line indicates chance performance (approximately 17% correct) for the six-alternative forced choice task. The difference in performance between clear and conversational speaking styles is indicated for the talkers with the greatest increases and decreases in performance from the convers ational to the clear speaking style. 0 10 20 30 40 50 60 70 80 90 100 234691012141520222324TalkerPercent correct Conv ClearSpeaking style +22% -11% Monolingual talkers

PAGE 57

47 Figure 7 Mean percent-correct word-identificat ion scores of each ta lker in the early bilingual (EB) talker group. Performance fo r the conversational styl e (Conv) is shown by solid lines and performance for the clear st yle (Clear) is shown by dashed lines. The dashed horizontal line indicates chance perfor mance (approximately 17% correct) for the six-alternative forced choice task. The di fference in performance between clear and conversational speaking styles is indicated for the talkers with the greatest increases and decreases in performance from the convers ational to the clear speaking style. 0 10 20 30 40 50 60 70 80 90 100 245681011121416171819202324252629303133TalkerPercent correct Conv ClearSpeaking style +24% +24% -11% Early bilingual talkers

PAGE 58

48 Figure 8 Mean percent-correct word-identificat ion scores of each ta lker in the late bilingual (LB) talker group. Performance fo r the conversational styl e (Conv) is shown by solid lines and performance for the clear st yle (Clear) is shown by dashed lines. The dashed horizontal line indicates chance perfor mance (approximately 17% correct) for the six-alternative forced choice task. The di fference in performance between clear and conversational speaking styles is indicated for the talkers with the greatest increases and decreases in performance from the conve rsational to the clear speaking style. As shown in Figure 6 (MO talker group) 9 of the 13 monolingual talkers (69%) showed improvement from conversational to clear speech conditions; 3 talkers (23%) performed worse in the clear speech conditi on; and 1 talker (8%) performed the same across speech conditions. Five talkers (38 %) produced a clear speech intelligibility 0 10 20 30 40 50 60 70 80 90 10013679101112131516181921TalkerPercent correct Conv ClearSpeaking style +20% -13% Late bilingual talkers

PAGE 59

49 benefit of more than 5%. The greatest improvement was seen for subject NA02, who improved by 22% from conversational to clear. The greatest decrement was seen in subject NA04, who was 11% less intelligib le in the clear speech condition. Conversationally, the monolingua ls ranged in intelligibil ity from 25% to 66% (41% intelligibility range in convers ational speech). In the clear speech condition, they ranged from 30% to 78% (48% intelligibility ra nge in clear speech). As shown in Figure 7 (EB talker group), 16 of the 22 early bi lingual talkers (73%) showed improvement from conversational to clear speech conditions; 5 talkers (23%) performed worse in the clear speech conditi on; and 1 talker (5%) performed the same across speech conditions. Thirte en talkers (59%) produced a cl ear speech in telligibility benefit greater than 5%. The greatest impr ovement was seen for subjects HI06 and HI23, who each improved by 24% from conversational to clear. The greatest decrement was seen in subject HI17, who was 11% less inte lligible in the clear speech condition. Conversationally, the ea rly bilinguals ranged in intellig ibility from 27% to 82% (55% intelligibility range in convers ational speech). In the clear speech condition, they ranged from 26% to 74% (47% intelligibility ra nge in clear speech). As shown in Figure 8 (LB talker group), 8 of the 14 late bili ngual talkers (57%) showed improvement from conversational to clear speech conditions, and 6 talkers (43%) performed worse in the clear speech conditi on. Five talkers (36%) produced a clear speech intelligibility benefit of more than 5%. The greatest improvement was seen for subject LO03, who improved by 20% from co nversational to clear. The greatest decrement was seen in subject LO11, who was 13% less intelligible in the clear speech condition. Conversationally, the late bilinguals ranged in inte lligibility from 21% to 72%

PAGE 60

50 (51% intelligibility range in conversational speech). In the clear speech condition, they ranged from 18% to 70% (52% intellig ibility range in clear speech). While there is variability amongst individua l talkers in all thre e talker groups, it can be concluded that the monolinguals and ea rly bilinguals performed most similarly to each other in both speech conditions; however the early bilinguals were found to be more intelligible overall in both speech condition s. Additionally, the late bilinguals were the least intelligible in both speech conditi ons, with two individual talkers performing only slightly above chance.

PAGE 61

51 Chapter Four Discussion The goal of oral communication is for th e talker’s ideas to be understood by the listener. Intelligibility is one way of measuring whether the message was transmitted successfully. Sometimes the talker’s intent of the message is misunderstood by the listener. This misunderstanding may be the resu lt of a decrease in the intelligibility of the message. Two barriers that can affect inte lligibility are bilingualism and level of background noise. When two people are having a conversation and there is a high level of background noise, such as in a crowded re staurant, chances are that parts of the conversation will be less in telligible, and thus, a breakdown in communication may occur. Imagine this same situation, but im agine that one of the communication partners is bilingual and they are communicating in the bilingual’s second language. The environmental factor of noise from the rest aurant compounded by the talker factor of bilingualism may further cont ribute to reduced intellig ibility of the message. Some bilinguals speak their second langua ge with a certain degree of foreign accent. This may pose a problem for these i ndividuals in their daily lives, whether at school, at work, at home, or in the social realm. They may fi nd that they have to repeat themselves frequently because their listeners may have diffi culty understanding them. In turn, this may also lead to increased levels of frustration for both the non-native talker and the listener. For these reasons, ongoing res earch is needed to a ddress the issues that present difficulty to both na tive and non-native talkers of the language. The present

PAGE 62

52 study adds to our understanding of the problem because the smaller cl ear speech benefit found for late bilinguals suggests that they ma y lack knowledge or control of which cues to enhance when asked to speak more cl early, reflecting differences in linguistic knowledge. Understanding the degree to whic h bilinguals can improve intelligibility when asked to speak clearly can be useful in accent modification therapy for those bilinguals who seek it, which can help comb at the problem of reduced intelligibility. The present study addressed the issues of bilingualism and noise through the following research goals. First, a compar ison across three talker groups, monolinguals, early bilinguals, and late bili nguals, was made to determine the differences in native and non-native talkers’ ability to produce a cl ear speech benefit in noise. Second, a comparison was made of the differences in the effects of cl ear speech across six individual target vowels for each talker group. Third, the performance of individual talkers was compared within each group. Summary of Results In the present study, there were three inde pendent variables: talker group, vowel, and speech condition. The talker group variable consisted of three levels: native English monolinguals, early Spanish/English bilinguals, and late Spanish/English bilinguals. The vowel variable had six levels: /i, I, e, , a/. The speech condition variable had two levels: conversational and clea r speech. All three talker groups were recorded while reading lists of carrier phrase s. All carrier phrases were embedded with a /bVd/ word which contained the six vowel variations. Reco rdings were made in two different speech conditions, conversational and cl ear. All /bVd/ words were isolated, mixed with noise, and presented to a group of native English-speaking mono lingual listeners via a six-

