Just another WordPress.com site

Archive for the ‘Linguistics’ Category

A sociophonetic analysis of L2 substitution sounds of American English interdental fricatives



Andrew Seibert

B.A., Wright State University, 2009

A Thesis

Submitted in Partial Fulfillment of the Requirements for the

Master of Arts Degree.

Department of Linguistics

in the Graduate School

Southern Illinois University Carbondale

August 2011


Andrew Seibert, for the Master of Arts degree in Applied Linguistics/TESOL, presented on May 5, 2011, at Southern Illinois University Carbondale.


          Previous research done in sociophonetic variation of second language speakers has often looked at constraints of formality affecting degree of foreign accent and how this degree of formality can have effects on what speaking styles speakers choose to employ. Furthermore, other social constraints of convergence and divergence of speech affect speaker speaking style. However, no known previous research has examined interdental fricative /θ ð/ substitution based on each speaker’s interlocutor. This study explores second language speakers’ English interdental fricative substitution sounds in terms of sociophonetic variation of formality and speaker interlocutor(s). Five native language pairs of Arabic, Cantonese, French, Portuguese, and Vietnamese origin were part of the study, comprising ten participants in total. The study finds age of English onset, as verified by the literature, to be the most determining factor for accurate articulation of these marked fricatives. However, other constraints for substitution choice are at hand including phonological limitations and estimated linguistic experience based on demographic information given by survey participants. The primary aim of the study is to associate some of the interdental fricative substitutions with a social variable. Data for the study include recordings of each participant reading a poem by him/herself, a dialogue with the other same native language participant, and a dialogue with a native speaker of American English. The data analysis examined the replacement sounds in terms of native language background, linguistic experience variables, and phonological constraints. In addition, quantities and ratios of specific replacement sounds for each participant per recording and per native language pair were compared and contrasted to find if speech accommodation theory (SAT), as proposed by Giles et al. (1991), played a role in any of the participants’ choices for substitution. The study finds both convergence and divergence of interdental fricative substitutions to be characteristic of speakers with less linguistic experience in English. An additional stronger finding is that most participants’ most common sound substitutions for the voiced and voiceless interdental fricatives were independent in place and manner, the voiced most commonly replaced by dental and alveolar plosives [ḏ d] and the voiceless most commonly replaced by labiodental fricative [f], which could be an indication of each fricative’s acoustic and phonemic representation in each non-native speaker’s phonological component, supported by findings of Brannen (2002). Some literature suggests that varying values of [continuant] in speakers’ native languages are the means by which speakers choose the replacement sounds they do. However, such an explanation cannot be the only valid one when inherent variability comes into play and different places and manners of articulation are chosen for both interdental fricatives.

 Keywords: L2 acquisition, Degree of foreign accent, The Critical Period Hypothesis, Stabilization, Markedness, Accommodation Theory, L1 transfer


CHAPTER                                                                                                                        PAGE

ABSTRACT ………………………………………………………………………………………………..i

ACKNOWLEDGEMENTS  ……………………………………………………………………………..iii

LIST OF TABLES  ………………………………………………………………………………………..vi


CHAPTER 1 – Introduction and Literature Review …………………………………………………..1

1.1 – Accommodation Theory ………………………………………………………………………1

1.2 Linguistic/Social Category………………………………………………………………………4

1.3 Foreign Accent In An L2 ……………………………………………………………………….5

1.4 English Interdental Fricatives ……………………………………………………………….12

1.4.1 Predicted Participant Substitution & Hypothesis ………………………..15

CHAPTER 2 – Procedure ……………………………………………………………………………………..17

2.1 – Participants ………………………………………………………………………………………20

2.2 – Instrument Description ………………………………………………………………………22

CHAPTER 3 – Phonetic Data Analysis ………………………………………………………………….25

3.1 – Phonological Processes Governing Substitution ……………………………………29

3.2 – Word Place Governing Substitution …………………………………………………….33

3.3 – Data Cumulatively …………………………………………………………………………….35

CHAPTER 4 – Sociophonetic Data Analysis & Discussion ………………………………………46

4.1 – Substitutions Per Social Realm  …………………………………………………………..47

4.2 – Accurate Articulation V. Demographic Information ………………………………51

4.3 – Substitutions Per Speaker …………………………………………………………………..54

4.3.1 – Arabic Speakers  ………………………………………………………………….55

4.3.2 – Cantonese Speakers  …………………………………………………………….57

4.3.3 – French Speakers  ………………………………………………………………….60

4.3.4 – Portuguese Speakers  ……………………………………………………………63

4.3.5 – Vietnamese Speakers  …………………………………………………………..66

4.4 – Accurate Articulation V. Realms of Varying Formality …………………………69

4.5 Hypothesis Revisited …………………………………………………………………………..70

4.6 Discussion Results ………………………………………………………………………………70

CHAPTER 5 – Conclusion  …………………………………………………………………………………..74

5.1 Limitations …………………………………………………………………………………………75

5.2 Areas of Further Research ……………………………………………………………………78

REFERENCES  ………………………………………………………………………………………………….79


Appendix A – Instruments  ……………………………………………………………………………………88

Appendix B – Data  ……………………………………………………………………………………………..91

Vita ………………………………………………………………………………………………………………….111




4.1.1 Voiceless replacement sounds in Monologue Reading……….. 47

4.1.2 Voiceless replacement sounds in dialogue with non-native speaker……….. 48

4.1.3 Voiceless replacement sounds in dialogue with native speaker……….. 49

4.1.4 Voiced replacement sounds in Monologue Reading……….. 50

4.1.5 Voiced replacement sounds in dialogue with non-native speaker……….. 50

4.1.6 Voiced replacement sounds in dialogue with native speaker……….. 51

4.2.1 Demographic information……….. …………………………………………………………………………52-3 Arabic1……….. ………………………………………………………………………………………………..55 Arabic2……….. ………………………………………………………………………………………………..56 Cantonese1…………………………………………………………………………………………..58 Cantonese2…………………………………………………………………………………………….59 French1………………………………………………………………………………………………..60-1 French2……….. ………………………………………………………………………………………………..61 Portuguese1……….. ………………………………………………………………………………………….63 Portuguese2……….. ………………………………………………………………………………………….64 Vietnamese1…………………………………………………………………………………………66-7 Vietnamese2…………………………………………………………………………………………67

4.4.1 Accurate articulation ratios per social realm………………………………………………..69


Speech diversity is the research topic of many studies in the sociology of language, primarily, looking at language variation in different social contexts and offering explanations for linguistic variation under many social conditions. Much of the research in the multidisciplinary field of variationist sociolinguistics moves to understand and identify the relationship between linguistic and social variables. A specific area of sociolinguistics that has been the study of many researchers in the field is second language acquisition and second language speakers in various social contexts. The literature on the sociolinguistics of L2 acquisition mainly focuses on pedagogical theories of L2 sociolinguistics, from consciousness in L2 learning (Schmidt, 1990) to L2 gender-based language use (Ehrlich, 1997) to theories toward keeping a culturally competent second language classroom (Van Lier, 1988). Literature concerning L2 speakers in various social parameters involves topics covering application of existing sociolinguistic theories to second language speakers (Wolfson & Elliot, 1983).


            The most relevant sociolinguistic theory for this study is Speech (or Communication) Accommodation Theory, henceforth SAT (Giles et al., 1991). The central concern of the paradigm focuses on the cognitive processes that link speakers’ perception of their interlocutors to their speech habits and other communicative behaviors. Specifically, looking at theories of convergence and divergence, we can find out underlying motivations for these linguistic behaviors. Convergence of speech styles is when a speaker makes his speech more similar to his or her interlocutor’s speech characteristics (speech rate, utterance lengths, pronunciations, etc.) Psychological/ cognitive processes motivating speakers to converge their speech include but are not limited to: desire for their interlocutors’ social approval, desire for a high level of communication efficiency, and/or social norms are not perceived to dictate alternative speech strategies (Beebe & Giles 1984). Divergence is the opposite and is where speakers tend to maintain their own speech styles or even diverge from the perceived speech styles of their interlocutors. Reasons motivating speakers to diverge include but are not limited to: defining the encounter in intergroup terms and the desire for a positive ingroup identity, disassociating oneself personally from another in an interaction with another speaker, and/or desiring to bring interlocutors’ speech behaviors to a personally acceptable level (Bourhis & Giles, 1977; Taylor & Royer, 1980; Cappella, 1981).

Looking specifically at research in SAT being applied to second language speakers, we find several studies looking at Giles’ theory in different conditions with different variables, including an altered overview of speech accommodation in terms of ingroups and outgroups, termed ethnolinguistic identity theory (Giles & Johnson, 1987). This theory seeks to define theoretical and experimental confusion by identifying common psychological processes that underlie various assorted speech acts. Giles’ ethnolinguistic theory draws on previous research on intergroup behavior (Tajfel, 1974). Tajfel’s social identity theory states that all human beings categorize society into various groups or categories, and, hence, perceive themselves and others as being a part of one or more of these diverse groups. This perception of our and others’ place in these categories is called social identity. However, this is not a social constructionist sense of social identity but simply Tajfel’s use of the term. Social identity can be both positive and negative in that it depends on how the in-group socially compares to relevant outgroups. Generally, it is argued that most people try to achieve a positive identity by performing actions or linguistic acts that are deemed favorable by an outgroup. Giles & Johnson (1987)’s study begins to clarify some of the complex social conditions that ethnic minorities use to weaken or strengthen their distinctive ethnolinguistic styles, however, Giles focuses mostly on intergroup communication.

Much of previous research in L2 speech patterns says little about substitutions for marked sounds in a second language. This sociophonetic area of research wants to explain why second language speakers choose the replacement sounds they do in fulfillment of marked or unfamiliar sounds in the second language, and such an explanation is one of the primary goals of this study. Beebe (1977) found that Chinese/Thai bilinguals’ accents were directly correlated to their interlocutors. Chinese-Thai bilinguals selected a “Thai” variant for the nine phonological variables a higher percentage of the time when they were speaking Thai to a Thai listener than when they were speaking Thai to a Chinese listener. And conversely, they chose a “Chinese” variant more frequently when speaking to a Chinese-Thai than to an ethnic Thai. However, Beebe’s findings could be the result of Thai and Chinese having a close genetic relationship linguistically. Many studies have also been carried out to understand the extent of L1 transfer into an L2 depending on the degree of formality. Dickerson & Dickerson (1977) and Gatbonton (1978) all find that the more formal the style of speech, the less likely there is to be transfer from the first language. That is that in more formal speech, second language speakers are more likely to have a closer-to-native accent.


            Non-native accents, like native accents, can be social markers. The finest variation in phonetic detail can make an individual speaker’s interlocutors perceive the speaker as belonging to a specific social category.  Social categories can be as broad as class distinctions between speakers or as narrow as mannerisms and sexual orientation. A good example study of the former (phonetic class distinctions) is Labov’s 1966 study of the differences in suppliance of the retroflex liquid /ɹ/ in the phrase “fourth floor” /fɔɹθ flɔɹ/ between social classes in New York City. Labov found that /ɹ/-ful speech is more typical of high class department store workers, as the employees at Saks Fifth Avenue, whom Labov classified as high class had 62% suppliance of the retroflex liquid, while employees in Klein’s department store, whom Labov classified as low class only supplied the liquid 20% of the time. In addition to this observation, Labov tried to capture the suppliance of the word-medial and word-final liquid in both casual and careful speech. He found that the employees at Macy’s, who he classified as middle class, had the biggest difference in suppliance of /ɹ/ between casual and careful speech. This reveals linguistic insecurity on behalf of the Macy’s employees, as they had the most /ɹ/-fulness in their careful speech as compared with their casual speech. The Macy’s employees by supplying more retroflex liquid sounds in their careful speech were trying to put as much distance between themselves and the lower class, whose speech was nearly half as /ɹ/-ful. Mack’s (2010) study is a good example of the latter, narrower identity. Mack’s study bridges the gap of sexual orientation identity data outside of English by surveying listener sensitivity and acute perception of vowel qualities in Puerto Rican Spanish. Mack asked native Puerto Rican Spanish speakers to identify each speaker’s height, age, and social class in addition to sexual orientation. She found that higher F2 frequencies of two tokens of the unrounded front vowel /e/ was a significant factor in listener perception of homosexuality.


            Degree of foreign accent in a second language has been the topic of much research since Asher & García (1969). Thompson (1991) offers several good reasons for studying this linguistic phenomenon, among which are possible answers for what social and educational variables hinder accurate phonological acquisition. Piske et al. (2001) commits to an extensive review of methodologies of experiments carried out to test degree of foreign accent factors including ‘elicitation techniques’ (Oyama, 1976; Piper & Cansin, 1988; Elliot, 1995), and  ‘rating techniques’ (Olson & Samuels, 1973; Patkowski, 1980; Fathman, 1975). The researchers also investigate literature researching factors affecting degree of foreign accent including ‘age of L2 learning’ (Scovel, 1969, 1988), ‘length of residence’ (Meador et al., 2000; Purcell & Suter, 1980), ‘gender’ (Snow & Hoefnagel-Höhle, 1977; Tahta et al., 1981), ‘formal instruction’ (Bongaerts et al., 1997; Moyer, 1999), ‘motivation’ (Suter, 1976; Flege et al., 1999(b), and ‘language use’ (Thompson, 1991; Flege et al., 1997). After the comprehensive literature review, Piske et al. (2001) finds that age of L2 onset and continued use of L1 are the largest contributing constraints to the strength of a native-like accent for L2 speakers.

            Second language accents do not mean the speaker is not proficient in the second language because proficiency is based on the ability and quality of communication with the target (or second) language as a medium. This study will look not at linguistic proficiency in English, but at the accuracy of two marked sounds in English, the interdental fricatives /θ ð/. Proficiency of language is so much more than being able to accurately articulate difficult or marked sounds in the language. Language proficiency is also about commanding other linguistic components, such as subsegmental elements like morphology and sentence structure (syntax), as well as suprasegmental elements like stress, prosody, and vocabulary use, along with modality and sociolinguistic performance (Hernandez-Chavez et al., 1978). These elements are all necessary components to master when producing language in order for one to consider him or herself proficient in a language. This study uses the term “experience” instead of “proficiency” to refer to an estimate of how accurately participants articulated the interdental fricatives. Such an English linguistic experience estimate is based on English onset and length of stay in the U.S. only.

Let’s take into account the Critical Period Hypothesis (Lenneberg 1967) as a predicator for linguistic proficiency, Piske et al.’s (2001) review of a large majority of the literature on the hypothesis finds that age of L2 onset of learning appears to be the most determining factor in degree of foreign accent in a second language. Other factors affecting degree of accent are variable and researchers tend to lack the control needed to conduct research on factors like motivation and other study participant characteristics. Various ages have been proposed for the end of the critical period, but the hypothesis’ developer, Eric Lenneberg (1967) proposes the pre-adolescent age of 12 in his Biological Foundations of Language. However, Piske et al. (2001) found something similar to Flege et al. (1997) that speakers who speak their native language more frequently tend to have a stronger accent than those speakers who do not. Both studies used Italian L2 speakers of English. After all, though, L1 use is still not as great a factor affecting foreign accent as age of L2 onset.

