|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
The Influence Of Dialect On The Perception Of Final Consonant Voicing by Stacy Nicole Kile 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: Ruth H. Bahr, Ph.D. Nathan Maxfield, Ph.D Jean Krause, Ph.D Date of approval: April 4, 2007 Keywords: Categorical perception, African American English, phonological processing, literacy development, final consonant devoicing Copyright, 2007, Stacy Nicole Kile
ii Table of Contents List of Tables . .iv List of Figures ... ..v Abstract ...vi Chapter 1: Literature Review and Purpose ..1 Dialect ..4 History and Influence of African American English ...5 Phonological Representations ..6 Theories of Storage... .10 Storage hypotheses .11 Effects of Altered Representations 12 Evidence that Phonological Representations may be Different in Speakers of AAE. ..17 Evidence from dialect studies 19 Evidence from literacy tasks ..21 Evidence of a Phonological Processing Deficit .24 Conclusion .27 Summary of the Problem ...28 Dialect Does Play a Role . ..28 Purpose ...31 Chapter 2: Method .32 Participants 32 Materials 35 Hearing and speech/language screeners .35 Language variation measure ..35 Phonological awar eness screener ..35. Stimuli 36 Stimuli selection .36 Stimuli recording ...37 Stimuli manipulation ..37 Testing stimuli selection 38 Experimental Tasks 39 Paired comparison task ..39 Continuum task ..40 Procedures ..40 Data Reduction ...41 Statistical Analysis .43
iii Same/different task 43 Continuum task ..43 Chapter 3: Results ..44 Same/Different Task Results .44 Continuum Task Results. ...45 Response patterns attr ibutable to dialect 45 Response patterns attributable to grade .46 Response patterns attr ibutable to speaker ..47 CTOPP Correlations ..48 Conclusions 49 Chapter 4: Discussion 50 Same/Different Task .. 50 Continuum Task ... ..51 Influence of dialect on perception ..51 Influence of grade on perception ...54 Influence of speaker a nd consonant on perception 56 Strengths of the Current Study ...57 Weaknesses of the Current Study ..58 Stimulus Generation and Manipulation .. ...59 Voicing parameters 59 Talker Variability ...60 Future Studies 61 Conclusion .62 References ..64 Appendices .70 Appendix A: Vowel dura tion and closure duration measurements ...71
iv List of Tables 1. Presenting issues when examining the role of dialect in perception 30 2. Summary statistics of participants 34
v List of Figures 1. Distribution of responses to the continuum task by dialect ..46 2. Distribution of responses to th e continuum task across grades 47 3. Distributions of mean speaker ratings by target consonant ..48
vi The Influence Of Dialect On The Perception Of Final Consonant Voicing Stacy Nicole Kile ABSTRACT Children at risk for reading problems al so have difficulty perceiving critical differences in speech sounds (Breier et al., 2004; Edwards, Fox, & Rogers, 2003; deGelder & Vroomen, 1998). These children rely more heavily on context than the acoustic qualities of sound to f acilitate word reading. Dialect us e, such as African American English (AAE) may influence literacy developm ent in similar ways. Dialect use has been shown to affect speech sound processing and can even result in spelli ng errors (Kohler, et al., in press). The purpose of this study is to determine if children who speak AAE process cues indicative of final consonant voi cing differently than children who speak a more mainstream dialect of English. Twenty-six typically developing children in grades K-2 who spoke either AAE or a more mainstream American English dialec t participated. The speech stimuli consisted of nonsense productions of vowel + plosive cons onant. These stimuli were systematically altered by changing the vowel and stop-ga p closure duration simultaneously, which resulted in the final consonant changing from a voiced consonant, like ib, to a voiceless consonant, like ip. Two tasks were develope d: a continuum task where the child had to indicate when the stimuli changed in voicing and a same-different task which involved determining if two stimuli were identical in voicing or not.
vii No significant differences between groups were found for dialect use or grade for the same/different task. In the continuum ta sk, chi-square analyses revealed significant differences in response patterns attributable to dialect and gr ade. In addition, a significant consonant by speaker interac tion was found for mean ratings Correlations between mean continuum rating and phonological awarene ss composites were not significant. In conclusion, it was evident that children who speak AAE present with differences in their perception of final consona nts in VC nonsense syllables. This finding suggests the dialect speakers may be using di fferent cues to make judgments regarding the speech signal, or that th e speakers of AAE have a less mature ability to extract fine phonetic detail due to the influence of th eir dialect (Baran & Seymour, 1979). More research is warranted to determine the exact role that dialect plays. .
Chapter 1 Literature Review and Purpose There is an ongoing concern for the deve lopment of literacy in the school age population. The profound need for literacy and other related skills is understandable. Despite intense efforts in the schools and through enrichment programs to facilitate literacy learning, there are many children who ha ve difficulties with literacy development and continue to academically fall behind th eir school-age peers. Children all over the United States are not achievi ng age-appropriate academic skills (Farkas & Beron, 2004; Fishback & Baskin, 1991). More specifically to our line of resear ch, children are not meeting age-appropriate reading levels (Re port of the National R eading Panel, NICHD, 2000). Those children are a topic of research because failure to reach expected age reading level has severe impacts on the a dvanced language and literacy skills that determine future success. Reading is a multi-faceted skill. It is evident that there are many processes involved in the development of reading (F ennel & Werker, 2003; Velluntino & Scanlon, 1987). Deficits in any of these areas coul d pose major threats to typical literacy development in children. Fundamental reading skills provide a backbone for the acquisition of advanced skills. For example, as children are devel oping in oral language, phonological skills become more advanced. The combination of these more advanced
2 skills leads to a strong phonological base. In turn a child has access to higher-level literacy and language capabil ities, such as reading. For this study, it is also important to show the relationship between phonology, reading, and perception. Children who have read ing difficulties have been found to have perceptual difficulties also. Research has shown that children with reading deficits have difficulty paying attention to the fine phonetic de tails of speech signals (Breier, Fletcher, Denton, & Gray, 2004; Edwards, Fox, & Rogers 2002). During the processes of reading, these children rely more on context and less on phonology to extract the necessary information. In turn, they miss out on im portant phonetic information from the speech signal. Acquiring the basic level of phonologi cal knowledge (i.e., phonetic distinctions) is essential in developing higher-level skills, su ch as reading. If these children are missing out on the early stages of phonological knowle dge, it possibly puts them at risk for deficits at the higher-level stag es of phonological awareness. If these higher-level literacy and language skills are e ssential, what is happening to the populations that have di fficulty developing these skills ? More specifically, who are the populations that are missing out, and what can be done to intervene? There are several different reasons why children may be missing out on the essential sk ills. Perhaps a lack of phonological knowledge is the core de ficit (Munson, Edwards, & Beckman, 2005). Perhaps it is due to hearing deficit, which is the case with hearing-impaired children or children and chronic middle ear infections (Nittrouer & Burton, 2005). There is also research to suggest that these de ficits result from deficits at the perceptual level, which is the case with dyslexic individuals (Blome rt, Mitterer, & Paffen, 2004). Hence, some children may be missing out on the development of the necessary phonological
3 knowledge due to impoverished early experien ces. Another possible influence on early experiences is the use of a non-standard di alect (Silliman, Bahr, Wilkinson, & Turner, 2002). Perhaps exposure to and use of a di alect influences the development of phonological knowledge. One dialect-speaking populati on that draws attention in the United States is the African American population. Past and pr esent research shows a major gap in achievement levels between this dialect-sp eaking population and their same aged peers (Farkas & Beron, 2004; Fishback & Baskin, 1991) While it is still a mystery why this gap is so persistent, it is clear that the ga p begins to broaden at a very young age and is still apparent in academic scores of olde r students. According to the Report of the National Reading Panel (NICHD, 2000), region al dialectal variati ons are considered moderator variables (p. 2-31), meaning th at dialect somehow contributes to the development of reading, but its exact role is still unknown. It is important to note, also, that speakers of dialect mi ght experience trouble when reading because written forms taught in school are not indi cative of AAE. A more standa rd dialect form is used. Therefore, they may have more difficulty pro cessing the standard form of dialect used in written forms. Some existing explanations for the Black -White achievement gap include poverty, classroom environments and attitudes towards schooling, and early family literacy practices (Craig & Washington, 2006; Evan s, 2005; Snow, Burns, & Griffin, 1998). There has also been evidence to suggest that th is gap is present very early in life. Farkas and Beron (2004) revealed that as early as 36 months, an oral vocabulary gap is already present related to both race and socioeconomic status (SES). It is therefore pertinent to
4 find reasonable solutions that can be taken adva ntage of to lessen the achievement gap so that children are not destined for acade mic failure (Silliman, et al., 2002). As mentioned before, literacy is one of the main areas where these children are falling behind. This poses a significant probl em since reading skills are necessary for achievement in other academic areas. The focu s of this research study is on a dialectspeaking population, and how processes involved in the development of literacy skills may be different in this population. This study considers dialect as one possible early experience that could interfere with the de velopment of strong phonological base suitable for literacy acquisition. Organization of the literat ure review is as follows. The first main section introduces the dialect of African American E nglish (AAE) and its history. A discussion of phonological representations and how they ma y differ in speakers of dialect is next. Theories of phonological awaren ess are addressed, as well as their role in reading development. The link between dialect, phonolog ical awareness and re ading will then be explored. Finally the literature review closes with a summary of the problem. Dialect Based on the evidence that early experi ence has a significant impact on language development (Nittrouer & Burt on, 2005), it is probable that there is a population of children whose acquisition of phonological awareness is impact ed due the inclusion of a specific dialect during their early experien ces. While acquisition of a dialect is not considered to be atypical development, ch ildren who speak a non-standard dialect may achieve aspects of phonological processing di fferently than speakers of Standard
5 American English (SAE) due to the characteri stics of the dialect they speak (Seymour & Seymour, 1981). It is important to note that everyone speaks a variation of SAE. The dialect that will be e xplored in this project is African American English (AAE). This dialect is of interest because of a continuous widening spread in the BlackWhite Achievement Gap. Research has indica ted that the number of African American students that acquire basic levels in reading, science, and math is considerably lower than the number of Caucasian students who acquire the same skills (Farkas & Beron, 2004). This gap presents major concerns for th e education of thes e students and those responsible for providing their education. A br ief review of the history and influence of AAE will help in determining the relationship between AAE and reading skills. History and Influence of Af rican American English Dialects are normal outcomes of language ; however, their underlying role in language development is still somewhat unde termined. AAE specifically has been spoken for decades; however, it was not until the 1960s that research intere st in this dialect increased (Green, 2002). It was at that time researchers became interested in finding patterns of the dialect in hopes to define it more accurately. There are still questions and speculations raised as to the origin of the dialect; however, the features and characteristics of the dialect are clearly defined and agreed on by several authors (Craig & Washington, 2006; Green, 2002; Pollock et al., 1998). Green (2002) reported that while AAE is continually changing, there are as pects of the dialect that are constant and have been present for a significant amount of time. Features that are present in AAE can be characterized as lexical, syntactic, and phonological (Craig & Washington, 2006; Gr een, 2002; Pollock et al., 1998). An
6 example of a lexical feature is using the lexica l entry -own- to represent self and to use it as a qualifier, (e .g., He cooked his food hisownself .). Slang terms are also considered lexical features (e.g., phat : an adjective meaning nice or good). An example of a syntactical feature is th e use of the habitual be (e.g., He be eating). An example of a phonological feature is replacing interdental fr icatives with labiode ntal fricatives (e.g., bath becomes baf). A nother phonological example is the devoicing of final consonants (e.g., bad becomes bat). It is nece ssary to identify the di fferent features in order to determine how they might be aff ecting the phonological acquisition of children who speak the dialect. Poplack (2000) suggests that AAE is one of the most widely spoken variations of SAE discussed in the sociolinguistic research, which is why it is so important to consider the effects of this dialect on read ing and literacy skills. It is just as important to consider how the use of this dialect may affect the development of a childs phonological knowledge (Silliman et al., 2002). Phonological knowledge is an entity that can be broken down into smaller components and arra nged in a hierarchy (Munson et al., 2005). The most basic level of phonologi cal awareness exists at the perceptual level and it must be developed in order to acquire knowledge of higher-level phonol ogical skills, like phonological awareness. How then do phonetic di fferences influence the development of phonological knowledge in dialect speakers? Phonological Representations The storing of language featur es into the lexicon is an intricate process. Research over the decades has concluded that our brai ns store phonological re presentations as a quick way to retrieve concepts when needed (Tunmer & Chapman, 1998). These
7 representations are needed for both accurate perception and production. In order to retrieve lexical information and use it for othe r purposes, it is essential to form and store a complete representation of information. Munson, Edwards, and Beckman (2005) describe four different levels of knowledge th at are necessary to ach ieve in order for the brain to receive a complete representation of information for storage. After the different types of knowledge are explored below, di fferent hypotheses will be addressed that account for the storage of this knowledge into phonological representations. Munson, Edwards, and Beckman (2005) suggest the following four different types of phonological knowledge: perceptual knowledge (understanding of acoustic and perceptual aspects of sounds) articulatory knowledge (understanding of the placement, voicing, and manner of articulation of sounds) higher-level phonol ogical knowledge (understanding of how words are divided into sounds and how sounds are put together to make words) and social indexical knowledge (understanding of how variations in production convey social identity) The first type, percep tual knowledge, is most important to our study. This type of knowle dge includes the developmental changes that are present in childrens perceptions of sp eech (i.e. knowing the difference between /s/ and / / with auditory cues only). Munson et al. (2005) stated that th is type of knowledge entails two different kinds of information. They are: a) information about the finegrained acoustic-perceptu al characteristics of words, and b) information about the categorical structure of sounds, to acc ount for the blindness to within-category variability, (p. 192). It would appear that the in tegrity of perceptual know ledge could influence other levels of phonological knowledge (Coady, Kluender, & Evans, 2005). While language
8 development differs between young children an d that of an adult (Munson & Babel, 2005, Nittrouer & Burton, 2005), the initial stra tegies that children use eventually develop into more sophisticated adult-like strategies. Resear ch also has suggested that children have perceptual immaturity until pos sibly the age of ten (Edwards et al., 2002; Hazan & Barret, 2000; Nittrouer, 1992). If early experiences, such as hearing loss or socioeconomic status (SES), prevent children from full exposure to language, perceptual deficits may occur very early in the developmental process (Nittrouer, 2004). To illustrate the importance of the four different types of phonological knowledge mentioned above, the following example was provided (Munson et al., 2005). Consider the words cake and cage. These words are both stored as lexical representations. Different aspects of the words are stored as di fferent types of repres entations that can be organized into the following categories: articulatory representations, semantic representations, and acoustic/perceptual repres entations. All of thes e categories compose the phonological representation that must be retrieved in order to identify the target word. At the articulatory level, information regard ing voice, manner, and placement is stored. At the acoustic/perceptual level, informati on regarding acoustical parameters, such as frequency and amplitude, is gathered. Once these different types of information are combined to form the lexical representation, th e semantic representation is achieved if the child has been exposed to the word previ ously. All of these ty pes of knowledge are needed to distinguish the two words as ha ving separate semantic representations. Showing the importance of phonological repr esentations is a study by Gaskell and Marslen-Wilson (1998). They demonstrated that different strategies or codes are used to make judgments during the perception of speech. In a two-part study assessing the