PAGE 63

53 alternative forced-choice perception task. Th e dependent variable in this study was the percent of words correctly identified by the listeners. To address the first research goal of talker group by speaking style, it was found that all three talker groups were able to produce a significant clear speech benefit, although the benefit was only about half as much for the late bilinguals as for the other two groups. It should be kept in mind also th at the late bilinguals were performing at a less challenging SNR than the other two groups, due to their overall lower intelligibility level. The degree of clear speech benefit obtained was similar for the monolingual and early bilingual talker groups; however, overa ll, the early bilinguals were significantly more intelligible than the monolingual talkers in both th e conversational and clear speaking styles. The fact that the early bilinguals behave d similarly to the monolinguals was not a surprise. Previous research has proven that the earlier a second language is acquired, the more native-like the talker will become in the L2. Since the early bilinguals in the present study all reported lear ning English intensively prior to age 12 years, many of them seemed to have achieved native-like pr oficiency in English. The late bilinguals, on the other hand, all reporte d learning English inte nsively after the age of 15 years, which would explain why they were less intelligible overall. It was, how ever, an interesting finding that the early bilinguals performed si gnificantly better than the monolinguals. This can possibly be attributed to the fact that bilinguals have had to manipulate two or more languages for most of their lives. As a result, bilinguals may have greater metalinguistic awareness, which in turn, may have enabled them to more easily manipulate the acoustic cues for the vowel studied here. Alte rnatively, the early

PAGE 64

54 bilinguals may have had a larger vowels spac e, due to the need to maintain phonetic categories for both L1 and L2 vowels (cf. Flege, 1995). To address the second research goal of talker group by speaking style by individual vowel, four general patterns we re found. First, for “bid,” the monolinguals and early bilinguals better in the clear th an in the conversational speech condition, although the benefit was not si gnificant for either group. The late bilingual talkers, however, actually performed significantly wors e for “bid” in the cl ear speech condition than in the conversati onal speech condition. The second vowel pattern was shown for “be d.” Results indicated that only the monolinguals were able to produce a significan t clear speech benefit. The two bilingual groups performed similar to each other in that there was no change in performance across speech conditions for both groups of bilinguals. Overall, the early bilinguals performed better than the monolinguals here, as well, however. Third, for “bad” and “bayed,” the early bilinguals showed a the largest clear sp eech benefit. For “bayed,” the benefit was significant only for the early bilinguals, but for “bad,” both the early and late bilingual talkers (but not the monolingual talkers) show ed a significant clear speech benefit. Finally, for “bead” and “bod,” all three gr oups showed a simila r degree of clear speech benefit, but the benefit was only significant for “bod.” The fact that the late bilinguals had a significant decrement for “bid” was an interesting finding. Since confusion analyses have not been completed yet, one can only speculate as to why this occurred. Most like ly, the target word “bid” was confused most often with “bead.” According to Flege’s (1997) Speech Learning Model, /I/ would be considered a “new” vowel for Spanish learners of English, while /i/ would be considered

PAGE 65

55 an “old” vowel. The vowel /i/ is longer in duration than /I/. On the production task, when instructed to use clear speech, the late bilingual talkers may have manipulated the duration cue by slowing down their rate of speech, and thus, lengthening the vowel. If they did this when producing the vowel /I/, they may have actually produced it acoustically more /i/-like, wh ich would result in confusions with /i/, and thus, would explain the significant decrement in the clea r speech condition for /I/. This hypothesis can be investigated by confusion analyses of the perception data a nd by acoustic analyses of the production data. The finding that neither bilingual group was able to produce a clear speech benefit for /bed/ was interesting. The hypothesis that / / is a “new” vowel to Spanish learners of English could possibly explain this resu lt. The bilinguals may have had difficulty accessing the appropriate cues to produce this vowel due to a limited previous experience with it. The last vowel pattern was expect ed. All three talker groups were able to produce a significant clear speech benefit for “bod.” Since this vowel occurs in both English and Spanish, it was not a surprise that all three groups were able to utilize the appropriate cues to make themselves more intelligible for this vowel. To address the third research goal of examining individual talker differences within each group across speech conditions, it was found that there was variability amongst individual talkers within each group. For the monolinguals, many of the talkers showed improvement in the clear speech c onditions, but there were a few who performed worse. Among the early bilinguals, there were also many talkers who improved from conversational to clear, while others showed the opposite pa ttern. Finally, for the late bilinguals, a similar degree of variability in performance across talkers was found,

PAGE 66

56 although overall, their clear speech benefit was smaller than for the other two groups. There were two individual talk ers who were performing slightly above chance in the late bilingual group. The results from the third research goal were not unexpected. Every individual is unique and there may be a combination of fact ors that contributed to the variability found amongst talkers across all three groups. Some of these fact ors can include vocal quality, fundamental frequency, vocal inte nsity, vocal resonance, linguist ic experience, etc. It is obvious that there are a number of factors that may contribut e to individual variability. Hypotheses At the start of the present study, the inve stigators formulated three hypotheses. First, they hypothesized that there woul d be no change in vowel production across conversational and clear speech conditions for th e late bilinguals for some target vowels. This hypothesis was supported to some de gree. Although no talker group showed a significant clear speech benefit fo r all six target words, only the late bilinguals showed a significant decrement for one target word. Mo reover, the average de gree of benefit for the late bilinguals was only a bout half that of the other tw o groups. The late bilinguals did produce a significant clear speech benef it for two vowels // and /a/, however. The second hypothesis was that the early bilinguals would produce a clear speech benefit in noise. This hypothesis was also supported. The early bilinguals did in fact produce a significant clear speech benefit in noise overall that was similar in size to that for the monolingual talkers. Moreover, although the clear speech benefit was only significant for target vowels /e / and /a/, performance was higher for the early bilingual talkers than for the mono lingual talkers for half of the target vowels.