Of course, there is the theory of fossilization that could make second language speakers stabilize in their progression toward obtaining a native-like accent, meaning that no matter how much exposure to the second language a participant has, if he or she has stabilized in their use of the language, they will not be able to change errors or mistakes without concerted, conscious effort. Han (2004) describes fossilization as a linguistic phenomenon of non-progression of learning despite continuous exposure to input, adequate motivation to learn, and sufficient opportunity to practice. Fossilization can occur in all areas of linguistics, but it is especially common in respect to non-native phonology. Any foreign accent of second language speakers when speaking English is L1 phonology interfering with L2 phonology. Causes of fossilization are speculative and numerous. A possible explanation is phonological habit formation. The muscles in our mouth become accustomed to articulating certain sounds for so long, that the nerves and muscles triggered in speech production have atrophied so far that articulating new sounds and new sequences of sounds is rather difficult (Tarone, 1987). However, such an argument may not be limited to only muscles and nerves in the mouth, but to neural functions in the central nervous system. Lenneberg’s (1967) says that the loss of flexibility in the brain affects pronunciation of the L2 more so than the syntax or semantics. Krashen (1977) supports a more psychological factor in the onset of stabilization. He concurs the ending of the critical period is the beginning of Inhelder and Piaget’s stage of formal operations the time in which adolescents begin to construct abstract theories about the world (Inhelder & Piaget, 1958). After the onset of this stage, it may inhibit natural language acquisition. A third type of factor causing fossilization is L2 speakers’ underlying lack of empathy for the TL’s native speakers. Guiora et al. (1972) finds that whenever empathy with speakers of a particular L2 is artificially induced, the L2 speaker’s pronunciation improves.

Foreign language accent derives from speaking a language that is not a first language, based on many researchers’ studies Tahta (1981), Derwing & Munro (1997). Many researchers believe such accents emanate from the phonological structure of an L1 not matching up with the phonological structure of the L2, so L1 phonological structures transfer into the L2 interlanguage. Whether that interlanguage stabilizes at some point or continues to change is a point of dispute in the literature.  Selinker (1972) believes that a second-language speaker’s interlanguage can become stabilized at any given stage in development, thus ending the evolution of their interlanguage. Evidence for this permanent stabilization, termed fossilization can be found in Lardiere (1998), in which a Chinese L1 speaker of English was audio recorded two different times: once after being in the US for ten years and then again eight years later. Her suppliance of past-tense marking in obligatory contexts was nearly the same for both recordings, around 34%, thereby showing that at least her past-tense marking had stabilized. Larsen-Freeman (2005) disagrees with Selinker and believes there is no end-state to a second-language speaker’s interlanguage. Therefore a second language speaker will always be in his or her interlanguage. Larsen-Freeman has a more dynamic view of second language acquisition, as she believes that permanent stabilization does not occur.

            Given this background, if we study the speech of speakers with the same L1 who are speaking the same L2, we may find similar phonology in each L2 speaker’s interlanguage. Such evidence of this similar interlanguage between L2 speakers of the same L1 is when native speakers of a given language try to imitate a certain foreign accent, the generalizations of this particular foreign accent are similar. That is, native speakers of a given language often perceive certain phonological characteristics to be of a certain native, or in accordance with this discussion, foreign origin because speakers who have the same native language usually have interlanguages with similar phonological characteristics, such as German final devoicing transferring into English as an L2. Same native language L2 speakers will also often pronounce and replace unfamiliar sounds with the same sound. For example, Russian speakers often replace the English voiceless interdental fricative /θ/ with [t], while Japanese native speakers tend to replace the voiceless fricative with [s]. Lombardi (2003) analyzes these replacement sounds from an Optimality Theoretic approach, finding that replacement sound [t] stems from speakers of a native language with highly ranked segmental markedness constraints, as /t/ is a less marked segment; while [s] replacement rises from speakers with a native language that ranks faithfulness of the input manner highly, as speakers are maintaining faithfulness to the continuancy of the fricative at the expense of the segmental markedness constraints. Of course there are other replacement sounds possible such as [f] by other native language speakers. Hancin-Bhatt (1994) finds that speakers of native languages with [continuant] in low prominence like Hindi and Turkish tend to replace the sounds with [t/d]. While speakers of native languages where [continuant] is of high prominence like German, tend to replace with [s/z]. However, native languages with varying values for [continuant] are not the only constraints affecting interdental fricative replacement by second language speakers of English. Through examination of this study’s data, such an explanation of substitutions is not quite so simple. Brannen (2002) finds that substitutions for the interdental fricatives are based on an auditory phonetic comparison of a fully specified surface form with native-language internal representations. Specifically for labiodental and coronal substitution phones, perceptual confusion of the features Strident (channel turbulence) and Mellow (spread turbulence) are at hand. As examined in later chapters, perceptual acoustic and articulatory features may be at hand in impacting the substitution choices this study’s participants produce for the fricatives.

The interdental fricatives are highly marked sounds. Their presence in linguistic sound systems is actually quite rare, which makes their accurate acquisition notoriously difficult. The interdental sounds are also usually one of the last sounds for children who are acquiring English as their L1 to actually produce correctly (Smit, 1986). The accurate production of these fricatives can develop as late as six years of age in child L1 acquisition of English. The fricatives themselves tend to be so easily replaced in English most likely due to the infrequency of these sounds, in both language phonetic inventories and English. The interdental fricatives are contrastive consonantal phonemes in standard varieties of British, American, and Australian English. Many other English dialects see an elimination of the interdental fricatives from the phonetic inventory, like Cajun English and African American Vernacular English ((Dubois and Horvath, 1998) (Laing 2003)). Blevins (2006) attributes these sound changes to neutralization. Though, many of the dialects that neutralize interdental fricatives in English are spoken by speakers with a previous L1 lacking the fricatives, because these sound changes are so innovative, contact seems a less likely reason for neutralization. Dubois and Horvath (1998) project that the fricatives are highly marked and learned late by children. But markedness alone does not predict that a phoneme will be removed from a language’s phonemic inventory. The Khoisan languages include highly marked clicks in their inventories, and these languages’ speakers have been able to maintain the click sounds for as far back as anyone can construct. Clicks have even been borrowed in contact situations, so high markedness and late acquisition cannot attribute to these sounds’ neutralization or deletion (Blevins, 2004). Blevins (2006) surveys English perception literature and finds labiodental replacement articulation is based on misperception, as the labiodental and interdental fricatives, /f v/ and /θ ð/ respectively, are not easily distinguishable based solely on auditory cues for both native and nonnative speakers. Nonnative speakers often differ from native speakers at the production level instead of the auditory level, which includes both comprehension and perception.

What is the reason for second language speakers’ difficulty in achieving native-like phonological proficiency? Brown (2000) notes that the lack of success in learning or acquiring a second language is often attributed to Universal Grammar (UG) not operating in second language acquisition, unlike first language acquisition. However, White (1989) says that other factors in addition to lack of UG operation, like adequate input and assorted learning techniques, are important in both first and second language acquisition. As most research suggests, second language speakers’ failure to achieve native-like production skills are more attributable to these others factors defined by White and not the lack of operation of UG. Long (1990) says that age of learning (aol),  the terminology used for this study to mean age of onset, is the most determining factor when it comes to achieving a native like accent. Long concludes that an L2 can be spoken without any kind of foreign accent as long as aol is before the age of six, which is different than Lenneberg (1967)’s hypothesis of 12 years of age.


            Both the interdental fricatives occur word initially, medially, and finally in English, but word medial position for the voiceless fricative is much less common than in word initial or final position, while word final position for the voiced fricative is limited to verbs like “breathe,” “teethe” and “sheathe,” all from English’s Proto-Germanic descent. This distribution is due more to historical accident rather than systematic gap, as in Old English it was the voiced realizations [v z ð] that were allophones of the voiceless phonemes /f s θ/. The voiced allophones occurred between two voiced segments, so word-medially. This allophonic distribution is why Modern English applies the distribution of the interdental fricatives it does today (Mitchell & Robinson, 1992) Both voiced and voiceless are very common in word-initial position, but voiced shows up in function and content words and the voiceless in just content words.  In intervocalic position, voiced is more common than voiceless.  And in word-final position, voiceless is more common than voiced.  This all derives from the original allophonic distribution of the two fricatives.

Since both of the sounds have different uncommon word positions, it’s possible that native speakers think of the two interdental fricatives as being two allophones of one phoneme. The interdental fricatives are the only English voiced/voiceless consonants not to have voiced and voiceless distinctions in the orthography. That is, while all English plosives /b p/, / d t/ /g k/, and all other English fricatives /f v/, /s z/, and /ʃ ʒ/ generally have distinctions in English spelling to sound correspondences  (/ʃ/ = sh, ce, ch, ci, s, sci, sch, se, ti  and /ʒ/ = g, s, z), /θ,ð/  do not (Yavas, 2006). Though, in some English words, a voiced/voiceless orthographic character will be used to represent its voiced/voiceless pair, ie. of [ʌv], is [ɪz], business [bɪznəs]. While we know that the voiced and voiceless interdental fricatives are two separate phonemes in English, as they have a minimal pair: ether /iθəɹ/ and either /iðəɹ/, it would be interesting to look at awareness of these two phonemes in both native and nonnative speakers.  Fowler (1991) tells the traditional view of phoneme awareness development being well before alphabetic literacy.

Because the two English interdental fricatives are difficult to learn and represented by the same orthographic characters and because non-native speakers’ pronunciation are not consistent across different formality tasks and different social situations (Tarone 1979, 1982, 1983), this study will collect and analyze data from speakers of several languages to find if social variables of formality and addressee’s realization of each fricative play any kind of a role in the interdental fricative substitutions nonnative speakers of English choose (Briggs 1986). Zampini (1994) finds native English speakers of Spanish decrease in accurate pronunciation during formal reading tasks. Her formality task was an excerpt from a culture text that each participant read aloud. This study’s “Monologue Reading” is similar to Zampini’s (1994) formal reading task. However, while Zampini’s (1994) informal task was answering a series of questions, this study’s informal (or casual) task, refered to as “Dialogue Reading” involves a dialogue to measure speech accommodation in terms of convergence and divergence of substitute phones for the interdental fricatives. Zampini’s findings for less accurate native pronunciation during formal reading tasks differ from the findings of Dickerson & Dickerson (1977) and Gatbonton (1978), which find the opposite.

Formal versus informal (or casual) speech is a matter in difference of social context according to Price (2007). Zampini (1994) describes formal reading tasks as consisting of reading word lists or paragraphs or reciting learned material, while dialogue readings and spontaneous speech typify more informal speech tasks. Major’s (1986, 1987) Ontogeny Model of phonological acquisition describes transfer and developmental errors in L2 pronunciation. Major’s model predicts the results found by Dickerson & Dickerson (1977) and Gatbonton (1978) that transfer errors will occur less often in formal speech versus informal speech. Labov’s (1966) terminology is casual versus careful speech, instead of formal versus informal speech. Labov describes careful speech occurring in a context which employs only one main speaking style, and he defines casual speech as occurring in situations where speakers are less conscious of their speech, so they employ a more relaxed style of speaking and thereby use more than one speaking style. Furthermore, he describes contexts devised by linguists to elicit certain stylistic variation. The interview situation often rates higher on the scale of formality, while reading styles are more characteristic of casual conversation. All of this study’s instruments are reading style contexts, so, inherently; this study is collecting casual speech only. Elicitation of the voiced and voiceless interdental fricatives would be too few and far between in an interview style context, ultimately giving us too few tokens of the fricatives to analyze. However, based on both Labov and Zampini’s description of formality tasks, this study’s analysis treats the Dialogue Readings as more casual speech and the Monologue Reading as more formal speech.


            This section states what cited literature say are possible substitution sounds for speakers with the native languages of the participants involved in this study. Also, this section presents the study’s ultimate hypothesis.

Kharma & Hajjaj (1997) state that the most common substitution sounds of the interdental fricatives for Arabic speakers are of labial and coronal origin, [f, t] for the voiceless and [v, d] for the voiced.

Chan & Li (2000) say that [f] and [t] are very common substitution sounds for Cantonese, while [v] and [d] are common substitutions for the voiced.

Collins & Rodd (1972) say that Francophonic (French speaking) West Africans’ substitution for the fricative can be labiodental [f v], but more likely; substitutions are going to be coronal [s z]. Hypercorrection often confuses phoneme /s z/ with [θ ð], giving the impression of a lisp. Also possible substitutions are dental plosive [ṯ ḏ].

Dreasher & Anderson-Hsieh (1990) find [t] to be the most common substitution of the voiceless interdental fricative /θ/ for Brazilian Portuguese speakers. Other replacement phones for the voiceless include [ḏ s z f]. The two most common substitutions for the voiced interdental fricative are [ḏ θ]. [ḏ] is most common word initially and word medially, while [θ] is more common word finally and in consonant clusters.

Santre (1992) finds that Vietnamese learners will most likely substitute the voiceless fricative with either [t] or [tʰ], or the sound will just be deleted. She goes on to say that the voiced fricative is especially difficult for Vietnamese speakers to master and they are likely to substitute the fricatives with [d] or [θ].

This study’s hypotheses for substitution phones for each participant are based off the cited literature in this section. Though, due to the different literature predictions and different possible phones in labial and coronal positions available for participants to use for substitution, inherent variability is ultimately the hypothesis for this thesis, though, this inherent variability has social constraints (particularly SAT) that can affect it (Wolfram, 1998). Also due to Brannen (2002) findings that substitution is largely based on perception of auditory and articulatory features help the hypothesis that accommodation can occur with interdental fricative substitution phones.




            Chapter 2 contains the study’s primary goals, the data collection procedure, demographic information of all the study’s participants, and a thorough description and justification of the instruments used.

This study specifically seeks to find explanations of non-native speakers’ production of the interdental fricatives due to their often-difficult acquisition. The spelling of words containing these fricatives is confusing and the distribution of these sounds in English words is unbalanced.  In addition, the study investigates what the substitution patterns are that are used by speakers of different L1s due to there being various sound options available that are similar in place, manner, or voicing to the fricatives.  By analyzing spoken data from speakers of several different L1 languages, we can discover if speakers of the same L1 use the same substitutions as each other in various social contexts, primarily formal (monologue) versus informal (dialogue) speech. Included in informal speech are two social contexts, one with a same native language speaker and one with a native speaker of English. The Monologue Reading was to be read like a speech alone, as it is a poem.

The materials used to carry out the data analysis procedure are the instruments, attached in Appendix A, which includes “Demographic Questionnaire”, “Monologue Reading” and “Dialogue Reading” (two copies, one for each speaker). Monologue Reading and two different Dialogue Reading recordings are referred to as the study’s realms of speech.