9 perception of words and non-words that ha ve had various phonological changes, the researchers discovered that li steners perceived speech at an abstract level and that perception required phonologica l inference and lexical know ledge. Their study used the carrier phrase freight bearer in the following sentence, Luckily the ship was only a freight bearer . Each time the sentence was presen ted, different aspects of the carrier phrase changed each time. Some examples include frayp bearer, frayp carrier, prayp bearer, prayp carrier, freight be arer, freight carrier , preight bearer, and preight carrier The participants were asked to click on the computer screen when they heard the carrier phrase freight bearer . Response times were recorded. They determined that listeners used phonological inferences when making judgments about speech signals. This is evident because the listeners were able to make judgments based on the surface form (i.e., freight bearer ) to evaluate the meaning of the non-words (i.e., frayp bearer ). The participants used high-level phonological ski lls to make judgments regarding these stimuli. Si nce their knowledge of phonology was more advanced, they were able to evaluate the meaning of nonwords based on a correct phonological code or representation stored in their brain for the surface form. Th ey suggested that phonological inference plays a large part in perceptual processing, howev er, this requires a developed phonological representation system. While this storing process is rather invol ved, it is usually a natural developmental process if there are no interruptions or pr ocessing of ambiguous information. There are different theories that explai n the process of storing these different types of information and to determine if these representation are deficient. First, a model representing two broad ways in which our brain stores information will be addressed (Storkel &
10 Morrisette, 2002). Then, three rather detailed theories will be desc ribed that serve as explanations to help speculate deficits th at may occur prohibiting a proper storage of information into phonological representations. Theories of Storage Spoken word processing, or the produc tion and perception of language, depends greatly on the organization of lexical and phonological items in our brain. Fortunately, our brain usually systematically stores the needed information for easy retrieval. The storage compartments that hold the repres entations are referred to as neighborhoods. Lexical information is stored in different neighborhoods based on lexical or phonological similarities. Storkel & Morrisette (2002) used the example of the word sit. Words, such as sip, hit, it, fit, etc., are stored in the same general neighborhood as the word sit because they are phonologically simi lar. Words that are more frequent have a denser neighborhood because of the number of other words th at are phonologically similar. Less common words, such as the se have a sparse neighborhood because there are not many words that are phonologically similar to these. Activating these neighborhoods, which contai n the representations, is part of the process by which we perceive and produce speech. Without a pre-conceived knowledge of sounds and words that are developed into meaningful chunks of information, or representations, our brains ha ve no way of processing what they hear, and no way of having the means to produce speech. Storing info rmation in our brains is generally an automatic developmental function. There are, however, people who have more difficulty establishing well-defined lexi cal boundaries of phonological representations based on
11 experiences. Hypotheses to speculate what ha ppens when people have weakly developed phonological libraries will be discussed next. Storage hypotheses. Three theories that account for deficits during the storing of phonological representations in our brains include the segmentation hypothesis (Brady, 1997; Fowler, 1991), the lexical restructuring deficit hypothesis (Metsala & Brown, 1998; Metsala & Walley, 1998), and the distinctness hypothesis (Elbro, 1996; Elbro, Borstrom, & Petersen, 1998). These three theori es will receive further explanation in the following paragraphs. The segmentation hypothesis attributes difficulties in phonemic access to subtle deficits in formulating, retrieving, and ma intaining phonological repr esentations and not only to retrieval problems. Fowler (1991) sugge sts that as children ge t older, they shift from comprehending lexical units as a unit to comprehending them as smaller, more individual segments. This shift is seen in the spoken example of come mere for come here (Silliman et al., 2002). The child may not realize this is two different words until they become older and are exposed to more lexical complexities. The lexical restructuring deficit hypothesis focuses on the role of vocabulary size and sound familiarity. Metsala and Walley (199 8) suggest that vocabulary development depends mostly on the neighborhood densities that are stored in the brain. Since neighborhoods are stored according to phonetic similarity, it is necessary that phonemic distinctions continually be made in order to gain a strong vocabulary. Three important results that may occur if these phonemic di stinctions are not continually made: a) phonemic access deficits (retrieval), b) gr apheme-phoneme relationship deficits
12 (phonological awareness), and c) deficits in unfamiliar word recognition (decoding). All three of these possibilitie s are associated with poor phonological sensitivity. The distinctness hypothesis attributes deficits in phonological representations to poor discrimination abilities. These distinc tions refer to how phonemically different representations are to their neighbors (Elbr o, 1996). Based on stress, coarticulation, and dialect features, a simple phrase has many vari ations that could re sult. Silliman et al., (2002) used the example of the phrase thats mine Variations of this phrase that could result are: /ds maIn, s maIn, dt maIn, t maIn, d maIn, maIn/. Elbro (1996) states that children with deficits in dis tinction have overlapping phoneme boundaries that make it difficult to specify the desired va riation. These underdeveloped boundaries make it difficult for the desired linguistic form to be achieved based on context. This theory is different from the others in that it su ggests, in regards to neighboring phonemic representations, that word reading is impacted more by underdeveloped boundaries as opposed to difficulties with segmentations. Effects of Altered Representations These three theories suggest that it is possible for individuals to have underspecified phonological representations. Theref ore it is necessary to determine how these representations become alte red. Coady, Kluender, and Evans (2005) postulated that because speech perception and representationa l facility are hard to differentiate in research, problems in one area erroneously im ply problems in the other area. Therefore, an immature ability to code phonological representations can be related to an immature ability to detect fine phonetic detail in speech information.
13 The following studies are perceptual studies that show how the phonological representation deficits of certa in populations are related to early experiences. Several of these types of studies have shown that defici ts in early language experiences can lead to delays in certain phonological processing abilities (Nittroue r, 1996; Nittrouer & Burton, 2005). Speech perception in children as it re lates to phonological awareness has an interesting connection. It is know n that children have a less de veloped ability to perceive characteristics of speech sounds as compar ed to adults (Mayo, Scobbie, Hewlett, & Waters, 2003), and that childre n rely on different paramete rs of the speech signal to detect phonological changes (Hicks & Ohde, 2005). Their knowledge of fine phonetic detail is not as developed as adults because of the differences in exposure. In fact, to account for these differences, Nittrouer ( 1996) proposed the Developmental Weighting Shift hypothesis, which states children perceive different aspects of the acoustic signal as they become more aware of the native la nguage. In another article, Nittrouer and Crowther (1998) stated that ch ildren pay particular attent ion to rapidly changing, more obvious phonetic differences. For example, ch ildren in their experiment used formant transitions to make judgments about individual words, whereas the adults used strategies that included more fine phonetic details. This sh ift represents the stages of developmental maturity that occur when individuals learn to detect advanced acoustic signals as the phonetic knowledge of the na tive language increases. It is possible that certain populations may be more at ri sk than others for a delay in acquiring phonological se nsitivity due to a possible defic it in perceptual abilities. These populations include the heari ng impaired (Leybaert, 1998; Mody, Schwartz,
14 Gravel, & Ruben 1999; Nittrouer & Burton, 2005), chil dren at risk for reading impairment (Blomert et al., 2004), children with specific language impairment (SLI; Burlingame, Sussman, Gillam, & Hay, 2005; Coa dy et al., 2005), children with dyslexia (Godfrey, Syrdal-Lasky, Milla y, & Knox, 1981), and possibly dial ect speakers due to the phonological features of the dialect (Green, 2002). While not much research has been done on the latter condition, the other c onditions will be discussed below. Nittrouer and Burton (2005) studied the effects of early experiences on the perception of speech. They described the proce ss that occurs in or der to store lexical information. During this process, children rely on perceptual weighti ng strategies in order to gain access to lexical information. These we ighting strategies are clues from the signal that help the child differentiate phonemes As childrens knowledge of the language matures, they use different weighting st rategies to make j udgments about their perceptions. Nittrouer and Burton were interested in seeing how this natural process was affected when the early experiences were dimi nished. In order to do this, they tested the adverse effects of otitis media with effusi on (OME) and SES on perception. Four and five year old subjects were placed into one of four groups. The groups were the low SES group, the OME group, the both (low SES and OME) group, and the control group. Using a variety of different phonological processing tasks, the researchers demonstrated how early experiences could affect speech perceptio n and other skills, such as verbal working memory and temporal processing. They conc luded that early experiences did indeed affect the development of language capabili ties. Those children with chronic OME and low SES showed less accurate knowledge of we ighting strategies th an children without these two conditions.
15 These findings present a great groundwork for the reasons why children who have impoverished early experiences miss out on certain phonological information that is necessary for the development of more advanced language ab ilities. Nittrouer and Burton (2005), explained that these deficits interfere with the learning of language-specific perceptual strategies for speech. Being delayed in the acquisition of a ppropriate strategies for speech perception are related to delays in gaining access to phonetic structure, and those delays appear to affect (negatively) the abilities of ch ildren to store and retrieve language in working memory (p. 54). Since ch ildren with perceptual deficits have difficulties recognizing the phonetic structure of words, their ability to store information regarding phonology is impacted, which, in turn creates delays in more advanced skills, such as decoding complex syntax. Another study that showed the effects of perception on phonological awareness is Rvachew (2006). During a longitudinal study, the author explored the relationship between vocabulary, articulation, and perception as predictor variables and their effects on the outcome variable, phonological awarenes s. For the purposes of this study, the perceptual relationship is the focus. An important correlation mentioned in the article is that abilities in perc eption reflect the preciseness of ac oustic-phonetic representations that the child stores in his/her brain. The researcher looked at perception of correctly and incorrectly produced words as a pre-kinde rgarten skill in children with speech/sound disorders. Later, she assessed the phonologica l awareness abilities of the same children when they were leaving kindergarten. The fo llowing relationships were determined from the study: 1) speech perception skills were associated with improvements in phonological awareness and 2) speech perception had an impact on speech production, or articulation
16 accuracy. In conclusion, this study showed that perception in 4-year-olds is associated with later development of phonol ogical awareness in kindergarten-aged children. Yet another study showing the relationship between perception and phonological awareness is Edwards, Fox, and Rogers (2002 ). They discovered that children with phonological disorders had trouble discrimi nating consonant-vowel-consonant (CVC) minimal pairs that differed in the final cons onant (i.e. cap vs cat, and tack vs. tap). The purpose of their study was to exam ine childrens ability to recognize familiar words when redundancy in the speech signal wa s reduced, as in a gating task. During the experimental tasks, the children were asked to identify a CVC unit when a portion of the necessary acoustic information was gated, or removed. They discovered that younger children had a more difficult time discrimi nating CVC words than older children who were typically developing. The authors explained that younger children paid less attention to fine phonetic deta ils due to immaturities in their phonological system. Their results indicated that younger children needed more acoustic information to identify the final consonant in similar sounding words and that those with phonological disorders were less successful than their same aged p eers at this task. Anot her important finding was that younger children and children with phonological disorders seemed to rely on the combination of visual and auditory cues more so than the older and typically developing children. Together, these articles pr ovide strong evidence that children from various backgrounds or learning conditi ons experience the phonological ch aracteristics of speech differently. Adding to the expl anations already offered is the effects of exposure. Nittrouer (1996) noted that children with hi stories of low SES backgrounds spoke with
17 their parents less than the children from higher SES backgrounds. This observation suggests a reason why children from low-SES backgrounds may have diminished lexical knowledge. Exposure is very important duri ng early stages of language development (Hart & Risley, 1995; Honig, 1982; Laosa, 1982; Schachter, 1979; Walker, Greenwood, Hart, & Carta, 1994). If children have a lack of lexical exposure due to conditions such as low SES, differences in acquisition may resu lt. Another factor that may add to the acquisition differences in these homes is langua ge variation. It is possible that a more varied form of SAE (i.e., AAE ) was used in these homes. Evidence that Phonological Representations may be Different in Speakers of AAE Dialect studies have shown that there may be differences in the way that phonological knowledge is stored in the brains of those that speak dialects (Baran & Seymour, 1976; Seymour & Ralabate, 1985) When working memory contains phonologically confusing information, the seman tic and syntactic pr ocesses involved in grammatical role assignment become mo re difficult (Gray & McCutchen, 2006, p. 326). With the addition of dialect characteristics th at alter certain portions of words, it is possible to conclude that assi gning semantic and syntactic role s may be difficult. This is especially true for children who are first learning the language. To specifically target AAE, it is necessary to understand how the dialect changes the phonology of certain words th at are being prepared for st orage into the lexicon. For example, consider the phonological process of devoicing final consonants. This is a phonological rule of AAE that produces a chan ge in the SAE rule of voicing (Green, 2002; Pollock et al., 1998; Rickford, 1999). Ri ckford (1999) states that devoicing of word-final voiced stops after a vowel is a distinctive phonological (p ronunciation) feature
18 of AAE. This is characterized by th e realization of [b] as [p], [d] as [t], and [g] as [k]. It is of interest to note how this dialectal rule in fluences the retrieval and perception of certain words. For example, in AAE the word had can sound like hat due to this phonological change. This can present a problem because the new word that is formed is also a frequently occurring word in English. Both the lexical and phonological representations are at risk for ambigu ity because changing the word phonologically produces another word that is used frequent ly in English. Therefore, questions can be raised such as; how is the word had stored in the childs le xicon? Does the potential ambiguity in the words had and hat influence speech processing in the speakers of dialect? A problem that could occur during the re trieval of phonological representations is a slowed processing time while the brain is trying to decipher wh at actually to retrieve to make sense of the context. In fact, one study revealed perceptual difficulties which resulted in a delay of perceptual processi ng. Floccia, Goslin, Gi rard, and Konopczynski (2006) discussed perceptual issues in regard s to foreign and regional accents in a French community. Over the course of several experi ments, the researchers collected data on the perception of five regional accents. The authors suggested that a regional accent can lead to modifications of the phonological repr esentations used for analyzing the incoming speech signal (p. 1278). Accent processing is divided into two phases: an initial period (where comprehension is disrupted) and an adaptation period (where comprehension is recovered fully or partially). Speakers in a French community were as ked to listen to sentences in other regional dialects and make perceptual judgmen ts about certain words in the sentences
19 (i.e., Ann has never seen any sheep .). The italicized word in dicates where a perceptual judgment was to be made. During the first ex periment, the outcomes suggested that in unfamiliar regional dialects, there is a 30ms processing time delay in word recognition during continuous speech for unfamiliar listener s. The second experiment attempted to show if the processing time delay from the firs t experiment was evident in isolated words also. No significant difference was noted. The purpose of experiment three was to determine if the length of the utterance a ffected the comprehension of the accent. They determined that as utterance length increased, comprehension difficulties increased also. Therefore, the processing delay while the li stener adapted to the accent could be related to inefficient retrieval of phonological inform ation because of dialect unfamiliarity. As the listener was further exposed to the accen t, eventually they reached the adaptation period where full comprehension was established. While the article does not dismiss the fact that part of the delay is due to th e nature of the accent (such as prosody, pitch, inflections, etc), they defin itely found evidence of a lexi cal access delay as listeners adapted to unfamiliar accents. Evidence from dialect studies Seymour and Seymour (1981) attributed differences between young children who use SAE and those who speak AAE to different emerging phonologies as opposed to delayed acquisitional patterns (p. 274). They argued that African American child rens articulation differences reflected developmental aspects of a Black adult syst em just as White childrens articulation differences reflect an emerging White adult system. Specifically, they found that both Black and White children produced the same type s of errors on an articulation test. They came to three conclusions: 1) there were a gr eater number of overall errors produced by