PAGE 67

57 Lastly, it was hypothesized that the nativ e talkers would also produce a clear speech benefit in noise. This hypothesis was su pported as well. The native talkers were able to produce a clear speech benefit in noise overall. However, like the early bilinguals, this benefit was only significant for some target vowels (/ / and /a/). Comparison with Previous Research Previous research comparing intelligibili ty of vowels in conv ersational and clear speech with monolingual talkers was been conducted by Ferguson and Kewley-Port (2002) and Ferguson (2004). Ferguson and Ke wley-Port (2002) was the first study to systematically investigate vowel identification in noise in both clear and conversational speech. In Ferguson (2004), recordings were made of multiple native talkers, reading sentences containing /bVd/ words in clear and conversational speech. The words were later isolated from the sentence and embedde d in 12-talker babble for presentation to native listeners in a vowel identification task. The present study is very similar to Ferguson (2004) in that /bVd/ words were produced both clearly and conversationally, their stimuli were embedded in multi-talker ba bble, and they presented their task to native listeners. While Ferguson (2004) addressed th e linguistic factor of speaking style, the talker factor of bilingualism was not addressed. When co mparing the two studies, the major differences are that the Ferguson (2004) study used only native talkers and ten target vowels, while the pres ent study used both native and non-native talkers and only six target vowels. Ferguson found that their native talkers were able to produce a clear speech benefit of 8%. In the present study, a clear speech benefit was found for all talker groups, but it was smaller than the benefit found in Ferguson (2004). Native talkers in the present study produced a clear speech benef it of about 5%, early bilinguals produced

PAGE 68

58 a 7% benefit, and late bilinguals produced a 3% benefit. Some possible reasons as to why the native talker s in the present study did not produce the same benefit as in the Ferguson (2004) study could be that there was a smaller number of talkers. The present study only had 13 native talkers, while the previous study had 41 native talkers. Individual talker differences may have had a greater impact on the present study than in the Ferguson (2004) study. Also, all of th e talkers in the present study were female, while Ferguson (2004) included both male and fe male talkers. The factors of sample size, number of vowels studied and gender may explain why differences were found across these studies. Bilingual research on clear speech perception has been conducted by Bradlow and Bent (2002). In their study, native English ta lkers were recorded reading sentences in both clear and conversational speech. The sentences were later embedded in white noise and presented to native and non-native listene rs in a sentence recognition task. Results indicated that non-native listeners received a much smaller clear speech benefit in noise than did native listeners. Th is study paralleled the present study in that both clear and conversational speech was presented to lis teners and the bilingualism factor was addressed; however, Bradlow and Bent ( 2002) used native talkers and native and nonnative listeners, while the pr esent study used native a nd non-native talkers and only native listeners. Also, Bradlow and Bent ( 2002) did not group thei r bilingual subjects by age of acquisition of English, wh ile the present study did. Sin ce early learners tend to be much more native-like than late learners (R ogers, et al. 2006), the question of whether a clear speech benefit differs amongst bilingual li steners from various L2 proficiency levels was not addressed by Bradlow and Bent ( 2002). Bradlow and Bent (2002) found that

PAGE 69

59 non-native listeners received less of a clea r speech benefit in noi se than did native listeners. In the present study, it was found th at early bilinguals produced a clear speech benefit that was similar to that produced by monolinguals, while late bilinguals produced a much smaller clear speech benefit than m onolinguals. Together, the results of Bradlow & Bent (2002) and the present study suggest th at late bilinguals ar e faced with greater challenges than early bilinguals when comm unicating in noisy environments due to differences in their ability both to receive a clear speech benefit when listening and to produce a clear speech benefit when speaking. The present study can also be assessed with regard to its implications for Flege’s (1995) Speech Learning Model. In Flege’s model there is a dis tinction between “old” and “new” vowel categories for second langua ge learners. In the present study, a comparison can be made of “old” versus “new ” vowels for the bilingual talkers. For the three “old” vowels, /i, e/ a nd /a/, both bilingual groups were able to produce a clear speech benefit for all three of these vowels. However, for the early bilinguals, this benefit was only significant for the vowels /e/ an d /a/. For the late bi linguals, this benefit was only significant for the vowel /a/. For the three “new” vowels, /I, / and //, neither bilingual group was able to produce any cha nge across speech condi tions for the vowel / /. For the vowel /I/, the early bilinguals produced a small, but not significant clear speech benefit and the late bi linguals actually performed signi ficantly worse in the clear speech condition. Finally, for the vowel //, both bilingual groups were able to produce a significant clear speech benefit.

PAGE 70

60 Implications Theoretically, early bilinguals seem to have the productive li nguistic knowledge needed to improve intelligib ility when asked to speak cl early, but the late bilinguals appear not to share this same knowledge or seem less able to acce ss it to manipulate the relevant acoustic cues. One hypothesis is that early bilinguals may have better awareness of language due to the early need to manage two languages. It is also possible that the early bilinguals are more caref ul in this type of task than the late bilinguals. Practically, the early bilinguals do not seem to be at a disadvantage when talking in noisy listening environments, in relation to the monolinguals. In fact, as mentioned previously, they actually perf ormed better than the monoli nguals. However, the late bilinguals appear to be at an increased disadvantage when talking under adverse listening conditions. Under normal listening conditions, th e late bilinguals are at a disadvantage to begin with, and when noise is added to the background signal, they appear to be at an even greater disadvantage by experiencing a reduced ability to make themselves more intelligible to compensate for the challenging environment. Clinically, the results of this research can be useful to bilinguals who wish to improve their intelligibility under adverse li stening conditions. For example, a late bilingual professor, who teaches in hi s non-native language, may be difficult to understand by many of his students. The professor may wish to improve his intelligibility for the benefit of himself and hi s students. As a result, he may wish to enroll in an accent modification program at the local university, where he would be provided with therapy from a speech-language pa thologist (SLP), specializing in foreign accent.

PAGE 71

61 Initially, the SLP may review communication strategies with the professor. These strategies may include: slow down rate of speech, repeat and rephrase utterances, summarize at the end of each lecture to ensure the stude nts understood the material presented, use visual aids such as a PowerPoi nt presentation or handouts, make sure there is nothing obstructing the m outh while talking, engage in face-to-face communication, maintain good eye contact with the students, stand in good lighting wh ere there is not a lot of glare, and if possible, lecture in a room that ha s good acoustics such as carpeting and heavy curtains. It is also important to reduce the amount of b ackground noise in the classroom in order to improve the signal-tonoise-ratio. If necessa ry, the professor can also use a microphone. Many of these strategies could be applied in ot her settings for the professor, as well. Following communication strategies traini ng, the SLP could then implement an accent modification program, that should be backed up by evidence-based practice such as the evidence presented in th e present study, in particular in formation as to which of the vowels studied would be most likely to be mi sperceived. SLPs can provide services to their clients by training them on specific phoneme cues that are needed to improve intelligibility. Training such as this can ha ve an enormous impact on many individuals who use their voice professionally includi ng professors, company CEOs, customer service representatives, political candidates, actors/actresses, and regular everyday people who wish to improve their accent in orde r to improve their quality of life. Limitations Some limitations to the present study include the following: all talkers were female so it is not known how male talkers w ould perform on this ta sk; therefore, this