Participants of the study were students who speak English as a second language at a Midwestern university. They were ten participants with five different native language backgrounds (each native language having two speakers: one male, one female). Participants were recorded in their native language pairs and came together in one session. One participant (called participant 2) was asked to step outside into the hall, closing the door behind him or her. The participant in the researcher’s office (called participant 1) was asked to say the phrase “to this thing I am speaking” into the microphone in order to set an appropriate recording volume. Once the recording volume was set, participant 1 was instructed on how to properly begin and end their recording by clicking the appropriate buttons on the Speech Analyzer program (www.sil.org) recording window. All participants were asked to read “Monologue Reading” (attached as part of “Instruments”) silently to him or herself, and then once he or she was ready, he or she was asked to read “Monologue Reading” into the microphone. The researcher was not in the room for any of the recordings, so not to confuse the social atmosphere in which the participants were recording. Upon completion of the “Monologue Reading” participant 1 was then asked to read “Dialogue Reading” (attached as part of “Instruments”) with a native American English speaker, who was preselected and was the same native speaker used for all participants. No participants knew the native speaker prior to the study’s data collection. This native speaker was instructed to read “Dialogue Reading” with every survey participant, as “Dialogue Reading” is a dialogue with two speakers. Participant 1 was then asked to read his or her lines in “Dialogue Reading” (either SA or SB, where S=Speaker) silently to him or herself and then to read it into the microphone with the native speaker. Once the participant and the native speaker finished reading “Dialogue Reading” together, participant 2 was asked to come back into the office. Both survey participants then read “Dialogue Reading” together, where participant 1 read the speaker (SA or SB) he or she did not read before with the native speaker. Both participants were asked to read their lines silently to themselves. When both were ready, they were asked to read “Dialogue Reading” into the microphone together. Upon completion of both participants’ reading, participant 1 completed his or her participation in the study and was asked to leave the office, closing the door behind him or her. Participant 2 was then asked to say the phrase “to this thing I am speaking” into the microphone in order to set an appropriate recording volume. Once the recording volume was set, participant 2 was instructed on how to appropriately begin his or her recording, asked to read the lines silently in his or her head, and asked to read “Monologue Reading.” After completion, Participant 2 was then asked to read the opposite Speaker’s lines in “Dialogue Reading” that he or she just read with participant 1. Once participant 2 was ready, he or she was asked to read “Dialogue Reading” with the native speaker.

Because this study is using three different stylistic environments, it’s important to understand their level on the formality scale. The Monologue Reading is meant to be read as a more formal reading not just because it’s a reading alone, but because the participants aren’t reading a narrative, they’re reading a poem, which is so unlike casual conversation that for this study, it is analyzed as having the highest level of formality. The dialogue reading between the native and non-native speakers have differing levels of formality as well. As the most casual form of speech is most likely to occur in contexts in which the speaker feels comfortable, relaxed, and less conscious of his or her speech, the dialogue with the native English speaker is more likely to be less relaxed, as this reading was the first time any of the participants met the native English speaker (Labov, 1966). It is important to understand the descriptive formality differences between the study’s instruments (realms of speech), as the main goal of this study is to associate interdental fricative substitutions with SAT, and convergence and divergence of speech will rely heavily on how formal the situation or context is.


            The participants chosen for the study were ten second language speakers of English from five different language backgrounds (as well as the same country). One male and one female were chosen from each of the five native language backgrounds. The status of the relationship between the male and female participants of the same native language background could not be one of matrimony, however, it was important for the study that participants of the same native language background know each other before joining in the study. This was essential to establish the atmosphere of the joint dialogue reading, “Dialogue Reading” (Appendix A) between the two participants as a more casual reading. This social atmosphere is contrasted with the social atmosphere of the joint reading between each participant and the native speaker, as well as the Monologue Reading alone.

The first of the participants were a male and female Arabic speaker. The female was 29 years old and began learning English at the age of 9. She had been in the United States for 5 years at the time of data collection. The male was 22 and began learning English at 13 years old. He had been in the United States for 8 months. Both Arabic speaking participants were from Saudi Arabia.

The second pair of participants was a male and female from China whose native language is Cantonese. The male was 22 years old and began learning English at the age of 3. He had been in the United States for eight months, but spent a semester in the United States the year prior to his arrival eight months ago at the time of the data collection. The female was 20 and began learning English at the age of 6. She had been in the United States for 6 months.

The third pair of participants was from the Democratic Republic of Congo and they speak French natively. The female was 32 years old and began learning English at 13 years old. She had been in the United States for two years. The male participant was 41. He speaks Lingala natively as well. He began learning English at 15 years of age and had been in the United States for 1 year and 2 months at the time of the data collection. These speakers, and perhaps others, are surely multilingual, speaking not just French and Lingala but also Swahili and possibly other languages.

The fourth pair of participants was from Brazil, and both speak Brazilian Portuguese natively. The female was 33 years old and began learning English at nine years old. She had been in the United States for 2 years. The male was 28 years old and began learning English at the age of 28. He had been in the United States for five months.

The last pair of participants was from Vietnam and speaks Vietnamese natively. The female was 32 years old and began learning English at 12 years of age. She had been in the United States for 2 years. The male was 20 years old and began learning English when he was 10. He had also been in the United States for 2 years at the time of data collection.

The participants will be called by a generic participant name: native.language1/native.language2. Each female will be the first participant and each male, the second. So, that the Arabic female participant is Arabic 1, the Arabic male participant is Arabic2, the Cantonese female participant is Cantonese1 and the male Cantonese participant is Cantonese2, and so on.


            In order to test whether second language speakers tend to use different replacement sounds or more accurate articulation of the interdental fricatives in different social contexts, it was necessary for them to complete recordings in different mock social contexts: by themselves, with another native speaker of the same native language background, and with a native English speaker.

The data attempts to make all participants produce both the voiced and voiceless interdental fricatives in all word positions in all social contexts. However, trying to place sensible words with the voiced interdental fricative in word final position proved to be rather difficult due to the sound’s rareness as a word final sound. Also, most English words voice the interdental fricative when it is a word medial sound, so finding words that contain a medial voiceless fricative was difficult as well, even more difficult than finding and using a word-final voiced fricative, so, there is one token of the word-final voiced fricative and none of the word-medial voiceless fricative. The words containing /θ/ extracted from the Monologue Reading sample are provided in (1).  We see in this list that word-initial and word-final are the only two environments elicited and that all the words on the list are content words rather than function words.

(1) Voiceless interdental fricatives in Monologue Reading

word-initial         word-final

think         earth

thrash         month

through         birth

thank         strength


The words from the Monologue Reading sample containing /ð/ are provided in (2).  Note here that while three environments are represented in this list, all of the word-initial environments are function words.  In contrast, the intervocalic and word-final items are content words.

(2) Voiced interdental fricatives in Monologue Reading

word-initial         intervocalic         word-final

the (x7)         bother         breathe

than (x2)         mother’s

then          brother’s

this (x3)         rather

In the Dialogue Reading sample, some of the same words appear as appeared in the Monologue Reading sample and a few are added.  The voiceless fricative again appears only in word-initial and word-final position, as shown in (3).

(3) Voiceless interdental fricatives in Dialogue Reading

Speaker A         Speaker B

word-initial         word-final         word-initial         word-final

thought         fourth (x2)         think (x3)         month

thinking         thrash         through         death (x2)

The voiced fricative does not appear in word-final position in the Dialogue Reading sample.  Again, the word-initial environment consists of function words while words in the intervocalic environment are content words. (4) shows this distribution.

(4) Voiced interdental fricatives in Dialogue Reading

Speaker A         Speaker B

word-initial         intervocalic         word-initial         intervocalic

that (x4)         brother         that (x4)         rather (x2)        

this         either         the (x2)                 

the         father         they’re        




As can be seen by the data above, the voiced interdental fricative is in much greater suppliance in the data due to the sound’s existence at the beginning of many function words. Because of this difference in number of the voiceless versus the voiced fricatives, the data analysis takes this into consideration, comparing and contrasting voiced and voiceless interdental fricatives using ratios instead of raw frequencies.



            The data analysis chapter involves looking at the data with a phonetic, phonological, and suprasegmental analysis. This chapter does not look at voiced fricative substitution differences between content vs function words because all function words have a word initial voiced fricative and all content words have the voiced fricative in medial or final word position, so an analysis of content vs function word replacement sounds would also depend on the factor of word place. Word place is taken into consideration in section 3.2. Throughout the data analysis section, the phrases “phonemic articulation” and “phonemically articulated” are used to describe all participants’ production of the English interdental fricatives. What exactly is meant by this phrasing is that it is assumed all participants were phonemically producing interdental fricatives, the manner and place of articulation/sounds native English speakers make, however, as pertains to most speakers of any language, what they thought they were producing was not always what they were actually producing. That is, their phonemic representation often does not match up with their phonetic output.

The data for this study was transcribed using two methods: subjectivity via listening and objectivity according to spectrographic analysis. Many of the replacement sounds are fairly easy to recognize in a spectrogram, such as the difference between a plosive and a fricative. Still other sounds are both difficult for speech perception and analysis using spectrographic data, namely, the labiodental fricatives [f v] versus the interdental fricatives [θ ð]. These sounds are usually the most difficult to perceive without context or visual cues. Acoustically, the labiodental and interdental fricative pairs are difficult to distinguish. Reetz and Jongman (2009) describe speech perception of English fricatives by citing Heinz and Stevens (1961) who found that synthesizing fricatives by varying the location of the spectral peak gave listeners different cues of fricative perception. Fricatives with a peak below 3 kHz were perceived as the postalveolar fricative /ʃ/ and fricatives with a peak between 4.5 and 6.5 kHz were identified as the alveolar fricative /s/. Any fricative that had a spectral peak above 6.5 kHz was perceived as a labiodental /f/ or an interdental fricative /θ/. Harris (1958) found that the sibilant fricatives /s, z, ʃ, ᴣ/ are perceived based on the fricative noise itself, while perception of /f, θ/ was largely based on the transitions into and out of the vowel. Harris concluded that listeners use the fricative noise to determine if the sound being produced is /s, ʃ/ or /f, θ/, and then use the vocalic transitions of the fricative to distinguish between /f/ and /θ/.  Spectrographic details of the labiodental and interdental fricative pairs reveal both to be so weak that they barely show up at all on the spectrograph. The labiodentals, /f v/, usually have their main noise around a frequency of 6 – 8 kHz (Rogers 2000). While interdentals’, /θ ð/, main noise exists at a frequency between 6 – 7 kHz. This is not much difference when trying to distinguish sounds via help of spectrographic data. Jongman (2000) finds additionally that phonetic cues for distinguishing the labiodentals from the interdentals are based on the transition information.

Also the differences between dental plosives, /ṯ ḏ/, and alveolar plosives, /t/, /d/ can be quite tricky. They are a little easier to distinguish perceptually than with help of a spectrogram. Dental plosives are often characteristic of some Native American English speakers’ speech when phonemically articulating interdental fricatives (Labov 1972). Many Native English speakers tend to turn dental fricatives into dental plosives, as dental plosives are a less marked, more natural way of producing meaningful, distinguishing sounds. Eckman et al. (2003) performed a study where they found that Portuguese second language speakers of English tend to under-differentiate the contrast between voiceless alveolar plosive /t/ and voiceless postalveolar affricate /tʃ/. Due to the speakers L1 phonology where [t] and [tʃ] are allophones of the same phoneme, [tʃ] occurring before high front vowels /i, ɪ/ and [t] occurring elsewhere. Novice Portuguese speakers of English will tend to produce a phonemic /t/ as [tʃ] before high vowels in English, e.g. “tip” /tɪp/ and “tear” /tiɹ/. In addition to this under-differentiation, Portuguese speakers will often replace voiceless interdental fricative /θ/ with [t]. However, when /θ/ appears before a high front vowel, like in the English word “theory” /θiɹij/, Portuguese speakers will tend to realize it as [t], even though phonemic /t/ is realized as [tʃ] before high front vowels by these speakers, according to Eckman et al. Therefore, “theory” would be expected to be realized as [tiɹij] and not [tʃiɹij]. By structuring their pronunciation this way, Portuguese speakers subconsciously prevent the neutralization of any contrast between the three English consonantal phonemes /t, tʃ, θ/. These phonemic mismatches can be a good indication of difficulties that L2 speakers face when trying to acquire an L2, but markedness, also explained as unnaturalness, plays a large role in how knowledge of an L1 affects the production or realization of an L2 (Yavas, 2006).

This study’s data analysis includes an analysis of the data as a whole, phonological processes governing interdental replacement, phonemic word position governing substitutions, replacement sound differences between speakers with different native language backgrounds, differences of replacement sounds by speakers with the same native language background, how these replacement sounds might change when speaking alone versus with someone else, and how the replacement sounds might change based on to whom the speaker is speaking and simultaneously listening to. These differences are correlated with social as well as linguistic experience variables in the discussion section following.

Throughout this section, you will notice the term “accurate articulation” (placing the tongue between the teeth and expelling air from the lungs to pass through the narrow openings between the teeth and tongue so that a frication noise can be heard). We will not call it right or correct articulation, as this terminology tends to see inaccurate articulation as wrong and incorrect, which are conceived as negative connotations. We cannot call a dental fricative’s inaccurate articulation wrong if his or her interlocutors can still understand him or her. Accurate articulation here is what is considered a standard General American pronunciation of the interdental fricatives. Any sound that is not technically an interdental fricative is considered inaccurate and not in-line with what is considered Standard American English, or better known as General American (GA), even though some members of GA typically stop the voiced fricative word-word initially, but there has to be a standard to measure from, and so this study measures from the fricatives’ phonemic articulation. All participants are assigned a rate of accurate articulation and rate of substitution. These ratios reveal how often a particular participant either accurately articulates each relevant fricative (the former ratio) and how often a particular participant substitutes a phonological process for an interdental fricative. The sum of these ratios for any speaker is 100%. The native English speaker used to perform the “Dialogue Reading” with all participants was asked to consciously and accurately produce the interdental fricatives. On a trial recording, the native speaker only produced accurately the interdental fricatives 72% of the time. The other 28% was a combination of dental plosives. However, for this data collection process, the native speaker accurately produced the interdental fricatives 100% of the time while performing the reading with the participants. This constant was necessary to find if participants were subconsciously converging or diverging their phonemic interdental fricative articulation.


            Little of the data has to do with phonological processes commanding the articulation of a phonemic interdental fricative, but it does occur some with a few of the participants and is definitely worth noting. The most common phonological process to change the articulation of phonemic dental fricatives is flapping word medial voiced interdental fricative /ð/, also discussed in sections 3.2 and 3.3. In these data, it only occurs in a word medial position; therefore a flap replacement in the data will only replace a voiced interdental fricative because the voiced interdental is the only one to occur word medially. It is not necessary that stopping will undergo flapping though. The stopping process creates a derived environment, so, participants that reduce a phonemic interdental fricative to an alveolar flap /ɾ/, are stopping the continuation of the fricative and then flapping it. Flapping is an important characteristic of General American, but is not a characteristic of British English/ Received Pronunciation (RP) (Rogers 2000). Native speakers of General American will reduce phonemically full-contacted alveolar plosives /t d/ to a mere flap of the tongue against the alveolar ridge when it lies on the border between a stressed and unstressed syllable.  It can occur in word medial position or sentence medial position, that is, while an alveolar plosive exists in a word final position, it still has a tendency to flap if the syllable before it is stressed and the syllable after it is unstressed. For example, let’s take a look at the phrase “I brought it.” Phonemically, it is transcribed /aj bɹɑt ɪt/, however in rapid speech, the phrase is phonetically transcribed as [aj bɹɑɾ ət]. The words “mother,” /mʌðəɹ/ “brother,” /bɹʌðəɹ/ and “rather” /ɹæðəɹ/ see the greatest frequency of an alveolar flap /ɾ/ replacing the articulation of the word medial voiced interdental fricative. This happens throughout the data, (5) gives a few examples of it. (*Note: MR = Monologue Reading, SA = Speaker A in the dialogue with the other same native language participant, SA/NS = Speaker A in the dialogue with the native speaker, SB = Speaker B in the dialogue with the same native language participant, and SB/NS = Speaker B in the dialogue with the native speaker)

(5) Flapping of interdental fricatives

Arabic2 SA/NS:            “brother” [bɹɑɾə]

Cantonese1 MR:             “bother” [bɑɾə]

French1 SB/NS:             “rather” [ɹɛɾə]

Portuguese1 SA:             “father” [fɑɾə]

Vietnamese2 SA:             “either” [iɾəɹ]

As can be seen in (5), flapping of the word medial voiced interdental fricative is characteristic of all language background participants, and is characteristic of each individual participant’s speech, though an alveolar flap /ɾ/ is not always the replacement sound for a word medial interdental fricative.