20 African American children compared to Wh ite children, 2) error distributions between place and manner features were different acros s dialects, and 3) less inconsistency was found in distinctive features among substituti on phonemes. These findings suggested that the dialects are distinct and emer ge in similar, but unique, ways. Another study showed similar evidence for a unique emergence of phonology across dialects. Seymour and Ralabate (1985) performed a perception/production study in order to evaluate the phonological feature of substituting / / with /f/, which is a common substitution in both AAE and developmental SA E. Their main goal was to evaluate the difference in both perception and production in words that reflected this substitution. Results indicated that both sets of children (speakers of AAE and speakers of SAE) performed similarly on the production and per ception of single words. Both groups were able to hear the / / as the correct sound. During conve rsational speech, however, the AAE speakers used the substitution persistently. These researchers concluded that, Productive mastery of the dialect form may be dependent on mastery of discrimination and recognition skills (p. 147) The authors also mentioned that as children get older, they are be tter able to code switch betw een dialects, which may be the reason they were able to perceive the s ound correctly yet included the substitution in conversational speech. Yet another study determined possibili ties for a unique emergence. This study suggested that a reason for the differen ce in emerging phonologies could be due to different phonemic cues available for speaker s of different dialects. Baran and Seymour (1976) studied the influence of dialectal phonological rules on th e discrimination of minimal word pairs. They suggested that there were certain p honological rules in AAE,
21 such as final consonant devoicing, that c ould result in two word s sounding homophonous. The purpose of their study was to examin e childrens discrimination of homophonous words without contextual clues. During th eir experiment, different listener/talker combinations were tested: Black/self, Black /Black, White/Black, a nd Black/White. Under these conditions, children listene d to words and were asked to point to the picture that represented which word they heard. The choices of pictures represented the homophonous word pairs (i.e., if the acousti c stimuli was the word pig, pictures representing pig and pick would be avai lable as choices). Base d on response patterns, the speakers of dialect perceived the African American talkers differently than the nonAfrican American talkers. Th e Black children perceived the Black talkers better than White children. Black children also perceived White children better than White children perceived Black children. The authors indica ted that there were phonemic cues available for the dialect speakers that were not availa ble for the non-dialect speakers. Although this article did not address the possible cues, it is possible that they were cues indicative of voicing, such as vowel duration and stopgap closure duration. While non-dialect speakers also used these cues, it is possibl e that the dialect speakers used different weighting strategies to make judgments regarding the acoustic information. Evidence from literacy tasks. Research has shown that dialect in fact impacts phonological activities, such as spelling (Treiman & Barry, 2000). Treiman, Goswami, Tincoff, and Leevers (1997) showed the effect of dialect on spelling samples of American and British dialect speakers. In this study, children were asked to spell words that contained a rhotic r. This dialect feature is the most prominent in distinguishing these two dialects from each one another. The sp ellings produced reflected which dialect the
22 child spoke based on inclusion or exclusion of the pronunciation of the rhotic r in their dialect. For example, in the British dial ect, the word hurt was more commonly misspelled as hut since the r is not as salient in British English. The American children more commonly misspelled hurt as hrt since the r is clearly pronounced in American English. Spellings of the control wo rds were consistent between both groups of dialect speakers. Hence, dialectal phonetic features were activated when a standard production of a word was presented. Another recent study showed the effects of dialect on literacy skills. In this case, Kohler, Bahr, Silliman, Bryant, Apel, and W ilkinson (in press) showed the effects of dialect density on nonword spelling scores. Nonw ords were chosen for this project in order to eliminate lexical effects. A total of 80 African American children were divided into two grade groups (1st and 3rd grade) and subsequently tw o dialect groups (low AAE users and high AAE users). Based on nine different characteristics of AAE, nonwords were developed to assess spelling skills (i.e., pen became len). The Kohler, Apel, Bahr, and Silliman Spelling Assessment ( KABS) was used for scoring the nonword spelling. Spellings were scored based on errors that could be attributed to AAE. Results suggested that high users of AAE in 3rd grade presented with more dialectal patterns in their nonword spellings than low users of AAE in 3rd grade. For the 1st graders, errors represented a number of phonologi cal errors, not only errors attributable to AAE. These findings indicate that dialect affects literacy skills incl uding nonword spelling. Another study by Sligh and Conners (2003) evaluated the possibi lity of dialect effects on the performance of a phonological pro cessing task. The inclusion of dialect in ones language could lead to relative strengths and weaknesses in phonological
23 processing depending on the features and knowledge of the dialect. This is true because it is possible that children who speak a dialect may be more in tune to changes that occur from the standard to their di alect. The opposite could also be true. Children could be less aware of the SAE features because of the inclus ion of dialect where they are less likely to use the SAE features regularly. Both initial and final consonant deletion tasks were used by Sligh and Conners (2003) to show these eff ects. Four different types of tasks were administered to 7-11 year-olds: 1. word initial/outside (say prain without the p ), 2. word initial/inside (say prain without the r ), 3. word final/outside (say hisp without the p), and 4. word final/inside (say hisp without the s ). The authors hypothesized that use of AAE dialect would have a greater impact on the word-final clusters because that is where most of the phonological changes o ccur between AAE and SAE. They also hypothesized that outside deletions would be easier than inside deletions for the same reason that there are more phonological ch anges on that position in AAE. Results supported their hypotheses. Outs ide deletions were signifi cantly easier than inside deletions for the AAE speakers. Speakers of SAE also performed better on the word-final deletion tasks than on word-initial deletion tasks, where AAE performed worse on wordfinal deletion tasks than word-initial deletion ta sks. This was possibly due to the fact that the speakers of AAE were analyzing word fina l consonant clusters that are reduced in their own dialect making the analysis more diff icult. However, it should be noted that the AAE speakers performed better overall on these tasks than the speakers of SAE and the authors attributed this abili ty to the AAE speakers knowledge of two dialects. Hence, this study supported the idea that speakers of AAE may have relative strengths and
24 weaknesses in phonological processing due to the phonological characteristics of their dialect. In another study of AAE dialect and phonological awareness, Thomas-Tate, Washington, and Edwards (2004) used two sta ndardized tests to an alyze the performance of AAE speaking children. The assessments used were the Test of Phonological Awareness ( TOPA ; Torgesen & Bryant, 1994), and the Comprehensive Test of Phonological Processing ( CTOPP ; Wagner, Torgesen, & Rashotte, 1999). These two standardized tests measure different aspects of phonologi cal awareness. The TOPA focuses on initial and final sound comparisons while the CTOPP assesses a more general phonological knowledge. The results of the study indicated that th e children who spoke AAE performed more poorly on the TOPA, which measured initial and final sound comparisons. Since the children performed better on the CTOPP, which is a more generalized assessment, it demonstrated th at their general phonological knowledge may be compensating for their weakened knowledge of finer phonetic details. These results are not surprising consideri ng the rules of AAE. Many of the rules in AAE change aspects of the final consonant resulting in the possibility of a weaken ed knowledge of that position of words. In summary, these ch ildrens general knowl edge of phonological awareness may compensate for more spec ifically defined skills, such as phonemic awareness, making it seem like there is no defic it, when in actuality, the weakness is at a more basic hierarchal le vel (i.e. phonetic level). Evidence of a Phonological Processing Deficit Evidence shows the link between read ing and phonological processing very clearly. Literature reveals, through categor ical perception studi es, that phonological
25 awareness is strongly tied to reading (B ertucci, Hook, Haynes, Macaruso, & Bickley, 2003; Blomert et al., 2004; Breier et al ., 2004). Somewhere in the developmental processes, the children men tioned in the following studies have missed out on the essential skills that would en able them to detect fine phone tic changes in speech signals which results in phonological processing deficits. Breier, Fletcher, Denton, and Gray (2004) during a categorical perception task, tested phonological awareness in children at risk for readi ng disability. Their results indicated that there was a relationship betw een reading disability and the categorical perception of phonemes. They found that children at risk for reading disability were less sensitive to phonological changes occurring in the presented speech stimuli (as determined by response to voice onset time [VOT] parameters). Those who had better sensitivity to the VOT changes also had better phonological processing of the speech stimuli. The authors suggested th at because of this deficit, children at risk for reading may have a more difficult time interpreting the underlying information in the speech signal. Their results indicated that difficul ties while perceiving speech could in turn contribute to difficulties with reading fluency. Similarly, Blomert, Mitterer, & Paffen ( 2004) found that children with dyslexia exhibited immature phonological processing ab ilities when compared to typically developing children. Those with reading difficu lties had to rely more heavily on phonetic context in coarticulation rather than acousti c cues available from the individual segment because the individual segment cues al one were not enough information to make judgments about the speech signal. Defic its were noted in the following tasks: a phoneme-deletion task, an auditory word-d iscrimination task, and a word-recognition
26 task. During the phonological portion of the experiment, no significant differences were noted between the two study groups. However, in experiment two, a significant context effect was found. Context influenced the res ponses of the children with dyslexia more than the children with normal reading skills. This finding suggests that children with dyslexia may weigh contextual cues more h eavily than phonemic cues in individual segments. While this is not necessarily in itself bad, it possibly prohibits children from developing a strong phonetic base. A str ong phonetic base is necessary for the development of higher-level liter acy skills (Munson et. al., 2005). In concert with the idea of a proce ssing deficit, Bertucci, Hook, Haynes, Macaruso, and Bickley (2003) found that childre n with reading disabilities exhibited processing difficulties that were manifest ed as weak phonological coding. Their experiment evaluated the perception and production of vowels in the following CVC words: /pIt/, /pt/, and /p t/ in children with and without reading difficulties. It was hypothesized that children w ith reading difficulties would have a harder time distinguishing vowels that we re phonologically similar. Di fferences were found between the subject groups both in the production and pe rception of the vowels. The children with reading difficulties showed less well-def ined categories in both perception and production when compared to normal reader s. Based on their perception, they had shallower perceptual slopes, and based on the production of vowel formants, they had more phonemic overlapping. In conclusion, th e children with reading difficulty had different perceptual and production patterns. Similarly, Godfrey, Syrdal-Lasky, Mill ay, and Knox (1981) hypothesized that children with dyslexia may exhibit deficits in perceptual tasks. They wanted to show how
27 important perceptual accuracy was in the proc ess of learning to rea d. They stated that learning to read requires the conversion of letters into phone tic equivalents. This, in turn, requires the availabili ty of some long-term representation of the phonetic units, independent of contextual variations, which mu st have been formed by abstraction in the process of perceiving speech (p. 403). When children are learning to read, if they are unable to convert a strand of letters to a percep tual equivalent that is stored in their brain, their ability to process wh at they read will be diminished. Identification and discrimination tasks were administered in order to show perceptual performance of dyslexic readers vs. normally developing reader s. In all of their perception tests, the dyslexic children differed from the normal children in performance, which provides evidence for an immature or different repres entation storage. It is possible that these children with reading difficulties had trouble discriminating between fine phonetic details in the speech signal. Therefor e, they also probably have yet to establish long-term phonetic equivalents to aid in perception. This same rela tionship between perceptual differences and reading ability is interesting to consider with the dialect population. Little to no research has been done in the area of dialect, which is why research is warranted. Conclusion In conclusion, phonological processing im pacts reading abilities. Likewise, deficits in phonological proces sing have potential to impact reading abilities negatively. Several studies have been conducted that suppo rt the literature show ing that perception, phonological processing, and readin g are all connected (Bertucci et al. 2003; Breier, et al., 2004; Blomert, et al., 2004; de-Gel der & Vroomen, 1998; McBride-Chang, 1996). Evidence has revealed that children with r eading difficulties have significant perceptual
28 difficulties (Breier, et al., 2004). Research al so demonstrates that children with reading difficulties pay less attention to the phonetic characteristics and rely more on context to make judgments regarding speech signals (Blomert, et al., 2004). Summary of the Problem Differences in the storage of phonological representations have the potential to create a problem for speakers of dialect in regards to spoken word processing. This is especially true for children who are first learning the phonological properties of their native language. When they begi n to store information as repr esentations into their brain, what effects do features of their dialec t have on the development of phonological representations? As mentioned earlier, ther e are three different hypotheses that explain the effects of a difference in developmenta l acquisition of phonological representations. They are the: 1) segmentation hypothesis (Brady, 1997; Fowler, 1991), 2) lexical restructuring deficit hypothesis (Metsala & Brown, 1998; Me tsala & Walley, 1998), and 3) the distinctness hypothesis (Elbro, 1996; Elbro, Borstrom, & Petersen, 1998). Not enough research has been done in this area to find the exact role that dialect plays, however; it appears to play a role in the development of phonological representations. Dialect Does Play a Role It is crucial for children to develop fundamental phonological awareness skills so that they are less likely to suffer from acad emic failure in the future. Since dialect influences higher-level activities such as sp elling (Kohler et al., in press; Treiman, & Barry, 2000; Treiman et al., 1997), it has the po tential to influence other, more basic skills, such as perceptio n and discrimination.
29 Nittrouer and Burton (2005) recognized th at impoverishing early experiences can impact the development of phonological skills. They described that for some reason (i.e., OME & SES) certain children did not have acces s to all the information they needed for a strong phonological base to develop. The curre nt study considered dialect as one of the possible early experiences that could imp act phonological skills. By examining studies, such as the ones mentioned in this paper, it is evident that dial ect influences phoneme acquisition, as well as the st orage of phonological informa tion (Baran & Seymour, 1979; Seymour & Seymour, 1981). Evidence exists to demonstrate that there are differences in phonological processing between speakers of AAE and speakers of SAE (Seymour & Ralabate, 1985; Seymour & Seymour, 1981; Sligh & Conners, 2003). Evidence also revealed that dialect influences activities, such as spelling and reading (Report of the National Reading Panel, NICHD, 2000; Kohler et al., in press; Treiman & Barry, 2000; Treiman et al., 1997). Evidence from nonword spel ling tasks (e.g., Kohler et al., in press) provides excellent justifica tion for this study because it taps into phonological representations in ways that other phonological processing tests cannot. As mentioned before, dialect does play a role in the development of reading. Its exact role is still unknown. However, in anothe r line of research, strong links were made between reading skill and categorical percep tion (Godfrey et al., 1981) Therefore, a link may be drawn between dialect, percepti on, and reading abiliti es. Since phonological representations may be different in the di alect population (Sillim an et al., 2002), it is necessary to examine this population at a phoneti c level to rule out lexical and contextual effects. In other words, how will this popul ation respond to speech stimuli when they have to rely more heavily on acoustic/audito ry information to ma ke their judgments?