PAGE 72

62 research may not be clinically applicable to male bilinguals who wish to improve their intelligibility. Another limitation was that only six vowels of American English were studied, when there are 14 vowels in the languag e. It is not known which cues the talkers will manipulate for the other eight vowels. The present study only investigated clear speech effects for monolinguals and Spanis h-English bilinguals. There are many bilinguals who speak languages other than Spanish, and those languages were not explored; therefore, the result s of this research can only be applied directly to SpanishEnglish bilinguals. In the present study, the speech produced by all talkers in the conversational style was not completely natural. Although the par ticipants were instruct ed to read sentences in a normal, conversational manner, it was not true conversation since they were reading lists of carrier phrases. Normal conversation is usually full of various intonation patterns, as well as, different types of sentences (decla ratives, interrogatives, exclamations, etc.). One final limitation was that the talkers were not talking under adverse listening conditions in the clear speech style. Instea d, adverse listening conditions were simulated by instructing the talkers to pretend they were speaking to some one with a hearing impairment or to someone who was having difficulty understanding them. Therefore, it is not known, if the clear speech produced by the talkers is reflective of the clear speech they would actually use in a noise or other demanding listening environments. Future Research Future research in the near future should begin with the confusion matrices for the perception data collected in the present study. This will provide information on what types of errors the talkers are making. Next acoustic analyses should be completed on

PAGE 73

63 all of the production data in both clear and conversational styles. This will provide information on which acoustic cues, speci fically, are being manipulated by the monolingual and bilingual talkers when instru cted to speak clearly. This information would also be helpful in designing an evidence-based accent modification program. Other future research could include re plicating the present study to compare whether the same results are f ound. Replications can also incl ude other variables such as young and old listeners, other vowels, cons onants, and sentences. Other foreign languages, besides Spanish, can be investigat ed, as well, in order to examine whether differences exist in vowel pr oduction across languages and what those differences are. One other possible suggestion that would be interesting would be to conduct an experimental treatment design, where late bi linguals are instructed to read sentences using clear and conversational speech at base line; their production data can be embedded in noise and presented to a group of native lis teners via a vowel iden tification task; next, the late bilinguals can be provided with co mmunication strategies training and accent modification therapy; then, the late bilinguals can participate in the same experiment at baseline; and finally, their da ta after treatment can be em bedded with noise and readministered to the same group of native listene rs in a vowel identification task, as done previously. An experiment such as this could be beneficial because it would either support or fail to support the effectivene ss of communication strategies training and accent modification therapy for late bilinguals.

PAGE 74

64 References Bilger, R. C., Neutzel, J. M., Rabinowitz, W. M., & Rzeczkowski, C. (1984). Standardization of a test of speech perception in noise. Journal of Speech and Hearing Research, 27 32-48. Boersma, P. & Weenik, P. (2006). Praat: Doing phonetics by com puter (Version 4.4.24) [computer program]. Retrieved June 19, 2006 from http://www.praat.org. Bohn, O. S., & Flege, J. E. (1997). Percepti on and production of a new vowel category by adult second language learners. In A. James & J. Leather (Eds.), Secondlanguage speech: Structure and process (pp. 53-73). Berlin: Mouton de Gruyter. Bradlow, A. R., & Bent, T. (2002). The clea r speech effect for non-native listeners. Journal of the Acoustical Society of America, 112 272-284. CoolEdit 2000 (version 1.1) [compu ter software] (2000). Syntrill ium, Inc.: Phoenix, AZ. Dalbor, J. B. (1969). Spanish pronunciation: Theory and practice (3rd ed.). Fort Worth: Holt, Rinehart and Winston. Ferguson, S. H. (2004). Talker differences in clear and conversational speech: Vowel intelligibility for normal-hearing listeners. Journal of the Acoustical Society of America, 116 2365-2373. Ferguson, S. H., & Kewley-Port, D. (2002) Vowel intelligibi lity in clear and conversational speech for normal-hearing and hearing-impaired listeners. Journal of the Acoustical Society of America, 112 259-271. Flege, J. E. (1995). Second language speech learning theory, findings, and problems. In W. Strange (Ed.), Speech perception and linguistic exp erience: Issues in crosslanguage research (pp. 233-277). Baltimore: York Press.

PAGE 75

65 Flege, J. E., Bohn, O. S., & Jang, S. ( 1997). Effects of experience on non-native speakers’ production and perception of English vowels. Journal of Phonetics, 25 437-470. Grosjean, F. (1989). Neurolingu istics, beware! The bilingual is not two monolinguals in one person. Brain and Language, 36 3-15. Hillenbrand, J., Getty, L. A., Clark, M. J, & Wheeler, K. (1995). Acoustic characteristics of American English vowels. Journal of the Acoustical Society of America, 97 3099-3111. Hillenbrand, J., & Nearey, T. M. (1999). Identification of re-synthesized /hVd/ utterances: Effects of formant contour. Journal of the Acoustical Society of America, 105 3509-3523. Kent, R. D., Dembowski, J., & Lass, N. J. (1996). The acoustic characteristics of American English. In N. J. Lass (Ed.), Principles of experimental phonetics (pp. 185-225). St. Louis: Mosby-Year Book, Inc. Kewley-Port, D., Akahane-Yamada, R., & Aika wa, K. (1996). Intelligibility and acoustic correlates of Japanese accented English vowels. ICSLP 96 (International Conference on Spoken Language Processing) Proceedings. Ladefoged, P. (1982). A course in phonetics (2nd ed.). San Diego: Harcourt, Brace Jovanovich. Miller, G. A., Heise, G. A., & Lichten, W. (1951). The intelligibility of speech as a function of the context of the test materials. Journal of Experimental Psychology, 41 329-335. Picheny, M. A., Durlach, N. I., & Braida, L. D. (1985). Speaking clearly for the hard of

PAGE 76

66 hearing I: Intelligibility differences between clear and conve rsational speech. Journal of Speech and Hearing Research, 28 96-103. Picheny, M. A., Durlach, N. I., & Braida, L. D. (1986). Speaking clearly for the hard of hearing II: Acoustic char acteristics of clear a nd conversational speech. Journal of Speech and Hearing Research, 29 434-446. Rogers, C. L., & Dalby, J. (2005). Forced-cho ice analysis of segmental production by Chinese-accented English speakers. Journal of Speech, Language, and Hearing Research, 48 1-17. Rogers, C. L., Dalby, J., & Nishi, K. ( 2004). Effects of noise and proficiency on intelligibility of Chinese-accented English. Language and Speech, 47 139-154. Rogers, C. L., Lister, J. J., Febo, D. M., Be sing, J. M., & Abrams, H. B. (2006). Effects of bilingualism, noise, and reverberati on on speech perception by listeners with normal hearing. Applied Psycholinguistics, 27, 465-485. Small, L. H. (1999). Fundamentals of phonetics: A practical guide for students Boston: Allyn & Bacon. Smith, D. D. (2007). Introduction to special educa tion: Making a difference (6th ed.). Boston: Pearson/Allyn & Bacon. Studebaker, G. (1985). A “rati onalized” arcsine transform. Journal of Speech and Hearing Research, 28 494-509. TDT System III [Computer hardware and soft ware] (2001). Tucker-Davis Technologies, Inc.: Gainesville, FL. Tye-Murray, N. (2004). Foundations of aural rehabilitati on: Children, adults, and their family members (2nd ed.). Australia: Thomas Delmar Learning.