While Arabic1 articulates a flap in the word bother in her Monologue Reading, she accurately articulates a voiced fricative later in the same recording in the words “mother’s,” “brother’s,” and “rather.” Arabic2 flaps his phonemic voiced fricative in “mother’s” and “brother’s,” in his Monologue Reading, but produces a dental plosive in the word “rather” [ɹɑḏə] in the same recording.

In her recording with Cantonese2, Cantonese1 flaps the word medial fricative in “either” [iɾɪ], but later in the same recording the phonemic dental fricative is replaced by a voiceless alveolar plosive [ṯ] in the word father [fɹɑtə]. Cantonese2, in his monologue replaces the word medial fricative with a flap in the word “mother’s” [mɑɾəɹs], but accurately articulates the word medial phoneme later in the same recording in the word “rather” [ɹɑðəɹ].

French1 flaps the word medial fricative in her Monologue Reading of “bother” [bɹʌɾə], but later in the same dialogue, she replaces the fricative with a voiced alveolar plosive [d] in the word “brother’s” [bɹʌdʊz]. The only alveolar flap replacement we see in French2’s data is the flapping of the word initial dental fricative /ð/ in the utterance “hit the” [hiɾə], while all of his word medial voiced dental fricatives are accurately articulated.

Portuguese1 flaps consistently the word medial dental fricative in all of her recordings except for one instance where contact between the tongue and the alveolar ridge lasts just a little longer than a customary flap in the word “rather” [ɹædəɹ], where the length of the sound is 18.45 ms, while Portuguese1’s flaps’ duration is generally under 10 ms. Portuguese2 flaps “brother” [bɹʌɾɪ], but later in the same recording with the native speaker, he replaces the word medial fricative with a voiceless dental alveolar plosive /ṯ/ in the word “father” [fəṯə].

In her monologue, Vietnamese1 flaps “brother’s” [bɹʌɾəs], but earlier in the same recording, she had replaced the medial fricative with an alveolar nasal /n/ in the word “mother’s” [mʌnəz]. This particular substitution could be a result of progressive assimilation in that the nasality of the initial bilabial nasal /m/ is affecting the nasality of the chosen replacement sound. Vietnamese2 flaps “either” [iɾəɹ] in his dialogue with Vietnamese1, but later in the same dialogue replaces the medial fricative with a full-on alveolar plosive /d/ in the word “father” [fɑdɪ].

Other examples of phonological processes affecting phonemic interdental articulation replacement are postalveolar affricate articulation of voiceless interdental fricatives before phonemic articulation of retroflex liquid /ɹ/, like in the words “thrash” and “through.” Cantonese1 replaces the voiceless interdental with the voiceless postalveolar affricate /tʃ/ in the utterance “look through” [lʊk˺ tʃu] in her Monologue Reading. French1 makes the same replacement in “I thrash” [aj tʃɹʌʃ] in her dialogue reading with the native speaker. Vietnamese1 replaces with the postalveolar affricate in the same phrase as French1 “I thrash (in)” [a tʃɹɛsən]. English sequences like /tɹ/ and /dɹ/ are often realized as the English postalveolar affricates /tʃ/ and /dᴣ/. This palatalization process becomes quite varied in many individuals’ speech, however according to Rogers (2000) it seems to be most commonly realized before high front vowels /i, ɪ/. This phonological process, like flapping above, is where stopping creates a derived environment. The affrication process cannot occur before any stopping process is governing the interdental fricatives.

Cantonese2 and Vietnamese1 replaced the voiced interdental fricative with an alveolar nasal /n/ in the utterance “on the,” so that their phonetic realization is [ɑ˜ nə].

Lastly, a phonological process that only occurs with the voiceless interdental fricative /θ/ is deletion in word final position. In many words in the data, “strength,” “with,” “death,” and “month,” the fricative is deleted or lost. Loss of an interdental fricative again only happens to the voiceless fricative and only in word final position. We can attribute this loss to syllable or word final coda deletion.


            This section looks at the variable of word place for each fricative having an effect on the interdental replacements participants chose. It does not take into account social context, just the most common sound substitution participants had for each fricative in each of its different word places: initially, medially, and a single token of final for the voiced fricative, and initial and final place for the voiceless. However, this section does not look at each participant individually. It sums up what the most common pattern is across all data from all participants, though a few individual examples are given to demonstrate the common findings, as well as individual scores that disagree with what is most common among all participants. This section notes any change in replacement sounds based on these particular word places for the fricatives.

All function words containing the voiced fricative contain it word initially except ‘rather’ which appears three times throughout the instruments: once in the Monologue Reading and twice during the Dialogue Reading by Speaker B. Hence, an analysis of function versus content words containing the voiced fricative will not yield different results. For the voiced fricative, the most accurately articulated word place is intervocalic. In most participants’ cases it is articulated nearly twice as accurately in intervocalic position than initial or final (though final word position for the voiced fricative relies on only one token: ‘breathe’ in the Monologue Reading). Participants’ accurate articulation of the voiced fricative that disagree with this commonality among participants are Portuguese1, Vietnamese1, and Vietnamese2. The only participant that accurately articulated the word final voiced fricative is Arabic1. All others devoiced the fricative, stopped the fricative, or labialized it.

Initial word position is the most accurately articulated word place for the voiceless fricative. Portuguese1 and Vietnamese2 are the only participants’ whose findings disagree. The most extreme difference of accurate articulation between word places for the voiceless fricative is Arabic1 (90% initial, 36% final), French1 (80% initial, 27% final), and Vietnamese1 (80% initial, 0% final). The voiced fricative doesn’t have quite as large of a gap between accurate articulations based on word place. However, the most accurately articulated word places for the fricatives lie with the word place that has the fewest number of tokens in the instruments. It may be that the lack of accuracy is related to the larger number of tokens (Wolfram, 1998).

Only one word place seems to choose a specific substitution characteristic of all participants. Word medial voiced fricative is most often realized as alveolar flap [ɾ]. See section 3.3 for further discussion on this substitution. The voiceless, however, doesn’t do anything similar to the voiced by all participants, but we do find that the voiceless fricative is most usually deleted in word final position. Other substitutions based on word place for each fricative vary according to each participant. The largest fluctuation of either fricative is the choice of dental or alveolar plosives. Some participants varied in this degree, having dental plosives more often word initially and alveolar plosives word finally, or vice versa, but even this varies per participant.

Accurate articulation of the fricatives based on word place suggests that word initial position for the voiceless and medial for the voiced does have effect on whether participants chose to accurately articulate the phonemic fricative or substitute it with another segment. This suggests that voiced and voiceless fricatives may operate differently in the nonnative speaker’s phonological component. Of course, this conclusion is drawn only by looking at interdental fricative word place and cannot account for all marked sounds. Further research could take place to be able to prove such a conjecture as true, taking into account not only marked sounds, but all particular L2 sounds unfamiliar to particular non-native speakers.


            Cumulatively throughout the data, the voiced interdental fricative /ð/ was most often replaced by another sound, while the voiceless /θ/ was more often accurately articulated. Looking at all the recordings (Monologue Reading, Dialogue Reading A and B), Cantonese2 has the largest difference between accurate pronunciation of the voiced and voiceless interdental fricatives. He accurately produces the voiceless /θ/ 80% – 100% of the time, while his accurate articulation of the voiced fricative /ð/ falls between 34% – 73% accurate articulation. We have a range here because the articulation accuracy of the interdental fricatives was different throughout all realms of speech for all participants. Each realm of speech was given a ratio of accurate articulation of the fricatives. Therefore, 80%, in his recording with the native speaker, was the lowest ratio of voiceless interdental fricative accurate articulation in all of Cantonese2’s realms of speech, while 100%, in Cantonese2’s recording alone, was the highest.

The only participant to have a higher articulation accuracy of the voiced over the voiceless interdental fricative is French2. In every single one of his recordings, his accurate articulation of the voiced interdental was greater than his accurate articulation of the voiceless, so much that the voiced was nearly twice as accurately articulated than the voiceless. French2’s accurate articulation of the voiced fricative ranged from 36% – 56%, while his accurate articulation of the voiceless varied from 0-14% in all his recordings. Other participants had more accurate articulation of the voiced over the voiceless, but not in every single recording. Arabic1 did so in her dialogue with Arabic2, Portuguese1 did in her dialogue with Portuguese2, and Vietnamese2 also had a higher accuracy of articulation of the voiced over the voiceless in both his dialogue with Vietnamese1 and the native speaker. In all other recordings, the voiced interdental fricative /ð/ was replaced more often than the voiceless /θ/. For accurate articulation of each participant in each social realm and overall, take each ratio given in the charts in section 4.3 of Chapter 4, and subtract them from 1 (or 100%). Percentages of particular substitutions are calculated by using the total number of possible tokens for substitution.

Something we will not look at is comparison of accurate interdental fricative articulation between participants of the same native language in this section. Look to section 4.2 of chapter 4, using demographic information to analyze these accurate articulations. We turn, instead, to the replacement sounds that each participant makes, whether it agrees or disagrees with the same native language participant’s replacement sounds and how these substitutions might be different or similar to other native language background participants’ replacement sounds. The most common replacement sounds found throughout the data are dental /ṯ ḏ/ and alveolar /t d/ plosives. In most circumstances, the voiced plosives /ḏ/ and /d/ replaced the voiced fricative /ð/ and the voiceless plosives /ṯ/ and /t/ replaced the voiceless /θ/. The voiced dental /ḏ/ and alveolar /d/ plosives are the most articulated replacement sound of the phonemic voiced interdental fricatives /ð/. The dental and alveolar plosives are the only voiced/voiceless pairs of replacement sounds to appear in every participant’s speech. We can analyze these as one sound substitution process (stopping) due to their extreme closeness in proximity of place in the mouth, that it can often be difficult for listeners to perceive a coronal plosive as dental or alveolar. So, for analysis of similar or dissimilar replacement sounds, perception could lump these into one sound, so we will do so too, though for each participant pair, replacement usage distinctions between these two pairs of four plosives are made.

In addition to some other substitutions, both Arabic participants used these plosives as replacement sounds often, more often than other participant pairs. Arabic1 replacing the interdentals with all four plosives, while Arabic2 replaced using only three of the plosives. See (6).

(6) Arabic plosive substitution

Arabic1 SA:                         “well this (is)” [wɛl ɪsəs]

Arabic1 SB/NS:             “dream that” [dɾim dɛt]

Arabic1 MR:                         “month of” [mɑn ɑf]

Arabic1 MR:                        “with hope” [wɪt hop]

Arabic2 MR:                        “rather” [ɹɑə]

Arabic2 SB:                        “dream that” [dɾim dɛt]

Arabic2 SA/NS:            “fourth place” [fot plejs]

The Arabic participants articulate the dental and alveolar plosives in every realm of speech that the study is examining for the voiceless fricative. As, we will see, every participant produces plosives for the voiced fricative in every realm of speech. The voiced dentals appear more often throughout the Arabic participants’ speech than the voiceless dentals, but the voiceless alveolar plosive /t/ appears more than the dental version. In her speech alone, Arabic1 produced these dental and alveolar plosives as replacement sounds 90% of the time, while Arabic2 replaced using them 64% of the time. In their dialogue together, Arabic1 uses them as replacements 33% of the time, while Arabic2 used them as replacement sounds 100% of the time during their dialogue together.

Both Cantonese participants replaced using all four plosives in all of the study’s realms of speech. Notice, however, Cantonese1 replaces the voiced fricative /ð/ with the voiceless dental plosive /t/. See (7).

(7) Cantonese plosive substitution

Cantonese1 MR:            “of this” [əf is]

Cantonese1 SA:            “like that” [laj das]

Cantonese1 MR:            “river the” [wivəɹ ə]

Cantonese1 SB/NS:            “I thrash” [ɑ tɹʌs]

Cantonese2 MR:            “then this” [ɛn is]

Cantonese2 SA/NS:            “dream that (I)” [dɹim dæɾɑ]

Cantonese2 SA/NS:            “kept thinking” [kæpt eŋkɪŋ]

Cantonese2 SB:            “month” [mənt]

Like the Arabic participants, both Cantonese participants replace using the dental pair /ṯ ḏ/ more than the alveolar pair /t d/. However, in her reading with the native speaker, Cantonese1 uses the alveolar plosive over the dental 6 times more. Cantonese1 uses these plosives as replacement sounds 66% of the time during her Monologue Reading, while Cantonese2 had them 73% of the time during his. During their dialogue together, Cantonese1 was at about 53% replacement with the plosives and Cantonese2 was at 66%.

Both French participants replaced using the four plosives, French2 replacing with all four, but his “replacee” of the voiceless dental plosive [ṯ] is the voiced fricative /ð/. French1 is missing the voiceless dental plosive [ṯ] and her replacement with the voiceless alveolar plosive /t/ could be influenced by the same sound following the phonemic articulation of the voiceless interdental fricative /θ/. See (8).

(8) French plosive substitution

French1 MR:                        “on the (pulse)” [ɑn ɛpɚs]

French1 SA:                        “well this (is)” [wɛl dɪsəs]

French1 MR:                        “strength to” [stɹeŋtu] /stɹɛŋθ tu/

French2 SA/NS:            “dream that” [tʃɹim æt]

French2 SB:                        “they’re” [dejə]

French2 MR:                        “than this” [ðɛn is]

French2 SB:                        “think I thrash” [θiŋgaj tɹɛʃ]

Both French participants see these plosives in all of the study’s realms of speech. For all of the study’s realms of speech, French1 realized the alveolar plosive more than the dental, but French2 had a mix of usages of these plosives. 78% of French1’s replacement sounds in her Monologue Reading are these plosives, while French2’s replacement usage lies around 44%. During their dialogue together, French1’s use is 71% and French2’s rises to 57%.

Both Portuguese participants used these plosives to articulate their phonemic interdental fricatives too, and both Portuguese participants used the voiceless dental plosive [ṯ] to articulate the voiced fricative /ð/. See (9).

(9) Portuguese plosive substitution

Portuguese1 SA:            “dream that (I)” [drɪm ɛɾɑj]

Portuguese1 SB/NS:            “rather” [rædəɹ]

Portuguese1 MR:            “have the” [ʔæv ə]

Portuguese1 SB/NS:            “I thrash (in)” [ə tɹɛʃṇ]

Portuguese2 SB:            “think that” [fɪŋ ɛts]

Portuguese2 MR:            “have the” [hɛf də]

Portuguese2 SB:            “though” [oz]

Portuguese2 SA/NS:            “keep thinking through” [kip tɪŋ tʊf ]

Like most other survey participants, the Portuguese participants saw these plosives in all realms of speech, Portuguese1 68% of the time and Portuguese2 68% as well in their Monologue Readings. In their dialogue together, 46% usage of the plosives went to Portuguese1 and Portuguese2 dropped as well, but to 58% usage. In all realms of speech Portuguese1 uses the dental plosives more frequently than the alveolar, but Portuguese2 uses alveolar [d] more in is reading alone and with the native speaker, but dental /ḏ/ more in his dialogue with Portuguese1.