30 It is important to consider all the po ssible issues that may play a role in determining the effects of dialect on literacy skills. All of the issues that must be considered in determining the role of dialect in perception are illustrated in Table 1. Table 1. Presenting issues when examini ng the role of dialect in perception. Issue Reason to consider Academic skills There is a Black-White achievement gap that continues to widen. Phonological representations Alte red or ambiguous storage of representations is possible with AAE. Early experiences Since experiences like OME and SES influence perception, AAE also could serve as an influential early experience. Phoneme acquisition Dialect studies show differences in phonemic acquisition. Spelling Dialect influences error patterns in spelling. Processing deficits Children w ith reading deficits have perceptual immaturities also. Perceptual level of Phonological Awareness Phonological awareness has different levels with perceptual knowledge as basic. Weighting strategies Speakers of AAE may weight acoustic cues differently than standard dialect speakers. Exposure to language aspects Lack of exposure could in turn lead to deficits or immaturities in language skill.
31 Purpose While evidence thus far has shown adequa te information describing AAE dialect, there is not enough information to demonstrate th e role that dialect pl ays in perception. It is important to discover if children w ho speak AAE are using different weighting strategies to make judgment s about auditory information, especially when there are phoneme overlaps in the speech signal due to phonological characteristics of their dialect. The purpose of this study was to determine wh at effects cues indicative of voicing, such as vowel duration and closure duration, may ha ve on the perception of minimal pairs in speakers of a dialect where final consonant devoicing is a prevalent feature. Three questions were specifically addressed. 1. Does the use of final consonant devoici ng (as in African American English) in production influence the perception of th e voiced-voiceless di stinction in VC nonsense syllables? 2. Does grade level influence the percep tion of the voiced-voiceless distinction in VC nonsense syllables? 3. Does the dialect of the speaker influe nce the perception of the voiced-voiceless distinction in VC nonsense syllables?
32 Chapter 2 Method Participants Twenty-six monolingual children between th e ages of 5-8 years participated in this study. The children were in Kindergarten through grade 2 and were recruited from a local elementary school in west central Florida. The project was approved by the Institutional Review Board (IRB) at the University of Sout h Florida (USF) and the local school district. Classroom teach ers assisted the experiment er by sending home parental consent forms. The parents had at least one we ek to respond. Child assent was obtained at the initiation of the experiment. The children had to meet the following inclusion criteria: a) speak AAE and be African American or speak SAE, b) pass a hearing screening, c) pass a speech and language screening, and d) have parental/guard ian consent to participate in the study. The total number of children tested was 30, but four had to be ex cluded for different reasons. One child was unable to do the task, two stude nts were classified as ESL (English as a Second Language) students, and one was currently enrolled in the schools speech/language program. The remaining children were eight who spoke AAE and 18 who spoke SAE. In the following table, SAE is represented as Mainstreamed American English (MAE) because the DELV uses that te rm. Summary statistics for the participants are listed in Table 2.
33 Table 2. Summary statistics of participants. Participant Age (yrs., mos.) Race Grade Dialect Gender AA1 7,7 African American 2 MAE F AA2 7,11 African American 2 MAE F W3 7,6 White 2 MAE F AA4 8,1 African American 2 AAE M H5 8,8 Hispanic 2 MAE F H6 7,11 Hispanic 2 MAE F M8 7,8 Other 2 MAE M M9 7,8 Other 2 AAE F W11 6,2 White K MAE F W12 7,1 White 1 MAE M AA13 7,3 African American 1 AAE F W14 7,0 White 1 MAE M AA15 5,01 African American K MAE M AA16 6,2 African American K AAE M W17 6,4 White K MAE M H18 5,8 Hispanic K MAE F M21 7,3 Other 1 MAE F W22 5,8 White K MAE F AA23 6,1 African American K AAE F W24 5,4 White K MAE M H25 5,9 Hispanic K MAE F H26 6,4 Hispanic K MAE F H27 5,7 Hispanic K MAE M AA28 6,2 African American 1 AAE M AA29 7,1 African American 1 AAE M AA30 6,0 African American 1 AAE M Four women recorded the experimental stimuli. Three of them worked at the local university as professors or clinical supervisor s. The other talker was a graduate student in the Communication Sciences and Disorders pr ogram at the local university. Two talkers were speakers of SAE and were Caucasia n, and two were speakers who could code switch between SAE and AAE and were Afri can American. Consent was obtained in order to record the stimuli. These talkers were used in order to de termine if there was a relationship between race of talker and perception.
34 Materials Hearing and speech/language screeners Prior to the experiment, hearing and speech/language screeners were administered to the children to rule out any hearing difficulties or speech/language delays. A calibrated audiometer was used to test the childrens hearing in a quiet room. Hearing levels at 20dB were screened at 1000, 2000, and 4000 Hz. The local school protocol was us ed to assess the childrens speech and language development. This protocol measur ed various aspects of language, as well as articulation. Teacher input also was obtaine d regarding speech, language, and hearing abilities to informally confirm the results of the screeners. Language variation measure. The Diagnostic Evaluation of Language Variation ( DELV ; Seymour, Roeper, & deVilliers, 2003) was administered to determine if the children were dialect speakers The DELV has a screening section that measures the childs Language Variation Status The first part requires th e child to repeat five sentences to assess phonology. The second part elic its utterances that contain verb tenses that could be affected by language variati on. The verb tenses that are assessed on the screener are 3rd person singular (have/has), 3rd person singular (-s,-es), 3rd person singular (do/does), and the copula (was/were). The resu lts specify the degree of language variation as a strong variation for Mainstreamed Amer ican English (MAE), some variation from MAE, or strong variation from MAE, which would classify them as speakers of AAE. Phonological awareness screener Portions of the Comprehensive Test of Phonological Processing ( CTOPP ; Wagner, Torgesen, & Rashotte, 1999) were administered in order to gain informati on regarding the child s phonological awareness abilities prior to testi ng. Two versions of the test were used, depending on the ages of the
35 participants. One version is for ages 5 and 6 years, while the other is for ages 7 to 24 years. Both versions measure phonological awareness skills and provide a phonological awareness composite score based on the scores of specific subtests. The three subtests that measure phonological awareness skills are: Elision, Blending Words, and Sound Matching. The Elision subtest asse sses the childs ab ility to say words when asked to say the words without one of the sounds in the wo rd (e.g., Say tan without the t), blending words requires the child to say a word when given only the sounds that comprise the word (e.g., What word do these sounds make h-a-t?), and sound matching which requires the child to select a word out of three words that starts with the given sound (e.g., Which word starts with the sound /n/ like nest? Nut, bed, or cake?). The screener for 7-24 year olds only uses the scores from the Elision and Blending Words subtest to formulate a phonological awareness composite. Standard scores for each subtest were obtained and added together to obtain th e phonological awareness composite standard score. Stimuli Stimuli selection All plosives were paired with vowels that occurred at different points on the vowel quadrilateral and resulted in the most instances of VC nonsense syllables. Therefore, the vowels that were selected were: /I/ as in lift, /a/ as in hot, // as in apple, and / / as in cut. Once all vowels were matched with the six American English plosives, the following nineteen VC uni ts resulted: /Ip, Ib, Id, Ik, Ig, ap, ab, at, ak, ag, p, b, k, g, b, t, d, k, and g/. Any plosive-vowel combinations that made real words were discarded to avoid lexical effects in the processing of the stimuli.
36 Stimuli recording The experimental stimuli were recorded by four different women in a sound proof booth in a speech lab. The talkers were speakers of SAE and speakers who code switch between SAE a nd AAE. Using a portable Optimus 33-3013 microphone, the nineteen different syllable s were recorded on a Sony Vaio laptop computer. The syllables were each written in phonetics and given to the speakers. The talker repeated each syllable in a set of three (eg. /Ip Ip Ip /. All of the final consonants were released. The middle syllable was used for the syllable manipulation from voiced to voiceless consonant. Praat (Boersma & Weenink, 2001) was used to record and edit the VC syllables. Time measurements for the following two para meters in each syllable were computed in milliseconds: vowel duration and stop-gap closure duration. The measurements were obtained by the experimenter from a spectral graph of the signal. Measurements of the parameters were extracted from the middle por tion of the signal to refrain from altering any existing transitions. These parameters were used because they are strong indicators of consonant voicing and they could be easily ma nipulated in the syllab les that were used (Hillenbrand, Ingrisano, Smith, & Flege, 1984; Krause, 1982; Li sker, 1967; Raphael, 1972). Stimuli manipulation It was necessary to change each of the stimuli across the vowel duration and stop-gap closure durati on by adding or subtracting milliseconds from each individual stimulus to turn a voiced phone into its voiceless cognate. Since the vowel duration prior to a voiced consonant is longer than the vowel duration prior to a voiceless consonant, milliseconds were cut from the vowel duration of the voiced syllable to incrementally shorten the length of the vowel duration to achieve perception
37 of a voiceless consonant (Borden, Harris, & Raphael, 2003). Likewise, since the stopgap closure duration is shorte r prior to voiced consonants, milliseconds were added to the stop-gap closure durati on to make it sound like a voi celess consonant. Segments were extracted from the center portion of the signal so that transitions were not included. Full pitch periods were extracted in orde r to achieve natural sounding stimuli. The stimuli were altered from voiced to voiceless because it was easier to produce a better sounding stimuli going in that direction. Since the vowel durations and stop-gap cl osure durations were different across the various stimulus items and speakers, it wa s necessary to normalize the changes that were to be made to each syllable during th e experimental manipulation so that the stimuli were changed in a similar fashion re gardless of VC composition. In order to do this, the changes made to each syllable (from all talkers) were computed as percentages of 25% change, 50% change, 75% change, a nd 100% change. At 100% change, the VC stimulus was the cognate of its original phone (i.e. /b/ became /p/). To get these percentages, the difference between vowel durations for each pair of cognates was divided by four. Likewise, the difference be tween the stop-gap closure durations for each pair of cognates was divided by four. M illiseconds were cut or added to the voiced consonant to achieve values for vowel duration and stop-ga p closure duration representative of a voiceless consonant. The two parameters were changed together for each syllable as opposed to individually beca use it was determined that by only changing one parameter, not a big enough difference in voi cing was achieved in the speech signal. Testing stimuli selection. The final manipulated stimuli yielded 160 individual VC syllables (10 voiced phonemes in VC syllables x 4 speakers x 4 percentage change
38 intervals). Depending on the experiment, these syllables were arranged in combinations of either pairs (same/different task), or categories of 5 (each percentage change represented) VC syllables (continuum task). The stimuli for the same/different task were paired to themselves and across all levels of change. The stimuli for the continuum task were arranged in a 5-point continuum fr om voiced to voiceless or vice versa in sequential order. The stimuli for each task we re randomly selected. Those stimuli that were not selected for testing tasks were used for the training tasks. Therefore, stimuli used in the training tasks were not part of the testing tasks. Experimental Tasks Paired comparison task The acoustically manipulated stimuli were placed into EcoS/Win (AVAAZ Innovations Inc., 2002) expe riment generator, which can generate different perceptual tasks. The first task was a paired comparison task where the participant was asked to click same or di fferent to note if they heard a difference between the two-presented stimuli. During th is task, a happy and sad face picture appeared on the computer screen so th at young children could easily recognize the meaning of their response choices. A happy face corresponded to stimuli that were the same, while a sad face corresponded to stimuli that were different. Prior to the experiment, the children had a training session where they could familiarize themselves with the task. The training session consisted of 20 stimuli. The first couple of stimuli were conducted without the headphones over the computer speakers with the help of the examiner. Reinforcements were provided to the participants during the training session. The participants were encouraged to ask questions if they did not understand. Once the children de monstrated they were able to perform the task, (i.e.,
39 by demonstrating an understanding of the task as judged by the examiner) the examiner presented the test stimuli. During the actual testing, 130-paired stimuli were administered. The children ha d the option to rep eat the stimulus if needed. Continuum task The second type of task was a continuum task. For this task, there were five numbered boxes on the com puter screen, each representing a sound along the continuum from voiced to voiceless or vi ce versa. When the participants heard a change or shift in the so und from the voiced to the voiceless phoneme, they were instructed to press th e box that corresponded to the sound where they heard the shift. On the screen, boxes with the numbers 1, 2, 3, 4, and 5 were displayed. The stimuli were presented in a sequential orde r starting from the either th e voiced or voiceless cognate and moving in order from 25% up to 100% or from 100% down to 25% manipulation, where the last stimuli was th e cognate of the first stimu li presented. The 100% change was a manipulated signal. An announcement (i .e., number 1) was made prior to each stimuli presentation to avoid any response c onfusion. The child also had the choice to repeat the stimulus item one time if desired. The children were trained prior to this experimental task al so. Eight continuum sets were used to train the children. When th e children demonstrated they were able to perform the task, (i.e. demonstrating an understanding of the task as judged by the examiner), the test stimuli were pres ented. Twenty-four continuum sets were administered during testing. Procedures The testing was conducted over two separate sessions. On the first day of testing, the children were brought from their clas sroom into a quiet, individual room on the
40 schools property. Consent was solicited upon arrival. The particip ants completed the hearing screening, speech/language screen ing, language variation screener, and phonological awareness screener For some of the childre n, the language variation screener was administered on the second day of testing due to time constraints. The order of the screening tests was randomized across pa rticipants in order to account for fatigue. The first session lasted approximately 30-45 minutes. Once the children were finished, they received a reward for their work. The second day of testing o ccurred within two weeks of the previous session. The children were again picked up from their classroom by the experimenter. They were brought into the same testing room and instru cted to sit down at the laptop computer. A Sony Vaio laptop was used to administer th e experimental tasks. The children were instructed as to the nature of the task pr ior to its administratio n and given a set of headphones to wear. The headphones were clea ned after each child with anti-bacterial wipes. A trial run using the ECoS/Win experiment generator was administered first in order to train the subjects to the experimental task. Once the researcher felt that the participant was acquainted with the equipment, real test stimuli were presented. Task order was randomized across par ticipants to avoid an orde r effect. A short break was given in between the two experiment types to give the children a small break. The second session lasted approximately 30-45 minutes. Once the children were finished, they received a reward from the experimenter.