PAGE 77

67 United States Census Bureau (2000). 2000 U. S. Census. Retrieved July 16, 2007, from http://www.factfinder.census.gov

PAGE 78

68 Appendices

PAGE 79

69 Appendix A.1: Consent Form for Production Experiment (Monolingual) Social Sciences/Behavioral Adult Informed Consent University of South Florida Information for People Who Take Part in Rese arch Studies For monolingual recordi ng and listening purposes—compensated The following information is being presented to help you decide whether or not you want to be a part of a minimal risk research study. Ple ase read carefully. If you do not understand anything, ask the Person in Charge of the Study. Title of Study: Speech perception and production by native and non-native speakers Principal Investigator: Catherine L. Rogers, Ph.D. Study Location(s): PCD 3008A You are being asked to participate because you are a monolingual native speaker of English, with normal hearing and no history of speech or hearing disorders. General Information about the Research Study The purpose of this research study is to examin e performance of native and non-native speakers of English in both listening and speaking tasks. Audio recordings of your speech will be made and presented to listeners in two additional studies for comparison with foreign-accented speech. Plan of Study To participate in this experiment you must fi rst pass a basic audiometric hearing screening. During the hearing screening, a series of short tones (beeps) of different frequencies will be played. Your task is to indicate when you hear a sound, by raising your hand. If you do not pass the hearing screening, it does not necessarily m ean that you have a hearing loss, however you may wish to pursue further testing at a doctor’s o ffice or clinic. The main experiment will take place in two sessions, on two separate days. On the first day, audio recordings of your speech will be made as you repeat a set of sentences in English that will be presented to you on a computer screen. Some of the sentences will al so be presented over headphones. For the audio recordings, you will speak into a microphone, co nnected to a digital recorder. The audio recordings of your voice will be stored on a com puter. Later, portions of these recordings will be presented to listeners in two additional studies. During the main portion of the study you will be asked to produce about 95 sentences. Each will be presented on a computer screen. You will have several practice items before beginning th e main task. You will be instructed in the pronunciation of any nonsense words. Following th is session, you may be asked to repeat some sentences more clearly. Next, you will be asked to listen to English words. On each of the 108 trials, you will hear a single word. Your task will be to choose the word you heard from several choices. The first day’s testing and completion of this form will take approximately two hours. On the second day, you will again be asked to liste n to English words and to choose the word you heard from several choices. Some of the words ma y have been altered from their original form. Completion of the 240 trials on the second day shoul d take about one and a half hours. On both days you will be given a short break after each 20 minutes of listening or speaking. Total time for participation will be about three and a half hours.

PAGE 80

70 Payment for Participation As a subject, you may receive either extra credit points (if you are enrolled in a participating course in a participating department) or monetary compensation for your participation. If you are to receive monetary compensation, you will be paid on an hourly basis ($10.00/hour) for your participation in the study. If you are to receive extra credit points, you will receive 1 point for every 30 minutes of participation. You will be compensated at the conclusion of the research study sessions. If you decide to withdraw or the experimenter decides to terminate your participation, you will be compensated only for th e hours of participation in the research study session prior to withdrawal/termination. Benefits of Being a Part of this Research Study No practical benefits of this study are likely to apply to you directly. On the other hand, by taking part in this research study, you may in crease our overall knowledge of human listeners’ ability to understand speech under different lis tening conditions. This knowledge may aid in developing improved methods of teaching English to speakers of English as a second language. Risks of Being a Part of this Research Study There are no known risks related to participation in this research study. The speech you will listen to will not be loud enough to be harmful under normal conditi ons. The level used will be about that of speech at a normal conversational level. Also, you are encouraged to notify the investigator immediately if any of the sounds are uncomfortable. If the situation is uncomfortable in any way, the experimental procedure will be changed. Confidentiality of Your Records Your privacy and research records will be kept conf idential to the extent of the law. Authorized research personnel, employees of the Department of Health and Human Services and the USF Institutional Review Board may inspect th e records from this research project. The results of this study may be published. However, the data obtained from you will be published using a code number or will be comb ined with data from other people in the publication. The published results will not include your name or any other information that would in any way personally identify you. As stated a bove, the sentences you produce will be presented to listeners in two additional studies. Your name will not be given to the listeners. No other personal information will be given to the listeners. Individual data will be stored and identified usi ng a number code. All data will be stored in the Speech Perception and Production Laboratory (PCD 3008A) or in the principal investigator’s office (PCD 4013). Both locations will be locked and secured. Authorized research investigators, agents of the Department of H ealth and Human Services and the USF Institutional Review Board may inspect your records from this research project. Volunteering to Be Part of this Research Study Your decision to participate in this research st udy is completely voluntary. You are free to participate in this research study or to withdraw at any time. If you choose not to participate, or if you withdraw, there will be no penalty or loss of benefits that you are entitled to receive. If participating for extra credit points, you will be credited for the time you participated and will receive no grade penalty for withdrawing. If pa rticipating for money, you will be compensated for the time (hours or part hours) that you participat ed at the rate stated above ($10.00 per hour).

PAGE 81

71 Questions and Contacts If you have any questions about this research st udy, contact Catherine Rogers, Ph.D. at (813) 974-7423. If you have questions about your rights as a pers on who is taking part in a research study, you may contact a member of the Division of Research Compliance of the University of South Florida at 813-974-5638. Your Consent—By signing this form I agree that: I have fully read or have had read and explaine d to me this informed consent form describing a research project. I have had the opportunity to question one of the persons in charge of this research and have received satisfactory answers. I understand that I am being asked to participate in research. I understand the risks and benefits, and I freely give my consent to participate in th e research project outlined in this form, under the conditions indicated in it. I have been given a signed copy of this info rmed consent form, which is mine to keep. _______________________ _______________________ _____________ Signature of Participant Printed Name of Participant Date Investigator Statement I have carefully explained to the subject the nature of the above protocol. I hereby certify that to the best of my knowledge the subject signing th is consent form understands the nature, demands, risks and benefits involved in participating in this study. _____________________ ______________________ ____________ Signature of Investigator Printed Name of Investigator Date Or Authorized research investigators designated by the Principal Investigator Institutional Approval of Study and Informed Consent This research project/study and informed c onsent form were reviewed and approved by the University of South Florida Institutiona l Review Board for the protection of human subjects. This approval is valid until the date provided below. The board may be contacted at (813) 974-5638. Approval Consent Form Expiration Date: Revision Date:_______________