Vietnamese1 replaced using the voiceless alveolar plosive /t/ only once and when doing so, she added aspiration. Her only replacement with the voiceless dental plosive was in place of a phonemic voiced interdental fricative /ð/. Her “replacee” of the voiceless dental plosive /ṯ/ was the voiced fricative /ð/. Vietnamese2 did not replace using the voiceless dental plosive, but did add aspiration to two of his voiceless alveolar plosives in his reading with Vietnamese1. See (10).

(10) Vietnamese plosive substitution

Vietnamese1 SB:            “they’re” [ɛə]

Vietnamese1 MR:            “have the” [hæf  də]

Vietnamese1 MR:            “breathe up” [bɹi ʌp]

Vietnamese1 SA/NS:            “I thought” [a ɑ]

Vietnamese2 MR:            “have the” [hev˺ ə]

Vietnamese2 SB/NS:            “think that (show)” [a tiŋ dɑʃo]

Vietnamese2 MR:            “kept thinking” [kæp iŋkɪŋ]

Vietnamese2 SA:            “fourth” [fɑ]

We see Vietnamese1 using these plosives as a phonemic interdental replacement sound 58% of the time, while Vietnamese2 is a bit higher with 68%. In their dialogue together, Vietnamese1’s usage drops to 42%, while Vietnamese2’s usage also drops, but only slightly to 63% usage of these 4 plosive sounds.

The next most common replacement sounds are the labiodental fricatives /f v/. The voiced labiodental fricative is most commonly used to replace the voiced. In the case of these pairs of fricatives, it is the voiceless sound /f/ that is used more often to replace the voiceless interdental fricative, but the voiced /v/ replaces the voiced sound a lot less often. Unlike the pairs of plosives above, the voiceless version /f/ is used nearly 100% of the time by some participants, while the voiced /v/ is only used twice by two participants: once by Cantonese1 in her dialogue with the native speaker. She replaces a phonemic voiceless interdental /θ/ in “I think” [jɵ vɪŋ]. Also, it’s produced by French2 in his Monologue Reading alone in the word medial position “brother’s” [bɾɑvəs]. French2 and Cantonese1 are also the only participants to replace the voiced interdental fricative /ð/ with the voiceless labiodental fricative /f/, as all other participants using this fricative as a substitution replaced in a phonemically articulated voiceless interdental fricative place.

The only pair of participants to not use labiodental sounds in fulfillment of phonemic interdental sounds are the Arabic speaking participants, which is odd, as in Arabic the voiceless labiodental fricative /f/ is in the Arabic language phonemic inventory (Yavas, 2006).

            Of the two Cantonese speakers, only Cantonese1 uses the labiodental fricatives as a replacement sound. Cantonese2 only uses the plosives mentioned before, the flap for phonemic voiced fricative /ð/ in word medial position, and opposite voicing interdentals for his replacement sounds. Cantonese1’s voiceless labiodental usage is 27% for her Monologue Reading, 23% for her dialogue with Cantonese2, and 17% for her dialogue with the native speaker.

Both French participants use the voiceless labiodental fricative in all realms of speech, but French1 uses it as replacement sound much less than French2. In their Monologue Readings, French1 uses the voiceless labiodental /f/ as a replacement sound 11% of the time and French2 uses it 44% of the time. In their dialogue together, French1 usage jumps 28% and French2’s jumps to 56%. In their dialogues with the native speakers, French1’s labiodental usage lowered to 8%, while French2’s usage levels out at 55%.

The Portuguese participants’ labiodental fricative replacement use is of rather similar proportion in all of the study’s realms of speech. Portuguese1’s usage is about 15% of her replacement sounds and Portuguese2’s usage is around 19%. In their dialogue together, Portuguese1’s usage jumps to 50% and Portuguese2’s usage jumps to 35%. In their dialogues with the native speakers, both participants’ usages are at the lowest of any of the study’s realms of speech, 10% for Portuguese1 and 0% usage for Portuguese2.

Only Vietnamese1 uses the voiceless labiodental /f/ as a replacement sound. She uses it as a single replacement per her monologue recording and her recording with the native speaker, but in her dialogue with Vietnamese2, she has two replacement sounds with a labiodental fricative, while Vietnamese2’s usage is 0% throughout all his recordings. (11) illustrates Vietnamese1’s labiodental replacements.

(11) Vietnamese1 Labiodental Replacement Sounds

Vietnamese1 MR:            “earth” [əɹf]

Vietnamese1 SB:            “death at” [dæf ɛd]

Vietnamese1 SB:            “death” [dæf]

Vietnamese1 SA/NS            “forth place” [fofples]

Notice the only time Vietnamese1 replaces the voiceless interdental fricative with the labiodental is in word final position. She replaces all word final voiceless interdental fricatives with a labiodental. Vietnamese1 leaves no other sound in place of phonemic word-final voiceless interdental fricative. In all other cases throughout her recordings where labiodental does not replace a word final interdental, Vietnamese1 simply deletes the sound altogether. (12) shows these deletions.

(12) Vietnamese1 Deletions

Vietnamese1 MR:             “month of” [mɑwɑwə]

Vietnamese1 MR:            “birth”            [bəɹ]

Vietnamese1 MR:            “with hope” [wɪ hɑp˺]

Vietnamese1 SB:             “month” [mʌn]

Glottal stop /ʔ/ is the next most common replacement sound among survey participants. It is a replacement sound that is characteristic among all of the study’s language backgrounds except Arabic. As, the glottal plosive takes the place of both fricatives throughout the data, as did [f] and [v], both Arabic participants not articulating these two “pairs” of substitutions is peculiar indeed. Either the [f v] and [ʔ] are somehow a related substitution for the Arabic speakers or the Arabic speakers’ phonological component simply favors dental and alveolar stopping above all other possible substitutions. (13) gives some of these replacement instances.

(13) Glottal Stop Replacement

Cantonese1 SB/NS:            “rather” [ɹæʔə]

French1 SB/NS:            “death at (least)” [dɛʔ æɾlis]

Portuguese1 SB/NS:            “think that (show)” [θʔæʃo]

Portuguese2 SA/NS:            “fourth place” [foɹɪʔ plɛs]

Vietnamese2 SA:             “fourth place” [ɤʔ ples]

Vietnamese1 SB:            “rather” [ɹæʔəɹ]

Vietnamese1 SA/NS:             “father” [fɑʔɪ]

Other replacement sounds of participants are rare and only happen in cases of one participant’s speech in one recording. Cantonese1 replaces a voiced interdental fricative with a voiceless alveolar fricative /s/ in her dialogue with the native speaker,  “think that” [tiŋk sæʔ]. Portuguese2 replaces a voiceless interdental fricative with a voiced postalveolar affricate /dᴣ/ in his Monologue Recording, “strength to” [stɹendᴣ tʃu]. Vietnamese1 replaces a voiceless interdental with a voiceless glottal fricative /h/ in her reading with Vietnamese2, “I think” [a hiŋ]. Vietnamese2 has both a voiceless alveolar affricate /ts/ and a voiceless velar plosive /k/ as replacement sounds in his monologue recording, “earth” [əɹst] and “to think” [kɛŋk], though this last velar substitution could be the result of regressive assimilation from the same sound being the phonemic word final sound.










            The previous Data Analysis section analyzed the data phonetically, that is to say it literally described the phonetic sounds the study’s participants chose in replacement of the phonemic interdental fricatives. This Discussion chapter is more speculative in its analysis of the data. The Discussion attempts to correlate some of the replacement sounds per each speaker with social variables of convergence and divergence (Giles et al., 1991) and correlates similarities/differences of replacement sounds per each native language pair with social variables as well as demographic information from section 2.1.

SAT examines linguistic variation in different social contexts, and according to Giles et al. (1991), speakers can and will often either converge or diverge (become more similar or different) in speech styles with their interlocutors. The theory was devised to explain motivations that underlie speech style choice for various native and nonnative speakers during different social encounters and the possible consequences that may arise. According to this theory, convergence of speech styles with one’s interlocutor(s) is usually given a positive evaluation by the listener, while divergence is given a negative evaluation. Speakers who converge their speech styles (make one’s speech more closely match an interlocutor’s) are considered to want to close the social gap between one another, while divergence is the opposite. A diverging speaker is considered to wish to increase his or her social distance from his or her interlocutor by emphasizing their distinctiveness linguistically. This convergence or divergence in speech styles can occur at any or all dimensions of language, from certain accent or dialect feature choices to other vocal characteristics like volume, rate of speech, etc. This study’s primary examination of interdental fricative substitutions would fall into the former description of speech style choice.

But first, let’s look at phonetically why the replacement sounds surfaced as they did. The replacement sounds examined in section 3.1 account for reasons as to why some of the substitutions surfaced as they did, but phonology can’t account for all the replacement sounds.


            Below, tables 4.1.1- 4.1.3 show voiceless fricative substitution sounds for all participants in each of the study’s realms of speech. Tables 4.1.4- 4.1.6 show voiced fricative replacement sounds for all participants in each of the study’s realms of speech.

(Note: marked symbols with diacritics like aspirated voiceless alveolar plosive [tʰ] and unreleased voiced alveolar plosive [d̚ ] are added to the prototypes voiceless alveolar plosive [t] and voiced alveolar plosive [d], respectively.

The most common replacement of the voiceless fricative in the monologue  appears to be voiceless labiodental fricative [f], and deletion seems to be the second most common.

Table 4.1.1 Voiceless replacement sounds in Monologue Reading

  f t ð d dᴣ st k Total
A1 2 9
A2 2 9
C1 4 1 1 9
C2 9
F1 2 1 9
F2 7 9
P1 3 1 1 1 9
P2 4 1 1 1 9
V1 1 4 9
V2 2 2 1 1 9
Total 21 8 4 2 1 1 1 1 1 1 1 90

Like the monologue replacements of the voiceless fricative above, table 4.1.2 shows labiodental [f] being the most common substitution of the voiceless in the dialogue with the speaker of the same native language, however, no other replacement or deletion tends to be common except alveolar plosive [t], produced by one less person than deletion in the Monologue Reading.

Table 4.1.2 Voiceless replacement sounds in dialogue with non-native speaker

  f t ð ɾ ʔ h Total
A1 1 1 1 5
A2 7
C1 3 1 5
C2 1 7
F1 2 5
F2 5 1 7
P1 5 5
P2 6 7
V1 2 1 1 1 7
V2 2 1 5
Total 23 5 2 1 1 1 1 1 60


The voiceless alveolar plosive [t] appears to be the most common replacement sound of the voiceless fricative in the dialogue with the native English speaker, which is both a different place and manner of articulation than the most common replacement sound of the voiceless fricative in the study’s other realms of speech. It was also produced by more participants than [f]. However, the labiodental fricative [f] appears to be the second most common replacement sound in the dialogue with the native speaker. Table 4.1.3 shows these replacements.

Table 4.1.3 Voiceless replacement sounds in dialogue with native speaker

t f ʔ v Total
A1 1 1 7
A2 1 5
C1 2 3 1 7
C2 1 5
F1 1 1 1 7
F2 5 5
P1 1 1 7
P2 4 1 5
V1 1 1 5
V2 6 7
Total 16 11 2 1 1 1 1 60

            Table 4.1.4 shows voiced fricative replacements in the Monologue Readings. Both voiced dental and voiced alveolar plosives are the most common replacement sounds of the voiced fricative in the participants’ monologues. All participants replace using the dental plosive three or more times, but the dental plosive is not a replacement sound of Portuguese1 The alveolar plosive, however, occurs in lower frequencies, as some participants only use the alveolar as replacement one or two times, but the alveolar also has the larger quantity of being used as a replacement sound by a single speaker. French1 and Portuguese2 both replace using the alveolar plosive ten times. Alveolar flap [ɾ] is a common replacement sound of most participants, but definitely occurs in lower quantities compared to the dental and alveolar plosives. This is most likely due to the fact the participants most usually used the flap to replace a word medial voiced interdental fricative, and since words containing the voiced fricative medially were not in great supply, the sound used to replace this sound in medial position is not nearly as common as the plosives.

Table 4.1.4 Voiced replacement sounds in Monologue Reading

  d ɾ θ n t f v Total
A1 4 2 1 1 18
A2 8 1 3 1 18
C1 8 4 3 1 1 1 18
C2 7 1 1 1 1 18
F1 4 10 1 1 18
F2 3 1 2 1 1 2 1 18
P1 11 3 1 1 18
P2 3 10 2 18
V1 1 8 3 1 3 18
V2 4 8 3 1 1 18
Total 53 45 18 7 5 5 3 3 2 180

In the dialogue with the participant of the same language background, alveolar and dental plosives still seem to be the most common substitution. Table 4.1.5 illustrates this.

Table 4.1.5 Voiced replacement sounds in dialogue with non-native speaker

  d ɾ t θ ʔ n Total
A1 2 1 9
A2 1 3 11
C1 3 3 2 1 9
C2 3 1 11
F1 5 9
F2 5 2 1 11
P1 3 3 2 9
P2 2 5 1 2 2 11
V1 3 3 1 2 1 11
V2 5 3 9
Total 27 24 10 3 2 2 2 1 100

Table 4.6 shows that again the most common substitution sounds for the voiced fricative are the alveolar and dental plosives for the dialogue with the native speaker. We see also that word medial flapping is less common in the dialogue with the native speaker, but this could also be do to the lack of word medial interdental fricatives in the Dialogue Reading. Table 4.1.6 illustrates these findings.

Table 4.1.6 Voiced replacement sounds in dialogue with native speaker

  d ɾ ʔ n θ t s Total
A1 2 5 11
A2 2 1 9
C1 6 1 1 1 1 1 11
C2 1 5 9
F1 8 1 1 11
F2 2 2 9
P1 1 5 1 1 11
P2 4 2 1 1 1 9
V1 3 1 1 2 1 9
V2 5 1 3 11
Total 32 23 9 3 2 2 2 1 1 100

The most common replacement sounds of all realms of speech of all the participants were either dental or alveolar plosives (stopping) for the voiced interdental fricative. The most common replacement sound for the voiceless interdental fricative was a labiodental fricative /f/ in the monologue and the dialogue with the speaker of the same L1, or in phonological processes terms, labialization was the most common process changing the voiceless fricative. However, the most common replacement sound for the dialogue with the native English speaker was [t], followed closely by [f].


            Looking at each participant’s demographic information from section 2.1, summarized in Table 4.2.1, we do not see an extreme difference in predicted accurate articulation. Because some participants have less ideal circumstances (late L2 onset, short time spent in an English speaking country) in their acquisition of English than others, we may be able to predict which participants will have a lower accurate articulation of the fricatives.

Included in Table 4.2.1 is each participant’s demographic information as provided by them on the Demographic Questionnaire. Also included is Interdental fricative Articulation Accuracy (I.A.A.) rate. Again, there is a range on these accurate articulation ratios because there are three social realms in the study. The lower ratio/percentage is the lowest accurate articulation of the relevant fricative in all of the study’s social realms for that participant and the higher ratio/percentage is the highest accurate articulation of the relevant fricative in all of the study’s realms of speech. The I.A.A. ratio/percentage of the third realm of speech lies somewhere in between the lower and higher percentages given for each fricative (except Portuguese1 whose lowest accurate articulation was 11% in both her Monologue Reading and her dialogue with Portuguese2 for the voiced fricative, and Vietnamese1 whose highest accurate articulation was oddly also 11% for both her monologue and her dialogue with the native speaker).