41 Data Reduction In order to prepare for sta tistical analysis, the experimental data were extracted from ECoS/Win and placed into Excel spread sheets for further analysis. To illustrate listener ability to identify voiced/voiceless distinctions during the same/different task, listener responses were analyzed using th e d measure of signal detection theory (Macmillan & Creel, 1991). This measure provides a numeric equivalent (d) representing the ability of the child to respond appropriate ly when taking into account listener response bias. The listener response bias is taken into account by considering the listeners hit rate and false alarm rate. The hit rate represents the percentage of time the listener responded correctly to the relationship between the CV pair (same or different). The false alarm rate represents the percenta ge of times when the listener responded that the stimuli were same when they were actually different. The percentages from these measures are converted to a normal distributi on z-score. The d measure is the difference between Z (Hit rate) and Z (False Alarm rate). A d of 1 or greater signifies that the listener was able to perform th e task greater than chance. To illustrate listener ability to identif y phonetic changes on a continuum, listener response patterns were analyzed using fr equency counts. For the continuum task, frequency counts were computed for each chil ds responses. The number of times a child chose each category on the continuum task wa s recorded. The ratings of the speakers were adjusted so that the ratings accurate ly reflected the voiced/voiceless change. In other words, when the continuum went fr om voiceless to voiced, the values were reverseda 2 rating became a 4 and a 1 rati ng became a 5 and vice versa. This way, the ratings consistently reflected the point of change from voiced to voice phoneme. Once
42 these ratings were obtained, they were summ ed across grade and di alect to prepare for statistical analyses. Statistical Analyses Same/different task The d from the signal detection theory analysis was computed. These values were analyzed in two separate Kruskal Wallis Analysis of Variances (ANOVAs), one considering diffe rences in dialect group and the other considered differences in grade. Continuum task. A chi-square analysis was us ed to show differences in performance across the continuum. The freque ncy counts computed were analyzed to show differences in response patterns across dialect groups and grades. A three way Multivariate Analysis of Variance (MANOVA) was run to show differences in mean listener ratings across sp eakers. Post hoc tests were used to show differences across speaker and consonants. Correlation statistics were run to find relationships signifi cant to phonological awareness abilities. The CTOPP scores were analyzed across grade and dialect with the ds and frequency counts to dete rmine possible relationships.
43 Chapter 3 Results This study was designed to determine if us e of AAE, a dialect that features final consonant devoicing, influences the percep tion of final consonants in VC nonsense syllables. Twenty-six typically developing ch ildren in grades K-2 participated. Four women (2 African American and 2 Caucasian) provided the speech stimuli, which consisted of nonsense productions of vowel + plosive consonants. These stimuli were then systematically altere d by changing the vowel and stop-gap closure durations simultaneously, resulting in the final cons onant changing from a voiced consonant, like ib, to a voiceless consonant, like ip. Two tasks were util ized: a same-different task which involved determining if two stimuli were identical in voicing or not and a continuum task where the child had to indi cate when the stimuli changed in voicing. The research questions focused on noted differe nces in perception by dialect use, grade, and speaker race. Same-Different Task Results To analyze the effects of dialect use a nd grade, differences in mean d across conditions were analyzed using two separate Kruskal-Wallis ANOVAs. The d served as the dependent variable and dialect and grade as the independent variab les. No significant differences between groups were found for dialect use, 2(1) = .020, p = .889 or for grade, 2 (2) = 1.223, p = .542. It is possible that there were no significant differences because
44 of the small number of participants used or because of the difficult y of the task. Further analysis revealed that 19/26 pa rticipants were able to perfor m the task with a d of 1.0 or more. With scores above 1.0, it can be assumed that the children were able to perform the task at levels greater than chance. Therefore, the present findings indicated that most of the participants were able to perform the tas k, but that the same/diffe rent task was indeed quite difficult. Statistics were run again with the children taken out who could not perform the task, and the results were still insignificant. This furt her supports the idea that neither dialect use nor grade influenced the perception of voicing in this task. Continuum Task Results Data from this task were analyzed using chi-square analyses to show differences in subject response patterns across dialect status and grade. In order to extract the needed information for this analysis, frequency count s of the individual re sponses were obtained for each child. Responses for each unit in the continuum task were then summed across participants by dialect status and grade. It was expected that re sponses would cluster synergistically around units 3 and 4. Actual responses show ed more variability than expected. Response patterns attr ibutable to dialect. The chi-square analysis for differences in response patterns a ttributable to dialec t was significant, 2 (3) = 24.35, p <.01. As illustrated in Figure 1, the speakers of AAE (n= 8) chose the last point on the continuum most frequently, while SAE speakers (n=18) selected points 3 and 4 more often, with 5 response occurring frequently. These findings would suggest that the use of final consonant devoicing may be a factor influenci ng the patterns of performances here. Since the stimuli were manipulated from voiced to vo iceless and vice versa, the selection of the
45 final point on the continuum would indicate th at the distinction between these two types of phonemes must be maximally differe nt for the children to respond. Figure 1. Distribution of responses to the continuum task by dialect. 0 20 40 60 80 100 120 140 160 180 200 2345 Voiced--------------------------------------------Voiceless Response CategoryFrequency SAE AAE Response patterns attr ibutable to grade. The chi-square analysis for differences in response patterns attributable to grade also was significant, 2 (6) = 12.55, p =.05. This finding would suggest that the children res ponded differently to the continuum stimuli depending on their grade level. As illustrat ed in Figure 2, students in Kindergarten showed a steady increase in response, with th e largest number of responses occurring at the final point in the continuum This pattern suggests that they may have found the task difficult and did not hear the difference in voicing until the maximum difference was apparent. First graders picked the second point on the continuum most frequently and the second graders maintained a relatively consta nt response across the continuum, with the last point (#5) receiving the largest number of responses.
46 Figure 2. Distribution of responses to the continuum task across grades. 0 20 40 60 80 100 120 140 2345 Voiced-------------------------------------------------------------Voiceless Response CategoriesFrequency K 1 2 Response patterns attr ibutable to speaker. To note differences on listener responses to the continuum task that may be attributable to speaker, a three-way Multivariate Analysis of Variance (MANOVA) was run with consonant as the within subject factor and speaker and dialect as the between-subject fact ors. The mean rating served as the dependent variable. This an alysis revealed a significant consonant by speaker interaction, F (6,192) = 6.27, p < 0.001, 2 = .164. This finding would suggest that the perception of consonants varied by speaker. Post hoc testing with the Tukey A procedure revealed that across speakers, 4/18 paired comparisons were significant; As illustrated in Figure 3, when looking across subjects for each consonant, talker AA2 (African American) was different from all talkers on /t/ and /d/ and talker AA1 was different from subject 2 on /k/and /g/. When considering differences that are within speaker and across consonant, 5/12 paired co mparisons were signifi cant. In this case, talker AA1 was different on /t/ and /d/ vs /k/ and /g/. For talkers AA2 & C3 (Caucasian), /t/ and /d/ were different from /p/ and /b/, and /k/ and /g/ (but in different directions). For talker C4, there were no significan t differences across consonants.
47 Figure 3. Distributions of mean speak er ratings by target consonant. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5b/pd/tg/kconsonantmean rating AA1 AA2 C3 C4 CTOPP Correlations The nature of the tasks warranted th e inclusion of a phonological awareness assessment. It was also important to include this assessment to determine if these results agreed with Thomas-Tate et al., (2004) who determined that speakers of AAE generally do well on phonological awareness a ssessments. Deficits occur, rather, in more specific tasks of phonological awareness as opposed to de ficits in tasks that assess a general knowledge. Correlation statistics were run to find any relationship between performance on the phoneme perception tasks, dialect, and phonological awareness sk ills. Correlations with the CTOPP scores and the d from the sa me/different task were analyzed by dialect and found to be non-significant. Correlations with the CTOPP scores and mean ratings from the continuum task were analyzed by dialect and also were non-significant. This means that these children are achieving gene ral phonological awarene ss. It is possible
48 though that these children are achieving phonol ogical awareness differently or that deficits occur at basic levels that are somehow compensated for. Conclusion In attempts to find correlations betw een dialect, grade level, and perception, several important results were obtained. Fi rst, during the same/different task, no significant differences between groups were found for dialect use or grade. During the continuum task, a chi-square analysis for diffe rences in response patterns attributable to dialect was significant. Likewi se, a chi-square analysis for differences in response patterns attributable to gr ade was also significant. Du ring the continuum task, a significant consonant by speaker interaction was also found. Po st hoc analyses revealed that 4/18 paired comparisons across speaker and consonant were significant. Post hoc analyses also revealed that 5/12 paired co mparisons within speaker and across consonant were significant. All correlations with the CTOPP were found to be non-significant, suggesting that phonological awar eness skill was not a important factor in these more phonetic tasks.
49 Chapter 4 Discussion The current study examined the relationshi p between dialect st atus and perception of final consonants in 26 children in grades K-2. Two different tasks were administered in order to determine this relationship. Analys es determined no significant differences in the response patterns of the pa rticipants during the same/different task. Three important findings were extracted from the continuum task. First, analysis showed significant differences in response patterns related to di alect use. Second, analysis also showed differences in response patterns across grad es. Third, significant differences were found in performances across percep tion of speaker by consonant. Explanation of these results follows. Same/Different Task The first task focused on the ability of children to make judgments regarding voicing when presented with two similar s ounding stimuli. Results indicated that the mean d showed no significant differences in pe rformances attributable to dialect use or grade. Closer examination of the d values revealed that 19/26 children achieved scores of 1.0 or greater. These findings would suggest that while most children were able to perform this task at levels slightly greater than chance, there were some participants that had difficulty. The noted difficulty in perfor mance could be due to the nature of the stimuli. Since less is known about the interaction of stopgap closure and vowel duration
50 in producing the voiced-voiceless contrast in the final word position than voice onset time (VOT) in the initial word position, it was difficult to manipulate the stimuli consistently across talkers. Percentages of each parameter were used as an attempt to be consistent, but it is possible that talkers uniquely alter these parameters and that changes in final consonant voicing are more repres entative of a continuous, as opposed to categorical, variable (Raphael, 1972). As a resu lt, the stimuli did not represent an abrupt change in voicing, as in VOT. Therefore, it is possible that the child rens difficulty with the task could be partly attributed to the weaknesses of the stimuli. Continuum Task Several important findings were noted in the continuum task. The differences in response patterns attributable to both dialect use and grade we re significant, as well as differences in mean ratings attributable to sp eaker and consonant. This task required that the participants choose when the speech si gnal changed during the presentation of a continuum of changes in parameters asso ciated with final consonant voicing. The participants had four different categories to choose from when deciding when the signal changed. Perceptual differences based on dialect use will be discussed in detail first. Influence of dialect on perception. As shown in the chi-square analyses, differences in response patterns existed for children who spoke different dialects. These results are consistent with previous resear ch. Baran & Seymour (1976) determined that children who speak AAE may experience homophony while making judgments regarding speech signals. Seymour & Seymour (1981) came to similar conclusions regarding differences in AAE developmental speech patt erns. While there were differences in the number and type of errors, the differences not ed in both dialects we re developmental in
51 nature. It could be that the children in this experiment were experiencing a difference in their processing of speech information. It is possible that these children had different phonological representations that were retrieved during the process of perception. This could be explained by the possibility of different emerging phonologies. Speakers of AAE may have to rely on diffe rent cues in order to make judgments regarding speech because of the characteristics of their dialect. Consider the following example. Suppose a first grade teacher writes the word hand on the board. She then asks the students to identify the /d/ sound. Th e children in the class who are speakers of SAE do not have much difficulty with this task. The children who speak AAE may have to process this word differently than SAE speakers. In their dialect, the word hand becomes han because of final cluster redu ction (Green, 2002). Hence, they may have difficulty perceiving the /d/ sound in a final cl uster if it is absent in their own production. It is possible they use other cues to dete rmine the presence of the sound. This project does not necessarily reveal what the different cues are, but th e current results suggest that children who use AAE might have different phonological representations of the word making perception different. Ambiguity in phonological representations c ould in turn lead to difficulties in lexical development (Brady, 1997; Elbro, 1996; Elbro, Borstrom, & Petersen, 1998; Fowler, 1991; Metsala & Brown, 1998; Mets ala & Walley, 1998). Language aspects, such as reading, rely on the retrieval of stored phonologica l representations. If these phonological representations are not stored e fficiently because of ambiguities, children will have difficulty retrieving these representations when needed in order to make lexical judgments. This creates a critical problem fo r children who are at the pre-school age. As
52 children who speak AAE are learning their langu age initially, it is possible that they are storing phonological representations differen tly. Then once they reach school, they may have difficulty in language tasks, such as r eading and spelling, because of the differences in phonological representations. In this study, there were differences be tween the response patt erns of those who spoke AAE and those who spoke SAE. Participants who spoke AAE chose number 5 on the continuum most frequently where speaker s of SAE chose 3 and 4 more frequently. This finding would suggest that dialect did influence the perception of final consonant voicing. The children who spoke AAE needed the maximum degree of change between stimuli in order to perceive a difference in voicing. Speakers of SAE were able to detect the shift a little earlier. These findings in no way suggest that dialect impedes the perception of voicing, instead th ey indicate that the speakers of AAE may be relying on different cues to detect a change in voic ing and the manipulations conducted in this experiment hindered voicing perception. This interpretation is more consistent with Nittrouer and her colleagues who found that ch ildren used different cues than adults when perceiving consonants (Nittrouer, 1996; Nittrouer & Crow ther, 1998). Further research should be conducted on how speaker s of AAE use voicing cu es in the perception of final consonants. This knowledge coul d be important in understanding the phonological representations of children who speak both AAE and SAE and how this knowledge impacts vocabulary developm ent (Storkel & Morrisette, 2002). These finding are also in concert with S ligh and Conners (2003). Their research showed that depending on the characteristic s of the spoken dialect, speakers may have relative strengths and weakne sses in regards to phonologica l awareness abilities. For