PAGE 82

72 Appendix A.2: Consent Form for Production Experiment (Bilingual) Social Sciences/Behavioral Adult Informed Consent University of South Florida Information for People Who Take Part in Rese arch Studies For bilingual recording and li stening purposes—compensated The following information is being presented to help you decide whether or not you want to be a part of a minimal risk research study. Ple ase read carefully. If you do not understand anything, ask the Person in Charge of the Study. Title of Study: Speech perception and production by native and non-native speakers Principal Investigator: Catherine L. Rogers, Ph.D. Study Location(s): PCD 3008A You are being asked to participate because you are a bilingual speaker of Spanish and English or speaker of English as a second language, with normal hearing and no history of speech or hearing disorders. General Information about the Research Study The purpose of this research study is to examin e performance of native and non-native speakers of English in both listening and speaking tasks. Audio recordings of your speech will be made and presented to listeners in two additional studies. Plan of Study To participate in this experiment you must fi rst pass a basic audiometric hearing screening. During the hearing screening, a series of short tones (beeps) of different frequencies will be played. Your task is to indicate when you hear a sound, by raising your hand. If you do not pass the hearing screening, it does not necessarily m ean that you have a hearing loss, however you may wish to pursue further testing at a doctor’s o ffice or clinic. The main experiment will take place in two sessions, on two separate days. On the first day, audio recordings of your speech will be made as you repeat a set of sentences in English that will be presented to you on a computer screen. Some of the sentences will al so be presented over headphones. For the audio recordings, you will speak into a microphone, co nnected to a digital recorder. The audio recordings of your voice will be stored on a com puter. Later, portions of these recordings will be presented to listeners in two additional studies. During the main portion of the study you will be asked to produce about 95 sentences. Each will be presented on a computer screen. You will have several practice items before beginning th e main task. You will be instructed in the pronunciation of any unfamiliar items. Following th is session, you may be asked to repeat some sentences more clearly. Next, you will be asked to listen to English words. On each of the 108 trials, you will hear a single word. Your task will be to choose the word you heard from several choices. The first day’s testing and completion of this form will take approximately two hours. On the second day, you will again be asked to liste n to English words and to choose the word you heard from several choices. Some of the words ma y have been altered from their original form. Completion of the 240 trials on the second day shoul d take about one and a half hours. On both

PAGE 83

73days you will be given a short break after each 20 minutes of listening or speaking. Total time for participation will be about three and a half hours. Payment for Participation As a subject, you may receive either extra credit points (if you are enrolled in a participating course in a participating department) or monetary compensation for your participation. If you are to receive monetary compensation, you will be paid on an hourly basis ($10.00/hour) for your participation in the study. If you are to receive extra credit points, you will receive 1 point for every 30 minutes of participation. You will be compensated at the conclusion of the research study sessions. If you decide to withdraw or the experimenter decides to terminate your participation, you will be compensated only for th e hours of participation in the research study session prior to withdrawal/termination. Benefits of Being a Part of this Research Study No practical benefits of this study are likely to apply to you directly. On the other hand, by taking part in this research study, you may in crease our overall knowledge of human listeners’ ability to understand speech under different lis tening conditions. This knowledge may aid in developing improved methods of teaching English to speakers of English as a second language. Risks of Being a Part of this Research Study There are no known risks related to participation in this research study. The speech you will listen to will not be loud enough to be harmful under normal conditi ons. The level used will be about that of speech at a normal conversational level. Also, you are encouraged to notify the investigator immediately if any of the sounds are uncomfortable. If the situation is uncomfortable in any way, the experimental procedure will be changed. Confidentiality of Your Records Your privacy and research records will be kept conf idential to the extent of the law. Authorized research personnel, employees of the Department of Health and Human Services and the USF Institutional Review Board may inspect th e records from this research project. The results of this study may be published. However, the data obtained from you will be published using a code number or will be co mbined with data from other people in the publication. The published results will not include your name or any other information that would in any way personally identify you. As stat ed above, the sentences you produce will be presented to listeners in two additional studies. Your name will not be given to the listeners. No other personal information will be given to the listeners. Individual data will be stored and identified usi ng a number code. All data will be stored in the Speech Perception and Production Laboratory (PCD 3008A) or in the principal investigator’s office (PCD 4013). Both locations will be locked and secured. Authorized research investigators, agents of the Department of H ealth and Human Services and the USF Institutional Review Board may inspect your records from this research project. Volunteering to Be Part of this Research Study Your decision to participate in this research st udy is completely voluntary. You are free to participate in this research study or to withdraw at any time. If you choose not to participate, or if you withdraw, there will be no penalty or loss of benefits that you are entitled to receive. If participating for extra credit points, you will be credited for the time you participated and will

PAGE 84

74receive no grade penalty for withdrawing. If pa rticipating for money, you will be compensated for the time (hours or part hours) that you participat ed at the rate stated above ($10.00 per hour). Questions and Contacts If you have any questions about this research st udy, contact Catherine Rogers, Ph.D. at (813) 974-7423. If you have questions about your rights as a pers on who is taking part in a research study, you may contact a member of the Division of Research Compliance of the University of South Florida at 813-974-5638. Your Consent—By signing this form I agree that: I have fully read or have had read and explaine d to me this informed consent form describing a research project. I have had the opportunity to question one of the persons in charge of this research and have received satisfactory answers. I understand that I am being asked to participate in research. I understand the risks and benefits, and I freely give my consent to participate in th e research project outlined in this form, under the conditions indicated in it. I have been given a signed copy of this info rmed consent form, which is mine to keep. _________________ ________________ ____ __________ Signature of Participant Prin ted Name of Participant Date Investigator Statement I have carefully explained to the subject the nature of the above protocol. I hereby certify that to the best of my knowledge the subject signing th is consent form understands the nature, demands, risks and benefits involved in participating in this study. _____________________ ________ ____________ __________ Signature of Investigator Printed Name of Investigator Date Or Authorized research investigators designated by the Principal Investigator Institutional Approval of Study and Informed Consent This research project/study and informed c onsent form were reviewed and approved by the University of South Florida Institutiona l Review Board for the protection of human subjects. This approval is valid until the date provided below. The board may be contacted at (813) 974-5638. Approval Consent Form Expiration Date: Revision Date:_______________