Table 4.2.1 Demographic information


Age at Data Collection

Age of English Onset

Time Spent in the US

I. A. A.




5 years








8 months








6 months








8 months








2 years








1 year,

2 months








2 years





Table 4.2.1 (Continued)




5 months








2 years








2 years





Taking into account the Critical Period Hypothesis (Lenneberg 1967), we can predict that study participants who have their English onset as after the proposed age of the period’s end, will be most likely to have lower accurate articulation of the fricatives. Almost half of the study’s participants began learning English before the age 12, while the others’ age of English onset were just a few years older than 12, except Portuguese2, whose age of onset was 28. He began learning English when he moved to the United States to complete his education. His English acquisition began only 5 months ago at the time of data collection. We can postulate that because of both his late acquisition and nearness to his L2 onset, these factors explain why his accurate articulation of the fricatives is so low, never getting above 20% accurate articulation of either fricative in all of the study’s realms of speech. Vietnamese2 also never had an accurate articulation of either fricative above 20%. His English onset began at age 10 and he’s 20 years old. He has been speaking English the shortest amount of time of all participants, aside from Portuguese2. This might account for his low accurate articulation, but fossilization (Han 2004) could definitely be a factor in Vietnamese2’s speech.  After speaking English for nearly ten years and after living in the US for two years, one might find it hard to believe that Vietnamese2 had such a low accurate articulation of the fricatives, but despite his learning advantages, Vietnamese2’s English phonology is most likely fossilized, of course, we can’t say this with confidence, as it would require carrying out a full study to find if Vietnamese2 or any of the participants have reached a fossilized state in their English phonology acquisition. For examples of such studies, refer to Lardiere (1998).

Cantonese1 never accurately articulated the voiced fricative in any of her readings, and she correctly articulated the voiceless fricative 14-25% of the time. Her English onset began relatively young, at the age of six and she is now 20, having lived in the US for 6 months. No other explanations can really explain her low accurate articulation except Han (2004)’s definition of fossilization. All other participant’s accurate articulation of the fricatives was near the same ranges except Cantonese2 and Arabic2’s accurate articulation, who had similar accurate articulations of both fricatives. Cantonese2’s unusually high accurate articulation could be attributed to the fact that he began learning English the earliest of all the participants, at the age of 3. Arabic2 started learning English much later than Cantonese2, after Lenneberg (1967)’s proposed age of 12. Arabic2’s high accurate articulation of both the fricatives can be attributed to many things, the quality of his instruction input, the amount of speaking English versus Arabic he does, as well as can be attributed to many personal variables like motivation and linguistic ability in general.


            This section of the Discussion looks at sound substitutions of each participant. Here we are looking at quantity of substitution sounds and their ratios (# of a certain substitution/ total number of phonemic interdental fricatives). Due to varied substitutions by each participant, this analysis looks most closely at which sounds each participant most commonly chose as substitutions in each of the three recordings and not sounds that they used as replacements rarely, which is why the sounds explained to have phonological constraints governing them in section 3.1 are included in this section’ analysis. (*Note: Accurate articulation rate + substitution rate = 100%)


            The Arabic speakers exhibit more plosive substitutions than any other substitution for both fricatives. This, as we will find out, is not parallel to substitutions provided by the other participants. Dental stopping was more common than alveolar stopping for both Arabic participants. The participants have near identical substitution rate overall for the voiced fricative (55 and 50%), Arabic2’s substitution rate (15%) for the voiceless was much lower than Arabic1’s (40%), as Arabic1 exhibits 5 more instances of substitution for the voiceless fricative than Arabic2.

Tables Arabic1


  t ð ɾ # %
MR 1 2 3/9 33%
SA/L1 1 1 1 3/5 60%
SB/NS 1 1 2/7 29%
Total 3 2 1 1 1 8/20 40%



  d ɾ # %
MR 4 2 1 1 8/18 44%
SA/L1 2 2 4/9 44%
SB/NS 5 2 7/11 63%
Total 11 4 3 1 19/38 50%


Arabic1’s voiceless fricative substitutions are not in great amounts in any of her recordings. We find she replaces using a voiceless alveolar plosive [t] the same number of times in all her recordings, while the dental plosive [ṯ] is used as a replacement sound twice in her reading alone and not at all in her dialogues with Arabic2 and the native speaker. She has the most accurate rate of articulation of the voiceless in her dialogue with the native speaker, however her rate of substitution for the voiceless is only 9% less in her dialogue with the native speaker than her recording alone where she supplies only one more substitution for the voiceless fricative. We can hardly call this higher accurate articulation with the native speaker convergence. Arabic1’s highest substitution rate for the voiceless fricative is in her dialogue with Arabic2.

Arabic1’s voiced fricative replacements are greater in number than her voiceless, but the rate of replacement for the voiced is almost as much as the voiceless. She replaces most often using dental plosive [ḏ] as she supplies it most frequently in all her recordings except her recording with Arabic2 where a flap [ɾ] was just as common as [ḏ]. Arabic1 replaces using an alveolar plosive [d] in all her realms of speech except her recording with Arabic2, who did provide 1 substitution with [d].

Tables Arabic2


  t # %
MR 2 2/9 22%
SB/L1 0/7 0%
SA/NS 1 1/5 20%
Total 2 1 3/20 15%




  d ɾ θ # %
MR 8 1 2 1 12/18 66%
SB/L1 3 1 4/11 36%
SA/NS 3 1 1 5/9 55%
Total 14 3 3 1 21/38 55%

Arabic2’s highest accurate articulation rate, contrary to Arabic1’s results is in his dialogue with Arabic1. His lowest accurate articulation of the voiceless fricative is in his speech alone. Deletion, unlike Arabic1’s alveolar and dental plosives, is the most common process affecting the voiceless fricative. Arabic2 doesn’t provide the voiceless fricative only three times, twice with deletion and replaces it ounce with an alveolar plosive [t].

Arabic2’s voiced substitutions are 7 times greater than the number of substitutions for his voiceless. He, like Arabic2, most commonly replaces using a dental plosive [ḏ], whose use is twice as much in his monologue recording than in his dialogue with Arabic1 and the native speaker. His suppliance of alveolar plosive [d] is completely steady throughout all of his realms of speech, using it as a replacement sound one time per recording. He also flaps the voiced fricative most often in his speech alone and not with Arabic1. Actually, like the voiceless fricative, Arabic2’s greatest accurate articulation rate is in his dialogue with Arabic1. This coincidence could be worth noting. Also, his lowest accurate articulation rate and therefore his highest substitution rate for the voiced and voiceless fricatives is in his speech alone. He only once devoices the voiced fricative in his speech alone, but doesn’t in his dialogues with Arabic1 or the native speaker.

Arabic2’s divergence from substitution sounds Arabic1 is making is a definite possibility, as his highest accurate articulation of both fricatives is in his dialogue with Arabic1. Arabic1, on the other hand, doesn’t seem to be using accommodation when choosing her interdental fricative replacements. Actually, her substitution sounds don’t seem to depend on social context at all.


            The Cantonese participants show drastically different replacement sounds and rates of accurate articulation. Cantonese1 has lower accurate articulation rates than Cantonese2 for both fricatives and has more varied substitution sounds. Like the Arabic participants, the Cantonese participants had higher substitution rates for the voiced fricative compared to the voiceless. Actually Cantonese1 doesn’t accurately articulate one single token of the phonemic voiced fricative, while Cantonese2 does so about half the time. Cantonese1 has a very high substitution rate for the voiceless fricative (80%) compared to Cantonese2’s unusually low substitution rate for the voiceless fricative (10%).

Tables Cantonese1


  f t v # %
MR 4 1 1 6/9 66%
SA/L1 3 1 4/5 80%
SB/NS 3 2 1 6/7 85%
Total 10 2 2 1 1 16/20 80%




  d ɾ t θ f s ʔ # %
MR 4 8 3 1 1 1 18/18 100%
SA/L1 3 3 2 1 9/9 100%
SB/ 6 1 1 1 1 1 11/11 100%
Total 13 12 6 2 1 1 1 1 1 38/38 100%


Cantonese1’s most common substitution sound for the voiceless fricative is labiodental [f]. She produces it the most frequently of any sound replacement for the voiceless in all realms of speech. The next most common replacement is alveolar plosive [t], only showing up in her speech with the native speaker, however in her dialogue with Cantonese2 and in her Monologue Reading, she produces a postalveolar affricate [tʃ], which is inexistent in her speech with the native speaker. This distribution of [t] and [tʃ] might be something to note because the difference between [t] and [tʃ] is place of articulation, even though they do differ in manner, but if we were to hypothesize that the frontness of the native speaker’s interdental sounds might have made Cantonese1 front her postalveolar affricate [tʃ] to alveolar [t], as no affricate lies at the alveolar place of articulation. This is rather far-fetched, though, but not out of the realm of possibility.

Cantonese1 had the most varied substitutions of all participants for the voiced fricative, most of them occurring only one time each. Her most common voiced replacement is alveolar plosive [d], used most in her dialogue with the native speaker and dental plosive [ḏ], used most in her recording alone. Each plosive was equal in number in her dialogue with Cantonese2. Her use of a flap as a substitution for the voiced fricative is nearly half as frequent as either of the plosives. Some of the other sound replacements for the voiced that only have one token in all of the realms of speech combined are due to phonological processes.

Tables Cantonese2


  t # %
MR 0/9 0%
SB/L1 1 1/7 14%
SA/NS 1 1/5 20%
Total 1 1 2/20 10%




  d ɾ θ n # %
NS 7 1 1 1 1 11/18 61%
SB/L1 3 3/11 27%
SA/NS 5 1 6/9 66%
Total 15 2 1 1 1 20/38 52%

Cantonese2’s voiceless fricative replacement rate is the lowest of all the participants. He only replaces the voiceless twice, once with a dental plosive [ṯ] in his dialogue with Cantonese2 and once with an alveolar plosive [t] in his dialogue with the native speaker.

Cantonese2’s most common voiced fricative replacement is by far dental plosive [ḏ]. It out numbers alveolar plosive [d] fifteen to two. Other sound replacements are infrequent occurring only once throughout all the realms of speech for Cantonese2.

Both Cantonese participants’ rates of substitution and rates of accurate articulation show no patterning in favor of accommodation. However, Cantonese1’s consistent accurate articulation (0%) for the voiced fricative and consistently low accurate articulation of the voiceless fricative could be an indication of low proficiency in English, or at least, lower than Cantonese2’s, but this is purely speculation because neither of the Cantonese participants’ demographic information foresee an experience difference in the participants. Though, Cantonese2 did begin his onset of English learning 3 years earlier in life than Cantonese1, both still have onset ages below Lenneberg’s proposed age of 12.


            The French speakers’ substitutions are more harmonious than the Cantonese speakers, where their overall accurate articulation rates/ substitution rates are not quite so. French1’s voiceless interdental substitution rate (40%) was more similar to French2’s  substitution rate for the voiced fricative (57%), and French1’s voiced fricative substitution rate (81%) was closer to French2’s voiceless substitution rate (90%). This is interesting because while their rates for each fricative don’t agree, their substitution sounds do.

Tables French1


  f t ʔ # %
MR 2 1 3/9 33%
SA/L1 2 2/5 40%
SB/NS 1 1 1 3/7 43%
Total 5 1 1 1 8/20 40%



  d ɾ θ t # %
MR 10 4 1 1 16/18 89%
SA/L1 5 5/9 55%
SB/NS 8 1 1 10/11 91%
Total 23 4 2 1 1 31/38 81%

French1’s most common voiceless fricative replacement was, like Cantonese1, labiodental [f]. She supplies this substitution a little more often in her monologue and dialogue with French2. In her dialogue with the native speaker, she has more varied substitutions for the voiceless fricative, including glottal stop [ʔ] and postalveolar affricate [tʃ]. She also replaces using alveolar plosive [t] once in her speech alone.

The voiced fricative sound replacements for French1 are most varied in her recording alone, as she replaces using [d ḏ ɾ θ] in her Monologue Reading alone. Her dialogues however see a bit less variation in voiced interdental fricative substitution. Her dialogue with French2 only reveals replacements using alveolar plosive [d]. She replaces with [d] most often in her speech alone where she also articulates dental plosive [ḏ] nearly half the time. [ḏ] is inexistent in the rest of French1’s recordings.

Tables French2


  f t # %
MR 7 7/9 77%
SB/L1 5 1 6/7 86%
SA/NS 5 5/5 100%
Total 17 1 18/20 90%




  d θ f ɾ t v # %
MR 1 3 2 2 1 1 1 11/18 61%
SB/L1 5 1 1 7/11 63%
SA/NS 2 2 4/9 44%
Total 8 6 2 2 1 1 1 1 22/38 57%

French2’s most common voiceless replacement sound is also labiodental fricative [f]. However, unlike French1 and Cantonese1, French2 uses [f] in abundance. 17/20 replacement sounds were [f] for French2. He only ever had one other replacement sound, an alveolar plosive [t] in his dialogue with French1. French1 did not provide any alveolar plosives in her dialogue with French2, so French1’s sole alveolar plosive cannot be attributed to convergence. Divergence is also probably unlikely, though French1 did only replace using [f] in her dialogue with French2, where her other recordings had variable substitutions. It might be that French1 was converging her interdental replacements to match the abundant amount of [f] replacements French1 was providing. However, French1’s use of [f] does not rise from her monologue to her dialogue with French2, so perhaps she wasn’t converging after all.

Alveolar plosive [d] is also French2’s most common replacement sound for the voiced fricative, but he articulates dental plosives [ḏ] in not just his monologue but in all his other recordings as well, unlike French1. He also devoices both plosives one time each, replacing the voiced fricative with [t] and [ṯ] instead of their voiced pairs. Like French1, French2 has the most varied substitutions in his speech alone, he replaces with a flap [ɾ] once in his dialogue with French1, but replaces using only dental and alveolar plosives in his dialogue with the native speaker. French1’s highest accurate articulation rate is in his dialogue with the native speaker, the dialogue in which he only used plosives as substitutions for the voiced fricative. French1 could be converging his voiced replacements to match more closely the native speaker’s articulation, but truly, these substitutions don’t match in place and manner to the native speakers 100% accurate articulation of all the fricatives.


            The Portuguese participants had differing rates of accurate articulation for both fricatives. Portuguese1 articulates the voiced fricative (11%) less accurately than the voiceless (35%) overall. Portuguese2 does too, but with less of a range between his rates. We find Portuguese2’s substitutions rate for both fricatives at about the same rate (90% and 92%).

Tables Portuguese1


  f t ð # %
MR 3 1 1 1 6/9 66%
SA/L1 5 5/5 100%
SB/NS 1 1 2/7 28%
Total 9 1 1 1 1 13/20 65%




  ɾ d ʔ v # %
MR 10 4 1 1 16/18 89%
SA/L1 4 2 2 8/9 88%
SB/NS 5 1 1 1 8/11 72%
Total 19 7 3 1 1 1 32/38 84%


Portuguese1’s most common sound substitution for the voiceless fricative is labiodental fricative [f], and it is the only sound substitution she uses for the voiceless fricative when speaking with Portuguese2. All other of Portuguese1’s sound substitutions for the voiceless fricative only occur one time in all her realms of speech.  Like French1 and French2, Portuguese1 has most varied substitutions when speaking alone. She substitutes using deletion, a voiced interdental fricative [ð], and a dental plosive [ḏ] in her reading alone. Also, she replaces using an alveolar plosive [t], also only once, in her dialogue with the native speaker.