53 example, they may be bette r processors of phonological in formation that is present frequently in their dialect, but the may be weaker at processing phonological information that is not as frequent in thei r dialect. In this case, it is po ssible that use of final consonant devoicing (as in AAE) lead to differences in their ability to perceive voicing in final consonants. Therefore, it is possible that children who sp eak AAE have a less mature ability to hear differences between two speech signals that differ in final voicing or experience ambiguity in accessing their phonolo gical representations because in their dialect, these phonemes are (at times) homophonous (Baran & Seymour, 1976). In order for children to discriminate between sounds, they must have clear phonetic boundaries in their phono logical representations. If therefore, these boundaries are not fully established, homophonous words may impede future vocabulary learning. It could be argued that children rarely ha ve to rely on phonetic aspects for speech perception, however, the developm ent of new vocabulary requir es this skill (Storkel & Morrisette, 2002). Further research is warrant ed to show what the noted perceptual differences actually mean, but it is eviden t that the differences are present. Influence of grade on perception. In the chi-square analysis, response patterns across grade were determined to be significan tly different. Review of the frequency table revealed that the three grade levels perfor med differently from one another during the perception task. Kindergartners appeared to c hoose the last step in the continuum most frequently suggesting that they did not hear fine phonetic differences until the greatest amount of phonetic difference was available. Th is is in concert w ith their schooling and age. Children in kindergarten have a basic understanding of letters and sounds. As they continue to develop their language skills, childr ens perceptual maturity also develops. It
54 must also be noted, as stated in the prev ious section, speakers of AAE choice number 5 on the continuum most frequently. This could suggest that the speakers of AAE have a perceptual maturity more si milar to the younger children. First graders picked the second point on the continuum most frequently. This level of the continuum represents minimal ch anges to the speech signal. These findings suggest the first grade children may be more in tune to the phonetic changes in speech because of the emphasis on word decoding and other aspects of phonological awareness as they learn to read. They are gaining more exposure to different cues present in speech. The exposure to those cues may have made it easier for them to make judgments regarding the speech stimuli. Patterns in second grade responses sh owed that they responded across the continuum rather consistently. Their most common choice was number 5 on the continuum, which like the kindergartners, sugges ts they had difficulty hearing differences in the speech signals until th e maximum amount of change was available. This finding may also be a consequence of their stage of schooling. Once children reach second grade their development of language and reading also advances. Th ese children are moving into more lexical-based learninga greater focus on whole words, as opposed to focusing on phonemes. Hence, these findings do not suggest a perceptual deficit, rather a shift in perceptual knowledge and a relian ce on different cues for perception. This information agrees with the literat ure regarding developmental changes in phonological knowledge. Munson et al. (2005) suggested four different types of knowledge essential for phonologica l processing. The first st age begins at the phonetic level and then gradually shifts to a higher-lev el knowledge as a pers ons understanding of
55 their native language matures. The higher-l evel knowledge allows a person to use multiple advanced cues to make judgments. Th is concept is apparent in the responses across grades. The younger children used th eir basic level of phonological knowledge to make judgments. Therefore, their responses s howed the least mature ability to perceive acoustic differences. As the chil dren progressed to first grade, their responses indicated a more developed perceptual knowledge at th e phonemic level. Then as the children reached second grade, they performed more like the kindergarteners. This may suggest that those children were looking for higher-level cues, such as lexical cues that were not available. It is also possibl e that the older children were lexically processing the carrier phrase (i.e. the number stated before the stimulus), which impeded their phonetic processing of the nonsense syllable unit. Fi nally, the older children could have been distracted by the nonsense stim uli because they were anticipa ting a longer lexical unit. Influence of speaker and consonant on perception. During the continuum task, a significant consonant by speaker interaction was found. In othe r words, the participants responded differently to consonants based on which talker was speaking. This suggests that children are able to adjust to different speaking models effectively. Children receive different talker models as they develop language. Their production models differ when they are at home (i.e. parents, siblings), sc hool (i.e. teachers, educat ors), or in different environments (i.e. church, doctors office). If dialect is heavily included in their home environment, it is possible that the language model they receive in school could be significantly different. While they receive variab le speech signals, they must be able to adapt in order to make judgments regarding pho netic details. This is in concurrence with Munson et al. (2005) and Rvachews (2006) hypotheses that state children receive
56 different phonemic models and are still capable of perceiving speech appropriately. Since there was much variability in the productions of the nonsense sylla bles, the children had to adjust to the differences in models. For this project, when looking across and w ithin subjects for each consonant, the listeners responded significantly different depending on the consonant. After analyzing the syllable information, it was evident that there were no phonetic differences attributable to ethnicity; Post hoc testing revealed that talk er AA2 was different from all talkers on /t/ & /d/, talker AA1 was different from AA2 on /k/ & /g/. Within talkers, talker AA1 was different on /k/ & /g/ vs. /t/ & /d/. For talkers AA2 & C3, /t/ & /d/ were different from /k/ & /g/ and /p/ & /b/. Talker C4 showed no differences across consonants. These differences in response pa tterns across speaker and consonant show the variability in speech productions that all children must learn to appropriately process.Strengths of the Current Study There were several strengths of this p ilot study that must be discussed. When considering what type of stimuli would be most appropriate, it was determined that nonsense stimuli would show the most effectiv e results for a couple of reasons. Use of real word stimuli would allow the listener to rely on lexical cues and not phonetic cues alone. The stimuli for this project accounted for that problem. By creating nonsense syllables, lexical effects were controlled thereby creating a purely phonetic task. Another strength of this study was th e inclusion of two tasks during the procedures. The significance of having two tes ting measures for this project was crucial for different reasons. Multiple tasks allowed the children to demonstrate their knowledge in different manners. In additi on, during statistical analysis, on e of the tasks did not show
57 significant differences. If only the same/d ifferent task was used, no significant differences in performance would have been found. It appeared that the children had difficulty with the same/different task. Ther efore, it was beneficial that the continuum task was also administered. In this instance, it was a more sensitive task. Yet another strength of this current project was the inclusion of the CTOPP phonological awareness composite scores. Due to the nature of th e project tasks, a phonological awareness screener (CTOPP) was administered to identify any children that might have difficulties with overall phonologi cal awareness. Results of the CTOPP screener showed that all of the children performed within + 1 sd of the mean regardless of dialect or grade. The fact that no significan t differences between groups on this task were found concurs with the present literature regarding emerging phonologies in speakers of AAE (Seymour & Seymour, 1981; Thomas-T ate, Washington, & Edwards, 2004). Seymour and Seymour (1981) suggested th at the developing phonology is intact for speakers of AAE. It does appear to be different, not lacki ng. Thomas-Tate et al. (2004) suggested a similar hypothesis. In their st udy, based on the outcomes of the CTOPP, overall phonological processing skill between dialect groups was similar. The results from this study concur with existing lite rature regarding overa ll phonological skills in speakers of AAE. Weaknesses of the Current Study When discussing the results of this pr oject, it is important to note the disadvantages. Firstly, it is important to note the small sample size (n=8) for the AAE speaking group. Ten children were African Am erican and four were mixed ethnicities (African American & Caucasian). In the subject group that volunt eered, only a small
58 group actually tested on the DELV as speakers of dialect. It is possible that more children tested were dialect speakers, but they were able to code switch and did so during the process of being tested. Anot her contributing factor could be the location of the school. The school involved was in a middle cl ass neighborhood, with government funding housing close to the school. Re sults may show greater differe nces in a population with a lower socioeconomic status and heavier infl uence of dialect in their environment. Stimulus Generation and Manipulation Voicing parameters. Difficulties with the stimuli mu st also be noted. During the generation and manipulation of the stimuli, a few problems were encountered. The first issue involved the parameters that were chosen for manipul ation. Since final voicing was the issue here, vowel duration and stop-gap cl osure duration instead of VOT needed to be manipulated. This was more difficult than expected. It was decided to alter these parameters simultaneously because changing onl y one parameter did not create detectable differences to the examiner. This decision may have resulted in unnatural manipulations of the desired parameters, in that it wa s assumed that each parameter could be manipulated equally (as demonstrated by the percentage changes used) and simultaneously. The latter idea makes the performance of the SAE group even more interesting because they were able to adjust to this form of manipulation and the AAE speakers experienced more difficulty. Another decision that was made concerni ng the stimuli was which direction to present and change the stimuli during the task s. Since the stimuli were being put into a categorical perception task, they needed to be changed from voiced to voiceless and vice versa. Bi-directional changes (voiced to voice less & voiceless to voiced) were made at
59 first. It was determined by the examiner that when the changes were made from voiceless to voiced, adding milliseconds to a sound resulted in an unnatural production. When milliseconds of speech were extracted from the signal, more realistic sounding stimuli were produced. Therefore, the stimu li only reflect a change from voiced to voiceless. It was also evident that the two parame ters did not interact systematically meaning that a change in one parameter did not necessarily affect the other parameter in the same manner. Therefore, changes made to the stimuli were not sy stematic. Instead of manipulating each sample individually by milliseconds, the stimuli were manipulated by percentages. In other words changes made to the stimuli differed for each individual stimulus. Therefore the changes made to the stimuli were not linear meaning that a change to one stimulus did not equate to the same change on another stimulus. Talker variability. Another reason that the stimuli were difficult to manipulate was the fact that the talkers showed signifi cant variability in thei r productions of the nonsense syllables. Speakers did not show any production patterns that would allow for linear changes to the stimuli. In fact, their pr oductions were extremel y variable in regards to vowel durations and stop-ga p closure durations. The lengths of the parameters varied both within and between subj ects. Therefore, since the vowel durations and stop-gap closure durations significantly differed in leng th across speakers, the changes had to be made based on percentages. For example, in the nonsense syllable /g/, Talker 3s vowel duration was 455 milliseconds, compared to Talker 4s vowel duration of 183 milliseconds for the same syllable. Their rela tive closure durations were 92 milliseconds verses 52 milliseconds. Since the millisecond durations differed between the parameters
60 substantially, it was difficult to make equiva lent changes by an actual millisecond length. Hence, the stimuli had to be changed indi vidually by percentage s (i.e., 25% change relative to the vowel duration or stop-gap closure duration) Therefore the changes made to the stimuli were not linear. It is also po ssible that the shift in voicing actually occurred in the middle of an item on the continuum maki ng the perception of ch ange occur later in the continuum than it actually did. Future Studies Future plans for this type of study ar e numerous. Since differences in the perception of phonetic details have been found, th ere is a need for research to continue investigating the reason for these differences. It is possible that studies using different characteristics of AAE would be beneficial (i.e., omissi ons, & substitutions). AAE has several rules that change th e phonology of the dialect (i.e. replacing interdental fricatives with labiodental fricatives, consonant cluster movement, postvocalic consonant reduction, monophthongization of diphthongs, et c.; Craig & Washington, 2006; Green, 2002; Pollock, et al., 1998). Investigating a ny of these characteristics may achieve differences in perception as well. More inve stigations must be conducted that focus on the phonetic level of perception where context ca n be excluded to be able to describe the influence of dialect on the developm ent of phonological representations. Another type of study that may be bene ficial is an evoke d response potential (ERP) study. ERP studies could detect di fferences in perception across dialect by investigating how the brain responds to phonetic changes in speech signals. By looking at brain waves, it would be possibl e to see when the differences occur, and what part of the speech signal receives the greatest response differences.
61 Reading studies are also a way that th is type of research could be headed. Research shows that reading deficits are of ten linked with perceptual difficulties, (Bertucci et al. 2003; Breier et al., 2004; Blomert et al., 2004 ). Likewise, development of reading and higher level skills requires the proper storag e and retrieval of phonological representations (Munson & Babel, 2005). Incl uding reading scores could help make a stronger link between percep tion, phonological representation and reading development. Since there is a large portion of children w ho have reading difficulties, including those children in these studies would be helpful. Including reading scores in future studies may help better pinpoint certain deficits. Conclusion In conclusion, it is evident that childr en who speak AAE present with differences in their perception of final consonants in VC nonsense syllables. AAE speakers response patterns suggest they may have perceived voi cing later on a continuum task than the speakers of SAE. This is suggestive that the dialect speakers may have been using different cues to make judgments regarding the speech signal, or that the speakers of AAE have a less mature ability to extract fi ne phonetic detail due to the influence of certain characteristics of their dialect (Baran & Seymour, 1979). It also suggests that they may have different emerging phonologies that may influence the storage of phonological knowledge (Seymour, & Seymour, 1981; Thomas-Tate et al., 2004). These results are important for several re asons. Although the lite rature says that these children perfor m fairly well on phonological proces sing tasks (Thomas-Tate et al., 2004), however other studies show that dialect may be influencing other linguistic tasks, such as spelling (Kohler et al ., in press; Treiman, & Barry 2000; Treiman, et al., 1997).
62 Therefore, dialect does play a role. The exact na ture of the role is still undetermined. The results from this pilot study have important imp lications for future research. It is hard to identify what these differences mean; however, with more research in this area, important conclusions may be drawn.
63 References Avaaz Innovations. (2000). Ecos/Win: Experime nt generator and controller for Windows [Computer software]. London, On tario, Canada: Author. Baran, J., & Seymour, H. N. (1976). The infl uence of three phonologi cal rules of Black English on the discrimination of minimal word pairs. Journal of Speech and Hearing Research, 19 467-474. Bertucci, C., Hook, P., Haynes, C., M acaruso, P., & Bickley, C. (2003). Vowel perception and production in adoles cents with reading disabilities. Annals of Dyslexia, 53 174-200. Blomert, L., Mitterer, H., & Paffen, C. (2004). In search of the a uditory, phonetic, and/or phonological problems in dyslexia: Cont ext effects in speech perception. Journal of Speech, Language, and Hearing Research, 47 1030-1047. Brady, S. A. (1997). Ability to encode phonological representations: An underlying difficulty of poor readers. In B. Blachman (Ed.), Foundations of reading acquisition and dyslexia: Imp lication for earl y intervention (pp. 21-47). Mahwah, NJ: Lawrence Erlbaum Associates. Breier, J. I., Fletcher, J. M ., Denton, C., & Gray, L. C. ( 2004). Categorical perception of speech stimuli in children at risk for reading difficulty. Journal of Experimental Child Psychology, 88 152-170. Boersma, P., & Weenink, D. (2001). Praat a system for doing phonetics by computer. Glot International, 5: 9/10, 341-345. Borden, G. J., Harris, K. S., & Lawrence, J. R. (2003). Speech science primer: Physiology, acoustics, and perception of speech (4th ed.). Baltimore, MD: Lippincott Williams & Wilkins. Burlingame, E., Sussman, H. M., Gillam, R. B ., & Hay, J. F. (2005). An investigation of speech perception in children with specifi c language impairment on a continuum of formant transition duration. Journal of Speech, Language, and Hearing Research 48 805-816. Coady, J. A., Kluender, J. R., & Evans, J. L. (2005). Categorical perception of speech by children with specific language impairment. Journal of Speech, Language, and Hearing Research, 48 944-959.