PAGE 85

75 Appendix B.1: Language Background Questionnaire for Monolinguals Participant Background Questionnaire (Form A) Name: _____________ Age: _____ Address (town & state): ______________ Phone (optional). Home: __________ Office: _________ Email address (opti onal): ____________ ____________ 1. Is English your first (nativ e) language? Circle one: Yes No 1a. If you answered “No” to (1) ab ove, list your first language here. 2. Did you speak any languages other than English while growi ng up (other than classroom instruction)? Circle one: Yes No 2a. If you answered “Yes” to (2) a bove, list those languages here__________ ________________________ ____________ ______________ 3. List any languages you speak other t han English and rate your degree of proficiency on a scale from “1” to “5” for each (1=beginner, can’t have a conversation; 5=like a native speaker): ________________________ ________________ _____________ 4. Have you ever been diagnosed with a speech or hearing disorder or had speech or hearing difficulties? Circle one: Yes No a. If you answered “yes” to (4), above, please explain in the space provided below (or on back if you need more room): ________________________ __________________ ____________ 5. How long have you lived in Florida (or current state)? ______________________ 6. What state were you born in and how long did you live there? __________________ (don’t answer #’s 7 or 8 if you’ve lived all your life in 1 state) 7. What state have you lived the longest in? ______________________

PAGE 86

76 a. How many years did you live there? ___________________ 8. List any other states that you’ve lived in for over a year (if more than 3, list top three): _____________________ _________________________ 9. On a scale from “1” to “7,” rate your experience wit h listening to speakers with a foreign accent (1=little or no experi ence; 7=every day or very frequent): ________________________ __________________ __________

PAGE 87

77 Appendix B.2: Language Background Questionnaire for Bilinguals Participant Background Questionnaire (Form B) Name: _____________ Age: _____ Addr ess (town & state): ______________ Phone (optional). Home: __________ Office: _________ Email address (opti onal): ____________ ____________ 1. How many years have you lived in your current area (town & state)? __________________ __________________ _______________ 2. Have you ever been diagnosed with a speech or hearing disorder or had speech or hearing difficulties? Circle one: Yes No a. If you answered “yes” to (2), above, please explain in the space provided below (or on back if you need more room): ________________________ _____________________ _____________ 3. What language(s) did your par ents speak with you? ________________ a. If you answered with more than one language in (1), above, which language(s) did each parent speak with you? ________________________ _____________________ ___________ 4. Where were you born (give city state, country ) _______________________ a. How many years did you live there? ________ b. List other cities or regions you’ve lived in for more than one year and note number of years you liv ed there for each. __________________ __________________ _______________ c. What city and country are your par ents from? Mother: ___________ ____________ Father : ___________________ 5. How old were you when you beg an learning English? _________ a. Why did you begin learni ng English? _______________________ ________________________ _____________________ _________

PAGE 88

78 6. If you moved to the United States from another country, how much did you speak English before moving here (descr ibe years of study, if you learned English in a classroom & percent of time speaking English)? ________________________ __________________ _____________ 7. If you moved to the United States from another country, how long have you lived here? ___________ years, ______________ months. 8. On a typical day, what percent of your time do you spend speaking English at work? _____ % At home? _________% Other (shopping, etc.)? _____% 9. On a typical day, what percent of your time do you spend speaking a language other than English at work ? ______ % At home? _____ % Other (shopping, etc.)? ____% (if mo re than one, answer below for each language) 10. What percent of your day do you spend with people with people who speak both (or more) languages that you do? ________ % 11. What language are you most comfortable speak ing? ______________ a. How much more comfortable are y ou in speaking that language on a scale of 1 to 5? (1=equal or nearly equal co mfort; 5=much more comfortable) ________________________ _____________________ ___________ 12. What language are you most co mfortable listeni ng in? ______________ a. How much more comfortable are you in listening in that language on a scale of 1 to 5? (1=equal or nearly equal comfort; 5=much more comfortable) ____________ _____________________ ____________ 13. What language are you most co mfortable reading in? ______________ a. How much more comfortable are you reading in that language on a scale of 1 to 5? (1=equal or nearly equal co mfort; 5=much more comfortable) ________________________ _____________________ ____________ 14. What language are you most comfortable writ ing in? ______________ a. How much more comfortable are you writing in that language on a scale of 1 to 5? (1=equal or nearly equal co mfort; 5=much mo re comfortable)

PAGE 89

79 ________________________ __________________ _____________ 15. Do you think your ability in t he language you are less comfortable in is still improving for any of the skills in questions 9-12? Circle one: yes no a. If you answered yes in 13 above, indi cate which abilities you believe are still improving. Circle any that apply: speaki ng listening reading writing 16. What academic degrees have you earned? (list language of education for each) __________________ ________________ 17. For all languages that you speak, rate your leve l of ability on a scale of 1 to 5 (1=not proficient, like a child or beginner; 5=very proficient, like a welleducated native speaker) for each of the following areas: a. Comprehension: ___ ________________________ _______________ b. Fluency: (ease of expre ssion) _________________ _______________ c. Vocabulary: ________ ________________________ _____________ d. Pronunciation: ______ ________________________ _____________ e. Grammar: _______________ __________________ ______________

PAGE 90

80 Appendix C: Race/Ethnicity Form Race and ethnicity questionnaire Subject code: _________ Please put an “x” in the box for th e items that apply to you. Gender: Female Male Ethnicity Hispanic or Latino A person of Mexican, Puerto Rican, Cuban, South or Central American, or other Spanish culture or origin, regardless of the race. T he term “Spanish origin” can be used in addition to “Hispanic or Latino.” Not Hispanic or Latino Race (check all that apply) American Indian or Alaska Native A person having origins in any of the original peoples of N orth, Central, or South America, and who maintains tribal affiliation or community attachment. Asian A person having origins in any of the original peoples of the Far East, Southeast Asia or the Indian subcontinent, including, for example, Cambodi a, China, India, Japan, Korea, Malaysia, Pakistan, the Philippine Islands, Thailand, and Vietnam. Black or African American A person having origins in any of the black racial groups of Africa Terms such as “Haitian” or “Negro” can be used in a ddition to “Black” or African American. Native Hawaiian or Other Pacific Islander A person having origins in any of the original peoples of Hawaii, Guam, Samoa, or other Pacific Islands. White A person having origins in any of the original peoples of Europe, the Middle East, or North Africa.