Portuguese1’s sound substitutions for the voiced fricative were, like most other participants, in great supply and more varied than her voiceless fricative substitutions. In her speech alone, Portuguese1’s most common replacement sound of the voiced was the dental fricative [ḏ], which goes down in suppliance in relation to other sound substitutions in her dialogue with Portuguese2, but then rises to the most common substitution when speaking with the native speaker. In her dialogues with the native speaker and Portuguese2, she uses alveolar plosive [d] as a substitution sound, where in her dialogue alone, she does not.

Tables Portuguese2


  f t d dᴣ ʔ # %
MR 4 1 1 1 7/9 77%
SB/L1 6 6/7 86%
SA/NS 4 1 5/5 100%
Total 10 5 1 1 1 18/20 90%




  d t ɾ θ # %
MR 10 3 2 15/18 83%
SB/L1 2 5 2 1 1 11/11 100%
SA/NS 4 2 1 1 1 9/9 100%
Total 16 10 3 2 2 2 35/38 92%

Like Cantonese1, both French participants and Portuguese1, Portuguese2’s most common replacement sound of the voiceless fricative is also labiodental fricative [f]. He uses this sound as a substitution in both his monologue recording alone and his dialogue with Portuguese1, but in his dialogue with the native speaker, we do not find [f] as a substitution sound at all. Instead, Portuguese2 has more alveolar plosives [t] as substitutions in his dialogue with the native speaker. Other substitutions only occur once voiced alveolar plosive [d] and voiced postalveolar affricate [dᴣ] occur as substitutions once each in Portuguese2’s reading alone and a glottal stop substitution exists once in his dialogue with the native speaker for the voiceless fricative.

            Portuguese2’s most common voiced fricative substitution is a voiced alveolar plosive [d], it occurs the most in his speech alone and the least in his dialogue with Portuguese1, although, he seems to vary between the voiced dental and alveolar plosives as replacement sounds throughout all of his recordings and there doesn’t seem to be a definite pattern. He substitutes using the voiceless interdental fricative [θ] in his speech alone twice, and has the same number of voiceless dental plosives as alveolar in both his dialogues with Portuguese1 and the native speaker.

The voiceless fricative substitutions for both Portuguese speakers matched exactly in number and sound substitution for their dialogue together. Both only substituted using a labiodental [f]. Portuguese1 varied between deletion, dental stopping, and voicing for her voiceless fricative replacements during her monologue. Portuguese2 also had varied substitutions of stopping, voicing, and affrication for the voiceless fricative during Monologue Reading. Both increased in usage of [f] as a replacement sound when speaking together. For their dialogues with the native speaker, Portuguese1 had the highest accurate articulation of the voiceless. Portuguese2’s was also the highest during his dialogue with the native speaker, in which Portuguese2 also increased in usage of [t] as a substitution.

The Portuguese pair of participants had a great deal of convergence occur when speaking with someone else. Portuguese2 was the most novice English speaker of all the participants and convergence of interdental replacement sounds occurred with him and Portuguese1 alone. The relationship between Portuguese1 and Portuguese2 was one of pretty close friendship. There is no evidence to document this fact except for the fact that the two lived in the same building near campus when this data was collected, and they have worked together on projects before. Portuguese2’s subconscious attempt to produce his phonemic interdental fricatives more like Portuguese1 makes sense, as Portuguese1 is perceivably a more proficient speaker. Portuguese1’s highest accurate articulation of the voiceless fricative was in her dialogue with the native speaker, showing convergence of manner, place, and voicing to that of the native speaker. The same is true for Portuguese1’s voiced fricative replacements, though not as extreme of a drop in substitution rate as with the voiceless fricative replacements.


            The Vietnamese participants vary most in substitution sounds of the voiceless fricative, there voiced substitutions are nearly identical, though not in the same distribution. Like the French participants, the Vietnamese participants’ substitution rates don’t agree for the fricatives. Vietnamese1, like most other participants, has a lower accurate articulation of the voiced fricative compared to the voiceless, as well as does Vietnamese2 who has higher accurate articulation of the voiceless over the voiced. While both participants don’t agree on most common voiceless substitution, they do agree on the most common substitution for the voiced fricative, which is the most common voiced fricative replacement – alveolar and dental plosives.

Tables Vietnamese1


  f t h # %
MR 4 1 5/9 55%
SB/L1 1 2 1 1 5/7 71%
SA/NS 1 1 2/5 40%
Total 5 4 1 1 1 12/20 60%



  d ɾ n ʔ θ # %
MR 8 1 3 3 1 16/18 89%
SB/L1 3 3 1 2 1 10/11 91%
SA/NS 3 1 2 1 1 8/9 89%
Total 14 4 4 6 3 2 1 34/38 89%


Vietnamese1 varies her substitution sounds of both interdental fricatives throughout all her recordings. Actually, the most common voiceless substitution is deletion for Vietnamese2. She deletes the voiceless fricative the most when she speaks alone, but a deletion process of her phonemic voiceless fricative is evident in her dialogue with Vietnamese2. In her dialogue with the native speaker, Vietnamese2’s most common replacement sound is an alveolar plosive [t].

Vietnamese1’s most common voiced fricative substitution is an alveolar plosive [d]. It is most common in her speech alone, but still surfaces as a substitution in both her dialogue recordings. Dental plosive [ḏ] shows up as a replacement sound in her reading alone, but is most common in her dialogue with Vietnamese2. Alveolar nasal [n] surfaces as a substitution in all of Vietnamese1’s realms of speech, but some of these are due to assimilation, described in section 3.1.

Tables Vietnamese2


  t k ʔ st # %
MR 3 2 2 1 8/9 88%
SA/L1 4 1 5/5 100%
SB/NS 6 6/7 85%
Total 13 2 2 1 1 19/20 95%



  d ɾ t n # %
MR 9 4 3 1 1 18/18 100%
SA/L1 5 3 8/9 89%
SB/NS 5 1 3 9/11 82%
Total 18 5 9 1 1 34/38 89%

            Vietnamese2’s most common voiceless fricative replacement is [t] and is in greatest suppliance in his dialogue with the native speaker, it being the only substitution that surfaces in this social context. He deletes the voiceless fricative in his speech alone, but deletion does not affect the voiceless fricative in any of the other realms of speech. His other substitutions are rare and often affected by phonology.

            Stopping is the only process affecting the voiced fricatives in his dialogues with the native speaker and Vietnamese1. However, in his speech alone there is one instance of an alveolar nasal substitution, but like Vietnamese1, this substitution was a result of assimilation.

The voiceless fricative substitutions for both Vietnamese participants vary throughout all of the study’s realms of speech. In their reading alone, the only substitution both had in common was deletion, and in their dialogue together, none of the substitution sounds matched in place or in manner. This could be a result of divergence in interdental fricative replacement sounds. In their dialogues with the native speaker, both replaced using [t] most frequently. Both participants’ highest accurate articulation for the voiceless fricative was in their dialogue with the native speaker.

For the voiced fricative, both most commonly replace using the dental plosive [ḏ] and the alveolar plosive [d] in all of their realms of speech. All alveolar nasal [n] replacements in for Vietnamese1 and Vietnamese2 are total or partial assimilation processes. Vietnamese2 uses flapping consistently throughout all of his recordings, while Vietnamese1 lacks flap replacements in her dialogue with Vietnamese2. The dialogue with the native speaker had the highest accurate articulation rate for the voiced fricative as well, though Vietnamese1’s rate was tied with her rate for her speech alone.


            As previously stated, the voiceless fricative was more often accurately articulated than the voiced fricative, with the exclusion of French2 who articulated the voiced fricative more accurately than the voiceless in every realm of speech, others were Arabic1 (S/L1), Portuguese1 (S/L1), Portuguese2 (S/NS),  and Vietnamese2 (SL1, S/NS). However in this section we are not concerned with differences in accurate articulation between fricatives, but between realms of speech. Examine Table 4.4.1. (Bolded scores are the highest rate of accurate articulation for the three realms of speech and parenthetical scores mean that the same value is tied for highest with another realm of speech for the same participant)

Table 4.4.1 Accurate articulation ratios per social realm

θ ð θ ð θ ð
Arabic1 67% (56%) 40% (56%) 71% 37%
Arabic2 78% 34% 100% 64% 80% 45%
Cantonese1 34% 0% 20% 0% 15% 0%
Cantonese2 100% 39% 86% 73% 80% 34%
French1 67% 11% 60% 45% 57% 9%
French2 23% 39% 14% 37% 0% 56%
Portuguese1 34% 11% 0% 12% 72% 28%
Portuguese2 23% 17% 14% 0% 0% 0%
Vietnamese1 45% (11%) 29% 9% 60% (11%)
Vietnamese2 12% 0% 0% 11% 15% 18%

Recall from Section 1.4 that the Monologue Reading (MR) is considered more characteristic of formal or careful speech and the Dialogue Readings (S/L1, S/NS) resemble more informal or casual speech. More generally, accurate articulation of the voiceless is higher in the Monologue Reading, and the voiced in either the Dialogue Reading with the native speaker or in the Monologue Reading. However, due to such varied results, it’s rather difficult to say if this study’s results compare more with Zampini (1994) or Dickerson & Dickerson (1977).


            The hypothesis, stated in section 1.4.1, speculates that labial [f v] and coronal [t d] are possible sound substitutions for Arabic, Cantonese, and French speakers (Kharma & Hajjaj, 1997; Chan & Li, 2000; Collins & Rodd, 1972). [s z] are also possible sounds in addition to [t d] and [f v] for French speakers. Portuguese speakers will most likely substitute [t] for the voiceless and [d] for the voiced, while Vietnamese speakers are also likely to substitute using [t d], but are likely to delete the voiceless and devoicing an otherwise accurately articulated voiced fricative to [θ] (Dreasher & Anderson-Hsieh, 1990; Santre, 1992).

While all participants (excluding the Arabic participants) exhibited substitution tokens of labial fricatives [f v] and coronal stops [ṯ ḏ] or [t d], [f] was the most common substitution for the voiceless and [ḏ d] were most common for the voiced. Plosive substitution was more widely used. Most participants did not use [v] to substitute the voiced fricative and those that did, did it rarely. The Arabic speakers had no labial fricatives as substitution sounds, which is contrary to possible sounds cited by Kharma & Hajjaj (1997). The Vietnamese speakers did have more deletions of the fricatives than any other participants which is in keeping with Santre (1992)’s findings.


            The results of the discussion indicate that accommodation can occur with interdental fricative replacement sounds of American English by non-native speakers. Arabic1 and Arabic2 seem to diverge some from one another in their dialogue together, as Arabic1 has the largest variation in her joint recording with Arabic2. Also Arabic2 has his lowest substitution rate of the voiced fricative in his recording with Arabic1. Arabic2 could have been trying to subconsciously highlight his and Arabic1’s distinctiveness by correctly articulating every voiceless interdental fricative in their reading together. However, Arabic1 and Arabic2 do have pretty similar replacement sounds for the voiceless in their recording together, so it’s a good possibility neither of the Arabic participants were using accommodation strategically, as seems to be true of both

Cantonese participants. Neither their rates nor substitutions seem to favor accommodation at all.

The French speakers exhibit more harmonious sound replacements than the two former participant pairs, but their rates of accurate articulation and substitution are opposing. Neither participant seems to be using accommodation with each other or the native speaker, though French1’s accurate articulation of the voiced and voiceless fricatives is lowest when speaking with the native speaker. Perhaps this could be called divergence, but really her rate of substitution in her dialogue with the native speaker is not much higher than her other realms of speech, so, then again, it probably isn’t divergence.

The Portuguese and the Vietnamese participants’ results were different, though. The Portuguese speakers see convergence of voiceless interdental sounds in their joint reading together. Both increase in usage of labiodental [f], and it becomes the only sound replacements both participants use in their Dialogue Reading together, but it can’t really be confirmed who was doing the converging, Portuguese1 or Portuguese2, or if perhaps they both were. Socially, it makes sense that Portuguese2 would be converging to Portuguese1’s sounds, as stated earlier in the chapter, but we find Portuguese1 having her highest accurate articulation rate of both fricatives with the native speaker, which suggests convergence here as well. Portuguese1’s convergence in her dialogue with the native speaker could suggest Portuguese1 was doing the converging in her dialogue with Portuguese2.

The Vietnamese participants’ substitution rates for both are lowest in their dialogues with the native speaker, which makes their accurate articulation rate of both fricatives highest with the native speaker. (Vietnamese2’s substitution rate for the voiced fricative is the same rate as her monologue, though).  None of the voiceless interdental fricative replacement sounds match for Vietnamese1 and Vietnamese2 in their dialogue together, yet in their other realms of speech, some do. Vietnamese1 increases in usage of stopping, while Vietnamese2 increases in his faithfulness to manner of the voiceless fricative. This suggests divergence on behalf of the Vietnamese participants in their joint reading, as well as convergence by both participants in their dialogues with the native speaker.

The study’s results also suggest that the voiced and voiceless interdental fricatives may operate differently in the non-native speaker’s mind. The realization of both is often a different place and manner of articulation, when phonemically, each fricative should only be independent in their voicing. The voiceless fricative has more varied substitutions for the most common replacements of each participant, but is most commonly labiodental fricative [f], while the most common voiced fricative replacement was a dental or alveolar plosive [ḏ d] for all participants. Such results seem to match Blevins (2006) findings of Southeastern English’s elimination of the interdental fricatives, the voiced turning to stopping [d] and the voiceless turning to labialization [f]. However the voiced seems to only stop word initially in Southeastern England’s English, and everywhere else it remains the voiced counterpart of [f] ([v]), which is definitely a pattern displayed by this survey’s participants. However, we only have 9 tokens of the voiced fricative word medially and 1 token word finally, and [v] was not the most common substitution for these other word places. Also, accommodation, seems to operate more freely with the voiceless fricative over the voiced.

How are we to explain this difference of the voiced and voiceless fricatives in the non-native speaker’s linguistic psyche?  Lombardi (2003) attempts to answer this question using Optimality Theory, but her analysis doesn’t explain different sounds chosen at different times by the same speaker. Brannen (2002) assumes full perception of interdental fricative features, but feature co-occurrence restrictions in the L1 grammar prevent second language speakers from combining certain features, thereby constructing a new output representation. Perhaps the perceptual features of each fricative do not transfer to the same co-occurrence feature restrictions, and thereby speakers realize the fricatives independently of voicing, place, and manner. Ultimately, it would take more data to prove this conjecture. Please see section 5.2 for further details.



            The findings of this study indicate that L2 speech accommodation in terms of matching or mismatching place, manner, and voicing of a particular sound replacement is

most characteristic of lower-level speakers, as we saw speech accommodation taking place with the Portuguese and Vietnamese pairs of participants, the pairs that have the participants with the lowest estimated linguistic experience in English, based on demographic information.  As hypothesized in the discussion, lower level speakers may be more likely to accommodate due to linguistic insecurity about their own non-native accents, while more proficient speakers are more confident about their abilities in English.

However, Portuguese1’s highest accurate articulation of both the voiced and voiceless fricatives are with the native speaker. This might indicate that subconscious convergence of phonemic interdental fricative replacement sounds occurred on behalf of Portuguese1 and not Portuguese2 after all.