64 Craig, H. K., & Washington, J. A. (2006). Malik goes to school: Examining the language skills of African American students from preschool-5th grade Mahwah, NJ: Lawrence Erlbaum Associates. deGelder, B., & Vroomen, J. (1998). Impair ed speech perception in poor readers: Evidence from hearing and speech reading. Brain and Language, 64 269-281. Edwards, J., Fox, R. A., & Rogers, C. L. (2002). Final consonant discrimination in children: Effects of phonological disorder vocabulary size, and articulatory accuracy. Journal of Speech, Language Hearing Research, 45 231-242. Edwards, T. J. (1981). Multiple features an alysis of intervocalic English plosives. The Journal of the Acoustical Society of America, 69 535-547. Elbro, C. (1996). Early linguist ic abilities and reading development: A review and hypothesis about distinctiveness of phonological re presentations. Reading and Writing: An Interdis ciplinary Journal, 8 453-485. Elbro, C., Borstrom, I., & Petersen, D. K. (1998). Predicting dyslexia from kindergarten: The importance of distinctiveness of phonologi cal representations of lexical items. Reading Research Quarterly, 33 33-60. Evans, R. (2005). Reframing the achievement gap. Phi Delta Kappa, 86 582-589. Farkas, G., & Beron, K. (2004). The detail ed age trajectory of oral vocabulary knowledge: Differences by class and race. Social Science Research, 33 464-497. Fennell, C. T., & Werker, J. F. (2003). Ea rly word learners ability to access phonetic detail in well-known words. Language and Speech 46 245-264. Fishback, P. V., & Baskin, J. H. (1991). Na rrowing the Black-White achievement gap in child literacy in 1910: The roles of school inputs and family inputs. Review of Economics and Statistics, 73 725-728. Floccia, C., Girard, F., Goslin, J., & Konop czynski, G. (2006). Does a regional dialect perturb speech processing? Journal of Experimental Ps ychology: Human Perception and Performance, 32 1276-1293. Fowler, A. E. (1991). How early phonologica l development might set the stage for phoneme awareness. In S. A. Bra dy & D. P. Shankweiler (Eds.), Phonological processes in literacy: A tri bute to Isabelle Y. Liberman (pp. 97-117). Hillsdale, NJ: Lawrence Erlbaum Associates. Gaskell, M. G., & Marslen-Wilson, W. D. ( 1998). Mechanisms of phonological inference in speech perception. Journal of Experimental Psychology, 24 380-396.
65 Godfrey, J. J., Syrdal-Lasky, A. K., Millay, K. K., & Knox, C. M. (1981). Performance of dyslexic children on speech perception tests. Journal of Experimental Child Psychology 32 401-424. Gray, A., & McCutchen, D. (2006). Young r eaders use of phonological information: Phonological awaren ess, memory, and comprehension Journal of Learning Disabilities, 39 325-333. Green, L. J. (2002). African American English: A linguistic introduction Cambridge: Cambridge University Press. Hart, B., & Risley, T. (1995). Meaningful differences in th e everyday experiences of young American children. Baltimore: Paul H. Brookes. Hazan, V., & Barrett, S. (2000). The developm ent of phonemic categorization in children aged 6-12. Journal of Phonetics, 28 377-396. Hicks, C. B., & Ohde, R. N. (2005). Deve lopmental role of static, dynamic, and contextual cues in speech perception. Journal of Speech, Language, and Hearing Research 48 960-974. Hillenbrand, J., Ingrisano, D. R., Smith, B. L ., & Flege, J. E. (1984). Perception of the voiced-voiceless contrast in syllable final stops. Journal of the Acoustical Society of America, 76 18-26. Honig, A. S. (1982). Language environments for young children. Young Children, 38 5667. Krause, S. E. (1982). Vowel duration as a perceptual cue to postvocalic consonant voicing in young children and adults. Journal of the Acoustical Society of America 71 990-995. Kohler, C. T., Bahr, R. H., Silliman, E. R., Bryant, J. B., Apel, K., & Wilkinson, L. C., (In press). AAE dialect, and performa nce on nonword spelling and phonemic awareness tasks. American Journal of Speech-Language Pathology Laosa, L. M. (1982). Families as facilitators of childrens intellectual development at 3 years of age. In L. M. Laosa & I. E. Sigel (Eds.), Families as learning environments for children (pp. 1-45). New York: Plenum Press. Leybaert, J. (1998). Phonologica l representations in deaf children, the importance of early linguistic experience. Scandinavian Journal of Psychology 39 ,169-173. Lisker, L. (1967). Closure duration and the in tervocalic voiced-voice less distinction in English Language 33 42-49.
66 Macmillan, N. A., & Creel, C. D. (1991). Detection theory: A users guide Cambridge: Cambridge University Press. Mayo, C., Scobbie, J., Hewlett, N. and Wate rs, D. (2003). "The influence of phonemic awareness development on acoustic cue weigh ting in children's speech perception" Journal of Speech, Language and Hearing Research, 46 1184-1196. McBride-Chang, C. (1996). Models of speech perception and phonological processing in reading. Child Development 67 1836-1856. Metsala, J. L., & Brown, G. D. A. (1998) Normal and dyslexic reading development: The role of formal models. In C. Hulme & R. M. Joshi (Eds.), Reading and spelling: Development and disorders (pp. 235-261). Mahwah, NJ: Lawrence Erlbaum Associates. Metsala, J. L., & Walley, A. C. (1998). S poken vocabulary growth and the segmental restructuring of lexical re presentations: Precursors to phonemic awareness and early reading ability. In J. L. Me tsala & L. C. Ehri (Eds.), Word recognition in beginning literacy (pp. 89-120). Mahwah, NJ: Lawrence Erlbaum Associates. Mody, M., Schwartz, R. G., Gravel, J. S., & Ruben, R. J. (1999). Speech perception and verbal memory in children with a nd without historie s of otitis media. Journal of Speech, Language, and Hearing Research, 42, 1069-1079. Munson, B., & Babel, M. E. (2005). The sequent ial cueing effect in childrens speech production. Applied Psycholinguistics 26 157-174. Munson, B., Edwards, J., & Beckman, M. ( 2005). Phonological knowledge in typical and atypical speech-sound development. Topics in Language Disorders, 25 (3), 190-206. National Institute of Child Health and Human Development. (2000). Report of the national reading panel: An evide nce-based assessment of th e scientific literature on reading and its implications for reading instruction Bethesda, MD: NICHD Clearing House. Nittrouer, S. (1992). Age-related differences in perceptual effects of formant transitions within syllables and across syllable boundaries. Journal of Phonetics, 20 1-32. Nittrouer, S. (1996). The relation between speech perception and phonemic awareness: Evidence from low SES children and children with chronic OM. Journal of Speech and Hearing Research, 39 1059-1070. Nittrouer, S. (2004). The role of temporal and dynamic signal components in the perception of syllable-final stop voicing by children and adults. Journal of Acoustical Society of America, 115 (4), 1777-1790.
67 Nittrouer, S., & Burton, L. T. (2005). The ro le of early language experience in the development of speech perception and phonological processing abilities: Evidence from five-year-olds with histories of otitis media with effusion and low socioeconomic status. Journal of Communication Disorders, 38 29-63. Nittrouer, S., & Crowther, C. S. (1998). Examin ing the role of auditory sensitivity in the developmental weighting shift. Journal of Speech, Language, Hearing Research, 41, 809-818. Pollock, K. E., Bailey, G., Berni, M., Fletch er, D., Hinton, L., Johns on, I., Roberts, J., & Weaver, R. (1998). Phonological features of African American Vernacular English (AAVE) Retrieved November 1, 2006, from website: http://www.rehabmed.ualberta .ca/spa/phonolgy/features.htm Poplack, S. (Ed.). (2000). The English history of African American English Malden, Massachusetts: Blackwel l Publishers Inc. Raphael, L. (1972). Preceding vowel duration as a cue to the percep tion of the voicing Characteristic of word-final c onsonants in American English. Journal of the Acoustical Society of America, 51 1296-1303. Rvachew, S. (2006). Longitudinal predictors of implicit phonological awareness skills. American Journal of Speech Language Pathology, 15 165-176. Rickford, J. R. (1999). Features and use. African American Vernacular English: Features, evolution, educ ational implications (pp. 4-5-14). Malden, MA: Blackwell Publishers. Schachter, F. F. (1979). Everyday mother talk to toddlers: Early Intervention New York: Academic Press. Seymour, H. N., & Ralabate, P. K. (1985) The acquisition of a phonologic feature of Black English. Journal of Communication Disorders, 18 139-148. Seymour, H. N., Roeper, T. W., & deVilliers, J. (2003). The Diagnostic Evaluation of Language Variation : Psychological Corporation. Seymour, H. N., & Seymour, C. M. (1981). Black English and standard American English contrasts in consonantal developm ent of four and fi ve-year-old children. Journal of Speech and Hearing Disorders, 46 274-280. Silliman, E. R., Bahr, R. H., Wilkinson, L. C., & Turner, C. R. (2002). Language variation and struggling reader s: Finding patterns in dive rsity. In K. G. Butler & E. R. Silliman (Eds.), Speaking, reading, and writing in children with language learning disorders: New paradigms in research and practice (pp. 109-148). Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
68 Sligh, A. C., & Conners, F. A. (2003). Rela tion of dialect to phonological processing: African Vernacular English vs. Standard American English. Contemporary Educational Psychology, 28 205-228. Snow, C. E., Burns, M. S., & Griffen, P. (1998). Preventing reading difficulties in young children. Washington, D.C.: National Academy Press. Storkel, H. L., & Morrisette, M. L. (2002) The lexicon and phonology: Interactions in language acquisition. Language, Speech, and Hearing Services in School, 33 24-37. Thomas-Tate, S., Washington, J., & Edwards, J. (2004). Standardized assessments of phonological awareness skills in low-income African American first graders. American Journal of Speech -Language Pathology, 13 182-190. Torgesen, J. K., & Bryant, B. R. (1994). Test of Phonological Awareness Austin, TX: Pro-Ed. Treiman, R., & Barry, C. (2000). Dialect a nd authography: Some differences between American and Bristish spellers. Journal of Experiment al Psychology: Learning, Memory,and Cognition, 26 1423-1430. Treiman, R., Goswami, U., Tincoff, R., & Leevers, H. (1997). Effects of dialect on American and British children's spelling. Child Development, 68 229-245. Tunmer, W. E., & Chapman, J. W. (1998) Language prediction skill, phonological recoding ability, and beginning reading. In C. Hulme & R. M. Joshi (Eds.), Reading and spelling: Development and disorders (pp. 33-67). Mahwah, NJ: Lawrence Erlbaum Associates. Vellutino, F. R., & Scanlon, D. M. (1987) Phonological coding, phonol ogical awareness, and reading ability: Evid ence from a longitudinal and experimental study. MerrillPalmer Quarterly 33, 321-363. Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. (1999). Comprehensive Test of Phonological Processing Austin, TX: Pro-Ed. Walker, D., Greenwood, C., Hart, B., & Cart, J. (1994). Prediction of school outcomes based on early language production and socioeconomic factors. Child Development, 65 606-621.
70 Appendix A: Vowel duration and closure duration measurements The following key explains what the values are: A. Beginning vowel: where the duration of the vowel began for the specified CV unit B. Ending vowel: where the durat ion of the vowel ended C. Difference: the difference between A and B D. Total ms: the total difference in milliseconds E. Beginning CD: where the closure duration began F. Ending CD: where the closure duration ended G. Differences: the differences between E and F H. Total ms: the total difference in milliseconds I. VD difference: the difference between th e two vowel durations divided by 4 J. CD difference: the difference between the two closure durations divided by 4 K. 0%: values representing 0% change to both vowel duration and closure durations L. 25%: values representing 25% change to both vowel duration and closure durations M. 50%: values representing 50% change to both vowel duration and closure duration N. 75%: values representing 75% change to both vowel duration and closure duration P. 100%: values representing 100% change to both vowel duration and closure duration Talker 1 ud ut A Beg Vowel 0.204001 A beg vowel 0.235091 B End vowel 0.4479 B end vowel 0.365298 C difference 0.243899 C difference 0.130208 D total ms 244 D total ms 130 E. beg CD 0.461084 E bed CD 0.393754 F end CD 0.569379 F end CD 0.567077 G difference 0.108295 G difference 0.173323 H total ms 108 H total ms 173 I. VD difference=114/4=28.5ms J CD difference=65/4=16.25ms
71 K 0% L. 25% M. 50% N 75% P .100% 224ms 215.5ms 187ms 158.5ms 130ms 108ms 124.25ms140.5ms 156.75ms173ms Talker 1 ug uk Beg Vowel 0.201352 beg vowel 0.246024 End vowel 0.446565 end vowel 0.369249 difference 0.245213 difference 0.123226 total ms 245 total ms 123 beg CD 0.45545 bed CD 0.405287 end CD 0.555845 end CD 0.569201 difference 0.100395 difference 0.163914 total ms 100 total ms 164 VD difference=122/4=30.5ms CD difference=64/4=16ms 0% 25%50%75%100% 245ms 214.5ms 184ms 153.5ms 123ms 100ms 116ms 132ms 148ms 164ms Talker 1 ib ip Beg Vowel 0.128645 beg vowel 0.224203 End vowel 0.324406 end vowel 0.360518 difference 0.195761 difference 0.136315 total ms 196 total ms 136 beg CD 0.34343 bed CD 0.388286 end CD 0.471688 end CD 0.559943 difference 0.128257 difference 0.171656 total ms 128 total ms 172 VD difference=60/4=15ms CD difference=44/4=11ms 0% 25%50%75%100% 196ms 181ms 166ms 151ms 136ms 128ms 139ms 150ms 161ms 172ms Talker 1 ig ik Beg Vowel 0.141534 beg vowel 0.125615 End vowel 0.385043 end vowel 0.218614 difference 0.243509 difference 0.092998