PAGE 91

81 Check here if you do not wish to provide some or all of the above information.

PAGE 92

82 Appendix D: Consent Form for Listening Experiment Social Sciences/Behavioral Adult Informed Consent University of South FloridaInformation for Peop le Who Take Part in Research Studies For monolingual listeni ng purposes--volunteer The following information is being presented to help you decide whether or not you want to be a part of a minimal risk research study. Ple ase read carefully. If you do not understand anything, ask the Person in Charge of the Study. Title of Study: Speech perception and production by native and nonnative speakers Principal Investigator: Catherine L. Rogers, Ph.D. Study Location(s): PCD 3008A; PCD3008 You are being asked to participate because you are a monolingual native speaker of English, with normal hearing and no history of speech or hearing disorders. General Information about the Research Study The purpose of this research study is to examine performance of native and non-native speakers of English in both listening and speaking tasks. Plan of Study To participate in this experiment you must fi rst pass a basic audiometric hearing screening. During the hearing screening, a series of short tones (beeps) of different frequencies will be played. Your task is to indicate when you hear a sound, by raising your hand. If you do not pass the hearing screening, it does not necessarily mean that you have a hearing loss, however you may wish to pursue further testing at a docto r’s office or clinic. The experiment will take place in four sessions, on four separate days. On each day, you will hear up to 216 practice trials and approximately 600 trials for the main e xperiment task. On the first day, you will also hear 12 example trials. On each trial (example practice, or main experiment), you will hear a word and six words will be displayed on the scr een. You will choose the word that best matches the word you heard. On example tria ls, words will be presented in order (you will know which word is coming next). On main experiment and practice trials, words will be presented in random order (you must decide which word you heard). On both main experiment and practice trials, words may be mixed with background noise (other voices). You will be given a short break after about 5 minutes of listening and a longer break after about 15 minutes of listening. Completion of the main task and additional tasks should take about 2 hours on the first day and about 1 hour on eac h of the three remaining days. Total time for participation, including completion of this fo rm should take between four and six hours. Payment for Participation You will not be paid for your participation in this study. Benefits of Being a Part of this Research Study No practical benefits of this study are likely to apply to you directly. On the other hand, by

PAGE 93

83taking part in this research study, you may in crease our overall knowledge of human listeners’ ability to understand speech under different liste ning conditions. This knowledge may aid in developing improved methods of teaching English to speakers of English as a second language. Risks of Being a Part of this Research Study There are no known risks related to participation in this research study. The speech you will listen to will not be loud enough to be harmful under normal conditions. The level used will be about that of speech at a normal conversational level. Also, you are encouraged to notify the investigator immediately if any of the s ounds are uncomfortable. If the situation is uncomfortable in any way, the experime ntal procedure will be changed. Confidentiality of Your Records Your privacy and research records will be kept confidential to the extent of the law. Authorized research personnel, employees of the Department of Health and Human Services and the USF Institutional Review Board may insp ect the records from this research project. The results of this study may be published. However, the data obtained from you will be published using a code number or will be comb ined with data from other people in the publication. The published results will not include your name or any other information that would in any way personally identify you. Individual data will be stored and identified using a number code. All data will be stored in the Speech Perception and Production Laboratory (PCD 3008A) or in the principal investigator’s office (PCD 4013). Both locations will be locked and secured. Authorized research investigators, agents of the Department of Health and Human Services and the USF Institutional Review Board may inspect your records from this research project. Volunteering to Be Part of this Research Study Your decision to participate in this research study is completely voluntary. You are free to participate in this research study or to withdraw at any time. If you choose not to participate, or if you withdraw, there will be no penalty or loss of benefits that you are entitled to receive. If participating for extra credit points, you will be credited for the time you participated and will receive no grade penalty for withdrawing. If pa rticipating for money, you will be compensated for the time (hours or part hours) that you partic ipated at the rate stated above ($10.00 per hour). Questions and Contacts If you have any questions about this research st udy, contact Catherine Rogers, Ph.D. at (813) 974-7423 If you have questions about your rights as a person who is taking part in a research study, you may contact a member of the Division of Research Compliance of the University of South Florida at 813-974-5638. Your Consent—By signing this form I agree that: I have fully read or have had read and explaine d to me this informed consent form describing a research project. I have had the opportunity to question one of the persons in charge of this research and have received satisfactory answers.

PAGE 94

84 I understand that I am being asked to particip ate in research. I understand the risks and benefits, and I freely give my consent to partic ipate in the research project outlined in this form, under the conditions indicated in it. I have been given a signed copy of this info rmed consent form, which is mine to keep. Signature of Participant Prin ted Name of Participant Date Investigator Statement I have carefully explained to the subject the natu re of the above protocol. I hereby certify that to the best of my knowledge the subject signi ng this consent form understands the nature, demands, risks and benefits involved in participating in this study. Signature of Investigator Or Authorized research investigators designated by the Principal Investigator Printed Name of Investigator Date Institutional Approval of Study and Informed Consent This research project/study a nd informed consent form were reviewed and approved by the University of South Florida Institutiona l Review Board for the protection of human subjects. This approval is valid until the date provided below. The board may be contacted at (813) 974-5638. Approval Consent Form Expiration Date: Revision Date:_______________

PAGE 95

85 Appendix E: Receipt for Producti on and Perception Experiments By signing this receipt, I am acknowledging that I have received the sum of $_________ for my participation in a University of S outh Florida research project, funded by the National Institutes of Health, National In stitute for Deafness and Other Communication Disorders (NIH-NIDCD) gran t #1R03DC005561-01A1, on the date noted below. _____________________________ Signature of Recipient _____________________________ Signature of Experimenter ______________________________ Date

PAGE 96

86 Appendix F: Instructions Handout for Listening Experiment Example Instructions: You will see a screen that looks like this. You will hear one of the following words: bead, bid, bayed, bed, bad, bod. Use the mouse to c lick on the word you hear. You will hear a total of 12 words with one break in between. Do you have any questions? Level Setting Instructions: Use the mouse to click on the word you hear. After the first 6 words, there will be background noise added to the signa l. At first, it may be diff icult to hear the word with noise in the background, but with practice it will get easier. Listen caref ully and click the word you think you hear. If you’re not sure, ma ke your best guess. After every 6 trials, there will be more noise added, making it harder to hear the word. There will be a total of 24 trials. Let me know if the volume is too loud or too soft and I will adjust it for you. Keep in mind: when I increase the volume, the noise will also get louder and when I decrease the volume, the word will sound soft er. Once we finalize a volume level that’s comfortable for you, we have to keep it on that setting for the remainder of the experiment. Do you have any questions?

PAGE 97

87 Practice Instructions: Use the mouse to click on the word you hear You will hear a total of 288 words. These words are broken up into 4 blocks. Each block contains 72 words. Th e first block will be presented in quiet. The second block will contain background noise. Each additional block, will contain more noise, making it harder to hear the word. If you like, take a short break in between each block. Do you have any questions? Main Experiment Instructions (Days 1& 2): Use the mouse to click on the word you hear. You will hear a total of 588 words presented in background noise. The noise leve l will remain the same for all 588 words. These words are broken up into 6 blocks. E ach block contains 98 words. Take a break after each block for a few minutes. Take a longer break (~10 minutes) when you’re halfway through the experiment (this will be after the third block). Do you have any questions? Main Experiment Instructions (Days 3 & 4): Use the mouse to click on the word you hear. You will hear a total of 588 words presented in quiet. These words are broken up into 6 blocks. Each block contains 98 words. Take a short break after each bloc k. Take a longer break (~10 minutes) when you’re halfway through the experiment (this wi ll be after the third block). Do you have any questions?