72total ms 244 total ms 93 beg CD 0.404217 bed CD 0.245772 end CD 0.495293 end CD 0.40461 difference 0.091076 difference 0.158838 total ms 91 total ms 159 VD difference=151/4=37.75ms CD difference=68/4=17ms 0% 25%50%75%100% 244ms 206.25ms168.5ms 130.75ms93ms 91ms 108ms 125ms 142ms 159ms Talker 1 ahb ahp Beg Vowel 0.139978 beg vowel 0.159293 End vowel 0.469936 end vowel 0.301164 difference 0.329958 difference 0.14187 total ms 330 total ms 142 beg CD 0.4795 bed CD 0.326498 end CD 0.571155 end CD 0.504681 difference 0.091655 difference 0.178183 total ms 92 total ms 178 VD difference=188/4=47ms CD difference=86/4=21.5ms 0% 25%50%75%100% 330ms 283ms 236ms 189ms 142ms 92ms 113.5ms 135ms 156.5ms 178ms Talker 1 ahg ahk Beg Vowel 0.256477 beg vowel 0.091675 End vowel 0.559814 end vowel 0.264954 difference 0.303336 difference 0.173279 total ms 303 total ms 173 beg CD 0.575025 bed CD 0.306015 end CD 0.660031 end CD 0.44973 difference 0.085006 difference 0.143715 total ms 85 total ms 144 VD difference=130/4=32.5ms CD difference=59/4=14.75ms
730% 25%50%75%100% 303ms 270.5ms 238ms 205.5ms 173ms 85ms 99.75ms 114.5ms 129.25ms144ms Talker 1 ab ap Beg Vowel 0.14152 beg vowel 0.158013 End vowel 0.4565 end vowel 0.301963 difference 0.31498 difference 0.14395 total ms 315 total ms 144 beg CD 0.472991 bed CD 0.333452 end CD 0.574411 end CD 0.515188 difference 0.10142 difference 0.181736 total ms 101 total ms 182 VD difference=171/4=42.75ms CD difference=81/4=20.25ms 0% 25%50%75%100% 315ms 272.25ms229.5ms 186.75ms144ms 101ms 121.25ms141.5ms 161.75ms182ms Talker 1 ag ak Beg Vowel 0.103587 beg vowel 0.10763 End vowel 0.427683 end vowel 0.259979 difference 0.324097 difference 0.15235 total ms 324 total ms 152 beg CD 0.441679 bed CD 0.296543 end CD 0.524912 end CD 0.452375 difference 0.083234 difference 0.155832 total ms 83 total ms 156 VD difference=172/4=43ms CD difference=73/4=18.25ms 0% 25%50%75%100% 324ms 281ms 238ms 195ms 152ms 83ms 101.25ms119.5ms 137.75ms156ms Talker 2 ud ut Beg Vowel 0.25438 beg vowel 0.319775 End vowel 0.542872 end vowel 0.435623 difference 0.288492 difference 0.115848 total ms 288 total ms 116
74 beg CD 0.550634 bed CD 0.458793 end CD 0.647661 end CD 0.594652 difference 0.097027 difference 0.135859 total ms 97 total ms 139 VD difference=172ms/4=43ms CD difference=42ms/4=10.5ms 0% 25%50%75%100% 228ms 245ms 202ms 159ms 116ms 97ms 107.7ms 118ms 128.5ms 139ms Talker 2 ug uk Beg Vowel 0.345005 beg vowel 0.341146 End vowel 0.569302 end vowel 0.471358 difference 0.224297 difference 0.130212 total ms 224 total ms 130 beg CD 0.598461 bed CD 0.491637 end CD 0.684815 end CD 0.605839 difference 0.086354 difference 0.114202 total ms 86 total ms 114 VD difference=94ms/4=23.5ms CD difference=28ms/4=7ms 0% 25%50%75%100% 224ms 200.5ms 177ms 153.5ms 130ms 86ms 93ms 100ms 107ms 114ms Talker 2 ib ip Beg Vowel 0.314669 beg vowel 0.182359 End vowel 0.500915 end vowel 0.295211 difference 0.186246 difference 0.112852 total ms 186 total ms 113 beg CD 0.511813 bed CD 0.305355 end CD 0.626731 end CD 0.44165 difference 0.114918 difference 0.13631 total ms 115 total ms 136 VD difference=73ms/4=18.25ms CD difference=21ms/4=5.25ms 0% 25%50%75%100%
75186ms 167.75ms149.5ms 131.25ms113ms 115ms 120.25ms125.5ms 130.75ms136ms Talker 2 ig ik Beg Vowel 0.282011 beg vowel 0.307454 End vowel 0.510322 end vowel 0.392682 difference 0.228311 difference 0.085228 total ms 228 total ms 85 beg CD 0.520641 bed CD 0.41073 end CD 0.605774 end CD 0.513004 difference 0.085133 difference 0.102274 total ms 85 total ms 102 VD difference=143ms/4=35.75ms CD difference=17ms/4=4.25ms 0% 25%50%75%100% 228ms 192.25ms156.5ms 120.75ms85ms 85ms 89.25ms 93.5ms 97.75ms 102ms Talker 2 ahb ahp Beg Vowel 0.223593 beg vowel 0.339081 End vowel 0.52793 end vowel 0.484116 difference 0.304337 difference 0.145036 total ms 304 total ms 145 beg CD 0.535586 bed CD 0.503034 end CD 0.623633 end CD 0.61654 difference 0.088047 difference 0.113506 total ms 88 total ms 114 VD difference=159ms/4=39.75ms CD difference=26ms/4=6.5ms 0% 25%50%75%100% 304ms 264.25ms224.5ms 184.75ms145ms 88ms 94.5ms 101ms 107.5ms 114ms Talker 2 ahg ahk Beg Vowel 0.348263 beg vowel 0.335198 End vowel 0.599914 end vowel 0.474217 difference 0.25165 difference 0.139018 total ms 252 total ms 139
76beg CD 0.654842 bed CD 0.511143 end CD 0.734042 end CD 0.604546 difference 0.0792 difference 0.93403 total ms 79 total ms 93 VD difference=113ms/4=28.25ms CD difference=14ms/4=3.5ms 0% 25%50%75%100% 252ms 223.75ms195.5ms 167.25ms139ms 79ms 82.5ms 86ms 89.5ms 93ms Talker 2 ab ap Beg Vowel 0.266631 beg vowel 0.338607 End vowel 0.606721 end vowel 0.507012 difference 0.340091 difference 0.168405 total ms 340 total ms 168 beg CD 0.615851 bed CD 0.521321 end CD 0.711716 end CD 0.651203 difference 0.095864 difference 0.129881 total ms 96 total ms 130 VD difference=172ms/4=43ms 172ms/ CD difference=34ms/4=8.5ms 0% 25%50%75%100% 340ms 297ms 254ms 211ms 168ms 96ms 104.5ms 113ms 121.5ms 130ms Talker 2 ag ak Beg Vowel 0.341048 beg vowel 0.385809 End vowel 0.62094 end vowel 0.568522 difference 0.27989 difference 0.182712 total ms 280 total ms 183 beg CD 0.642767 bed CD 0.599156 end CD 0.731357 end CD 0.687777 difference 0.08859 difference 0.088621 total ms 89 total ms 89 VD difference=97ms/4=24.25ms CD difference=0ms/4=0ms 0% 25%50%75%100% 280ms 255.75ms231.5ms 207.25ms183ms
7789ms 89ms 89ms 89ms 89ms Talker 3 ud ut Beg Vowel 1.055511 beg vowel 2.364414 End vowel 1.255361 end vowel 2.683895 difference 0.19985 difference 0.319481 total ms 200 total ms 319 beg CD 1.260342 bed CD 2.68951 end CD 1.407272 end CD 2.787207 difference 0.14693 difference 0.097697 total ms 147 total ms 98 vowel difference=119ms/4=29.75ms CD difference=49ms/4=12.25ms 0% 25%50%75%100% 200ms 229.75ms259.5ms 289.25ms319ms 147ms 134.75ms122.5ms 110.25ms98ms Talker 3 ug uk Beg Vowel 0.375865 beg vowel 0.417305 End vowel 0.716877 end vowel 0.576818 difference 0.341012 difference 0.159513 total ms 341 total ms 160 beg CD 0.718292 bed CD 0.604669 end CD 0.817341 end CD 0.784438 difference 0.099049 difference 0.179768 total ms 100 total ms 180 VD difference=181ms/=45.25ms CD difference=80ms/4=20ms 0% 25%50%75%100% 341ms 295.75ms250.5ms 205.25ms160ms 100ms 120ms 140ms 160ms 180ms Talker 3 ib ip Beg Vowel 0.975756 beg vowel 0.028068 End vowel 1.199667 end vowel 0.168343 difference 0.223911 difference 0.140276 total ms 224 total ms 140 beg CD 1.217641 bed CD 0.176291
78end CD 1.326002 end CD 0.317362 difference 0.108361 difference 0.14107 total ms 108 total ms 141 vowel difference=84ms/4=21ms CD difference=33ms/4=8.25ms 0% 25%50%75%100% 224ms 203ms 182ms 161ms 140ms 108ms 116.25ms124.5ms 132.75ms141ms Talker 3 ig ik Beg Vowel 0.34395 beg vowel 0.328025 End vowel 0.592343 end vowel 0.495644 difference 0.248393 difference 0.167619 total ms 248 total ms 168 beg CD 0.594903 bed CD 0.510368 end CD 0.702455 end CD 0.631552 difference 0.107552 difference 0.121184 total ms 108 total ms 121 VD difference=80ms/4=20ms CD difference=13ms/4=3.25ms 0% 25%50%75%100% 248ms 228ms 208ms 188ms 168ms 108ms 111.25ms114.5ms 117.75ms121ms Talker 3 ahb ahp Beg Vowel 0.339963 beg vowel 0.367821 End vowel 0.719502 end vowel 0.586706 difference 0.37954 difference 0.21885 total ms 380 total ms 219 beg CD 0.724247 bed CD 0.598502 end CD 0.829806 end CD 0.726949 difference 0.105559 difference 0.128447 total ms 106 total ms 128 VD difference=161ms/4=40.25ms CD difference=22ms=4=5.5ms 0% 25%50%75%100% 380ms 339.75ms299.5ms 259.25ms219ms
79106ms 111.5ms 117ms 122.5ms 128ms Talker 3 ahg ahk Beg Vowel 0.058436 beg vowel 0.04361 End vowel 0.362615 end vowel 0.23326 difference 0.304179 difference 0.189651 total ms 304 total ms 190 beg CD 0.394665 bed CD 0.260506 end CD 0.482351 end CD 0.405281 difference 0.087686 difference 0.144775 total ms 88 total ms 145 vowel difference=114ms/4=28.5ms CD difference=57ms/4=14.25ms 0% 25%50%75%100% 304ms 275.5ms 247ms 218.5ms 190ms 88ms 102.25ms116.5ms 130.75ms145ms Talker 4 ud ut Beg Vowel 0.130525 beg vowel 0.120764 End vowel 0.291632 end vowel 0.226691 difference 0.161107 difference 0.105927 total ms 161 total ms 106 beg CD 0.297698 bed CD 0.246371 end CD 0.368478 end CD 0.321041 difference 0.070779 difference 0.07467 total ms 71 total ms 75 VD difference=55/4=13.75ms CD difference=4/4=1ms 0% 25%50%75%100% 161ms 147.25ms133.5ms 119.75ms106ms 71ms 72ms 73ms 74ms 75ms Talker 4 ug uk Beg Vowel 0.174006 beg vowel 0.168975 End vowel 0.304053 end vowel 0.274496 difference 0.130047 difference 0.105521 total ms 130 total ms 106 beg CD 0.309103 bed CD 0.292689
80end CD 0.368445 end CD 0.358186 difference 0.059342 difference 0.065496 total ms 54 total ms 65 VD difference=24/4=6ms CD difference=11/4=2.75ms 0% 25%50%75%100% 130ms 124ms 118ms 112ms 106ms 54ms 56.75ms 59.5ms 62.25ms 65ms Talker 4 ib ip Beg Vowel 0.116005 beg vowel 0.173576 End vowel 0.288275 end vowel 0.297004 difference 0.172269 difference 0.123429 total ms 172 total ms 123 beg CD 0.293584 bed CD 0.305821 end CD 0.36143 end CD 0.409727 difference 0.067846 difference 0.103907 total ms 69 total ms 104 VD difference=49/4=12.25ms CD difference=35/4=8.75ms 0% 25%50%75%100% 172ms 159.75ms147.5ms 135.25ms123ms 69ms 77.75ms 86.5ms 95.25ms 104ms Talker 4 ig ik Beg Vowel 0.131211 beg vowel 0.130459 End vowel 0.267728 end vowel 0.248742 difference 0.136517 difference 0.118283 total ms 137 total ms 118 beg CD 0.287554 bed CD 0.258545 end CD 0.33627 end CD 0.385323 difference 0.048716 difference 0.126779 total ms 49 total ms 127 VD difference=19/4=4.75ms CD difference=78/4=19.5ms 0% 25%50%75%100% 137ms 132.25ms127.5ms 122.75ms118ms 49ms 68.5ms 88ms 107.5ms 127ms
81 Talker 4 ahb ahp Beg Vowel 0.140102 beg vowel 0.157893 End vowel 0.344593 end vowel 0.324204 difference 0.204491 difference 0.166309 total ms 204 total ms 166 beg CD 0.356705 bed CD 0.334745 end CD 0.432944 end CD 0.416728 difference 0.076239 difference 0.081983 total ms 76 total ms 82 VD difference=38/4=9.5ms CD difference=6/4-1.5ms 0% 25%50%75%100% 204ms 194.5ms 185ms 175.5ms 166ms 76ms 77.5ms 79ms 80.5ms 82ms Talker 4 ahg ahk Beg Vowel 0.099181 beg vowel 0.168676 End vowel 0.276578 end vowel 0.30627 difference 0.177397 difference 0.137594 total ms 177 total ms 138 beg CD 0.300621 bed CD 0.326024 end CD 0.374049 end CD 0.396865 difference 0.073428 difference 0.070841 total ms 73 total ms 71 VD difference=39/4=9.75ms CD difference=2/4=.5ms 0% 25%50%75%100% 177ms 167.25ms157.5ms 147.75ms138ms 73ms 72.5ms 72ms 71.5ms 71ms
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 001917283
007 cr mnu|||uuuuu
008 071119s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002012
Kile, Stacy Nicole.
The influence of dialect on the perception of final consonant voicing
h [electronic resource] /
by Stacy Nicole Kile.
[Tampa, Fla.] :
b University of South Florida,
ABSTRACT: Children at risk for reading problems also have difficulty perceiving critical differences in speech sounds (Breier et al., 2004; Edwards, Fox, & Rogers, 2003; de-Gelder & Vroomen, 1998). These children rely more heavily on context than the acoustic qualities of sound to facilitate word reading. Dialect use, such as African American English (AAE) may influence literacy development in similar ways. Dialect use has been shown to affect speech sound processing and can even result in spelling errors (Kohler, et al., in press). The purpose of this study is to determine if children who speak AAE process cues indicative of final consonant voicing differently than children who speak a more mainstream dialect of English. Twenty-six typically developing children in grades K-2 who spoke either AAE or a more mainstream American English dialect participated. The speech stimuli consisted of nonsense productions of vowel + plosive consonant.These stimuli were systematically altered by changing the vowel and stop-gap closure duration simultaneously, which resulted in the final consonant changing from a voiced consonant, like "ib", to a voiceless consonant, like "ip". Two tasks were developed: a continuum task where the child had to indicate when the stimuli changed in voicing and a same-different task which involved determining if two stimuli were identical in voicing or not. No significant differences between groups were found for dialect use or grade for the same/different task. In the continuum task, chi-square analyses revealed significant differences in response patterns attributable to dialect and grade. In addition, a significant consonant by speaker interaction was found for mean ratings. Correlations between mean continuum rating and phonological awareness composites were not significant.In conclusion, it was evident that children who speak AAE present with differences in their perception of final consonants in VC nonsense syllables. This finding suggests the dialect speakers may be using different cues to make judgments regarding the speech signal, or that the speakers of AAE have a less mature ability to extract fine phonetic detail due to the influence of their dialect (Baran & Seymour, 1979). More research is warranted to determine the exact role that dialect plays.
Thesis (M.S.)--University of South Florida, 2007.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 81 pages.
Advisor: Ruth H. Bahr, Ph.D.
African American English.
Final consonant devoicing.
x Speech Language Pathology
t USF Electronic Theses and Dissertations.