The pair with the participant who had the second-lowest estimated level of linguistic experience in English was the Vietnamese participants. In this case, the data indicated a divergence in chosen speech sounds to replace the fricatives. This divergence, according to Giles et al. (1991) would indicate a want to emphasis distinctiveness and a desire to increase the social fissure between the speaker and listener. If we follow that less experienced speakers are more likely to use speech accommodation when choosing interdental fricative substitutions, then it was on behalf of Vietnamese2 that this divergence occurred. But what reason could Vietnamese2 have for wishing to increase the social distance between him and his two other interlocutors in the study? Perhaps he wanted not to necessarily make the social distance between Vietnamese1 and himself greater, but instead to subconsciously declare their linguistic distinctiveness from Vietnamese1 and the native speaker. However, this finding may seem weak for we drew its conclusion from only 2 out of the 5 participant pairs.

The next finding of this study is much stronger than the previous as it applies to at least one participant in each of the native language pairs (except the Arabic participants). The study’s participants’ (excluding Arabic1, Arabic2 Cantonese2, and Portuguese2) most common substitution sounds for each of the interdental fricatives did not only differ by voicing, as phonemically, that’s truly the how the American English interdental fricatives are realized, except, of course, the stop modification of the voiced fricative (Wolfram & Johnson, 1982). Instead the most common replacement sounds for the opposite voicing fricatives differed in both place and manner. For most participants, the only characteristic of each fricative they kept was voicing. That is that manner and place of articulation were not as faithful to the original characteristics of the interdental fricative phonemes as voicing. Though opposite voicing of a substitution sound does occur and quite often throughout the data, it is not the most common characteristic of all substitution sounds supplied by participants.


            This section describes the integrity that can be given to the study’s data collection, its instruments, and its analysis. Due to the exploratory nature of the data collection and the data analysis’ inherent speculation, such an analysis of the study’s limitations is necessary.

As previously stated, the researcher was not in the room while the recordings were taking place. This was one of the steps taken to create a more authentic context. The context itself was not an exact replica of an actual dialogue that would occur between two people because the participants were reading these lines from a script. The best kind of data collection to draw conclusions from is spontaneous speech, and although spontaneous speech could have been collected from these participants, due to the interdental fricatives infrequency in lexical English words, there was no way to guarantee participants would say enough of these sounds to create analyzable data.

Also, another constraint that could have possibly affected interdental fricative replacements is that the participants had knowledge of their being audio recorded. Such knowledge can definitely change the atmosphere/social context and ultimately change the replacement sounds that the participants would have otherwise had in spontaneous speech without knowledge of being audio recorded. Wolfram (1998) states that careful examination of particular forms of a certain speaker’s speech may often be differentiated on the basis of how frequently particular forms are used. That is, individual speakers may fluctuate in their choices of variants, sometimes using one segment and other times using a different variant segment. Wolfram (1998) calls this inherent variability and Weinberger (1994) calls the phenomenon differential substitution. Inherent variability or differential substitution reflects the fact that fluctuation of variants is an internal part of a single linguistic system and should not be considered the result of importations. Such variability is what is most likely happening with most of the participants’ interdental fricative substitutions.

The instruments were created solely for this study’s data collection. So, the instruments succeeded in keeping as authentic of speech as possible, while having as many analyzable tokens of the phonemic interdental fricatives as possible. Of course the observer’s paradox always comes into play when a study participant consciously knows of their being audio or video recorded. Such conscious knowledge can change the speech style choice for each speaker (Briggs 1986). The Monologue Reading was written to be read like a poetic reading, a reading participants would normally read alone or in front of an audience perhaps. The Monologue Reading instrument was created and adapted from a dramatic poem by Dr. Maya Angelou that had been read publically at Bill Clinton’s inauguration in 1993 that she had composed for that day specifically. As there were already quite a few words containing interdental fricatives in the poem already, additional words containing the fricatives were inserted, replacing some of the poem’s last stanza’s original wording.

So many other variables can be affecting the substitution sounds chosen by participants: phonetic environment, native language, linguistic experience in English, that it is difficult to say if any of the substitutions provided by the study’s participants can be attributed to social variables alone.

We see our most interesting sound substitutions for the interdental fricatives occurring with the Portuguese and Vietnamese participants. The Portuguese participants seemed to converge in the sound replacements they chose, while the Vietnamese participants seemed to diverge. Also Portuguese1 and both Vietnamese participants had their highest rate of accurate articulation when speaking with the native speaker. This bit of information could tell us that the Portuguese and Vietnamese participants were, at least in part, using speech accommodation according to Giles et al. (1991).


            Although there is much research in the area of L2 SAT, much of this sociophonetic cross-area of research is unexplored. Based on the findings of this study, an extension of the procedure and analysis of this study could try to find participants that have more varied degrees of linguistic experience in the second language as the speech convergence and divergence in terms of interdental substitutions occurred with pairs of participants that had one predicted ill-experienced member. This lack of experience estimate is based solely on the interdental replacement sounds and nothing more about each participant’s linguistic ability or strength level in English. That is, other indicators of English experience (rate of speech, other sound substitutions, natural fluidity of speech, etc.) were not observed. Since speech accommodation can occur with any individualistic linguistic variable, possible studies in second language speech accommodation are virtually infinite.

Other findings of the study indicate research outside the area of L2 SAT. Research on phonological awareness similarities and differences can be drawn out of the findings for interdental fricative replacements chosen by each participant, indicating that phonological awareness is acquired and represented in the brain similarly or differently between first and second language speakers.


Asher, J. J., & Garcia, R. (1969). The optimal age to learn a second language. The Modern Language       Journal, 53: 334-341.

Beebe, L. M., and Giles, H. (1984). Speech-accommodation theories: A discussion in terms of second-language acquisition. International Journal of the Sociology of Language, 46, 5-32.

Beebe, L. M. (1977). The influence of the listener on code-switching. Language Learning, 27, 332-39.

Blevins, J. (2004). Evolutionary Phonology: The emergence of sound patterns. Cambridge: Cambridge University Press.

Blevins, J. (2006). New perspectives on English sound patterns: Natural and unnatural in             evolutionary phonology. Journal of English Linguistics, 34, 6-25.

Bongaerts, T., van Summeren, C., Planken, B., & Schils, E. (1997). Age and ultimate attainment in the pronunciation of a foreign language. Studies in Second Language Acquisition, 18, 447-65.

Bourhis, R. Y., & Giles, H. (1977). The language of intergroup distinctiveness. In H. Giles (ed.), Language, Ethnicity, and Intergroup Relations (119-135). London: Academic Press.

Brannen, K. (2002). The Role of Perception in Differential Substitution. The Canadian  Journal of Linguistics, 47 (1/2), 1-46.

Briggs, C. L. (1986). Learning how to ask: A sociolinguistic appraisal of the role of the interview in social science research. Cambridge: Cambridge University Press.

Brown, C. (2000). The interrelation between speech perception and phonological acquisition from infant to adult. In J. Archibald (Ed.), Second. Oxford, England: Blackwell.

Cappella, J. N. (1981). Mutual influence in expressive behavior: Adult-adult and infant-adult dyadic interaction. Psychological Bulletin, 89: 101-32.

Chan, A. M. & Li, D. S. (2000). English and Cantonese Phonology in Contrast: Explaining Cantonese ESL Learners’ English Pronunciation Problems. Language, Culture and Curriculum, 13(1), 67-85.

Collins, B. & Rodd, J. (1972). English pronunciation problems of francophonic West Africans. English Language Teaching Journal, 27(1), 79-88.

Derwing, T. & Munro, M. J. (1997). Accent, intelligibility, and comprehensibility (evidence from four L1s). Studies in Second Language Acquisition, 19: 1-16.

Dubois, S., & Horvath, B. (1998). Let’s tink about dat: Interdental fricatives in Cajun English. Language Variation and Change, 10, 245-261.

Dickerson, L. B. & Dickerson, W. B. (1977). Interlanguage phonology: Current research and future directons. In S.P. Corder and E roulet (Eds.), The notions of  simplification, interlanguages and pidgins and their relation to second language pedagogy (18-30). Geneva: Droz.

Dreasher, L. M., & Anderson-Hsieh, J. (1990). Universals in interlanguage phonology: The case of Brazilian ESL learners. Papers and Studies in Contrastive Linguistics, 26, 69-92.

Eckman, F., Elreyes, A., & Iverson, G. (2003). Some principles of second language phonology. Second language Research, 19, 3: 169-208.

Ehrlich, S. (1997). Gender as social practice: Implications for second language acquisition. Studies in Second Language Acquisition 19, 421–446.

Elliot, R. E. (1995). Field independence/dependence, hemispheric specialization, and attitude in relation to pronunciation in Spanish as a foreign language. The Modern Language Journal, 79, 356-71.

Fathman, A. (1975). The relationship between age and second language productive ability. Language Learning, 25, 245-53.

Flege, J. E., Frieda, E. M., & Nozawa, T. (1997). Amount of native-language (L1) use affects the pronunciation of an L2. Journal of Phonetics, 25, 169-186.

Flege, J. E., Yeni-Komshian, G., & Liu, H. (1999). Age constraints on second language acquisition. Journal of Memory & Language, 41, 78-104.

Fowler, A. E. (1991) How early phonological development might set the stage for phonological awareness. In S. Brady & and D. Shankweiler (eds.). Phonological Processes in Literacy (97–117). Hillsdale, NJ: Erlbaum.

Gatbonton, E. (1978). Patterned phonetic variability in second-language speech: A gradual diffusion model. Canadian Modern Language Review, 34: 335-47.

Giles, H., Coupland, N., & Coupland, J. (1991). Accommodation theory: Communication, context, and consequence. In H. Giles & J. Coupland (Eds.), Contexts of Accommodation (1-68). Cambridge: Cambridge University Press.

Giles, H., & Johnson, P. (1987). Ethnolinguistic identity theory: A social psychological approach to language maintenance. International Journal of the Sociology of Language, 68 (6).

Guiora, A. Z., Beit-Hallahmi, B., Brannon, R. C. L., Dull, C. Y., & Scovel, T. (1972). The effects of experimentally induced changes in ego stages on pronunciation ability in a second language: An exploratory study. Comprehensive Psychiatry, 13, 421-28.

Han, Z. (2004) Fossilization in adult second language acquisition. Tonawanda, NY: Multilingual Matters Ltd.

Hancin-Bhatt, B. (1994) Phonological Transfer in Second Language Perception and Production. Unpublished Ph.D. dissertation. University of Illinois at Urbana-Champaign.

Harris, K. S. (1958). Cues for the discrimination of American English fricatives in spoken syllables. Language and Speech 1, 1-7.

Heinz J. M., & Stevens, K. N. (1961). On the properties of voiceless fricative consonants. Journal of the Acoustical Society of America 33, 589-96.

Hernandez-Chavez, E., Burt, M. & Dulay, H. (1978). Language dominance and proficiency testing: Some general considerations. NABE Journal, 3:41-54.

Inhelder, B. & Piaget, J. (1958). The growth of logical thinking from childhood to adolescence. New York: Basic Books.

Jongman, A., Wayland, R., & Wong, S. (2000). Acoustic characteristics of English fricatives. Acoustical Society of America,. 108 (3), 1252-63.

Kharma, N. & Hajjaj, A. (1997). Errors in English among Arabic Speakers: Analysis and Remedy. Beirut, Lebanon: York Press.

Krashen, S. (1977). Some issues relating to the monitor model. Paper presented at the Eleventh Annual TESOL Convention, Miami, Florida.

Labov, W. (1966) The Social Stratification of English in New York City. Washington, DC: Center for Applied Linguistics.

Labov, W. (1972). Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press.

Laing, S. P. (2003). Assessment of phonology in preschool African American Vernacular English speakers using an alternate response mode. American Journal of Speech-Language Pathology, 12: 273-81

Lardiere, D. (1998) Case and tense in the fossilized steady state. Second Language Research, 14: 1-26.

Larsen-Freeman, D. (2005). Second language acquisition and the issue of fossilization: There is no end, and there is no state. In Z. Han & T. Odlin, Studies of fossilization in second language acquisition (189-200). Clevdon, UK: Multilingual Matters.

Lenneberg, E. H. (1967). Biological Foundations of Language. New York: Wiley.

Lombardi, L. (2003). Second language data and constraints on Manner: explaining substitutions for the English interdentals. Second Language Research, 19,3, 225–250.

Long, M. H.  (1990). Maturational constraints on language development. Second Language Acquisition, 12, 251-85.

Mack, S. (2010). A sociophonetic analysis of perception of sexual orientation in Puerto Rican Spanish. Laboratory Phonology 1, 1, 41-64.

Major, R. C. (1986). The ontogeny model: Evidence from L2 acquisition of Spanish rLanguage Learning 36: 453-503.

Major, R. C. (1987). A model for interlanguage phonology. In G. Ioup and S. H. Weinberger (Eds.), The Acquisition of a Second Language Sound System (101-24). Cambridge, MA: Newbury, 101-24.

Mitchell, B., & Robinson, F. C. (1992). A Guide to Old English, Fifth Edition. Cambridge: Blackwell.

Meador, D., Flege, J. E., & Mackay, I. R. A. (2000). Factors affecting the recognition of words in a second language. Bilingualism: Language and Cognition, 3, 55-67.

Moyer, A. (1999). Ultimate attainment in L2 phonology. Studies in Second Language Acquisition, 21, 81-108.

Olson, L. L. & Samuels, S. J. (1973). The relationship between age and accuracy of foreign language and pronunciation. Journal of Educational Research, 66, 263-68.

Oyama, S. (1976). A sensitive period for the acquisition of a nonnative phonological system. Journal of Psycholinguistic Research, 5, 261-83.

Patkowski, M. S.  (1980). The sensitive period for the acquisition of syntax in a secondary language. Unpublished doctoral dissertation, New York University.

Piper, T. & Cansin, D. (1988) Factors influencing the foreign accent. The Canadian Modern Language Review, 44, 334-42.

Piske, T., MacKay, I. R. A., Flege, J. E.  (2001). Factors affecting degree of foreign accent in an L2: A review. Journal of Phonetics, 29, 191-215.

Price, D. (2007). Language skills: Formal speaking versus informal speaking. Retrieved from http://www.helium.com/items/261609-language-skills-formal-speaking-versus-informal-speaking.

Purcell, E. T. & Suter, R. W. (1980). Predictors of pronunciation accuracy: A reexamination. Language Learning, 30, 271-87.

Rogers, H. (2000). The Sounds of : An Introduction to Phonetics. Pearson Education Limited.

Reetz, H & Jongman, A. (2009). Phonetics: Transcription, production, accoustics, and perception. Henning Reetz and Allard Jongman.

Santre, P. A. (1992). The way South Vietnamese pronounce English. Unpublished Thesis. Victoria University, Australia.

Scovel, T. (1969). Foreign accents, language acquisition, and cerebral dominance. Language Learning, 19, 245-53.

Scovel, T. (1988) A time to speak: A psycholinguistic inquiry into the critical period for human speech. New York: Newbury House/Harper & Row.

Selinker, L. (1972) Interlanguage. International Review of Applied Linguistics, 10:209-31.

Schmidt, R. (1990).  The role of consciousness in second language learning. Applied Linguistics, 11(2), 129–15.

Smit, A. B. (1986). Ages of speech sound acquisition: Comparisons and critiques of several normative studies. Language, Speech, and Hearing Services in Schools 17: 175-186.

Snow, C. E. & Hoefnagel-Höhle, M. (1977). Age differences in the pronunciation of foreign sounds. Language & Speech, 20, 357-65.

Suter, R. W. (1976). Predictors of pronunciation accuracy in second language learning. Language Learning, 26: 233–253.

Tahta, S., Wood, M., Loewenthal, K. (1981). Foreign accents: Factors relating to transfer of accent from the first language to a second language. Language and Speech, 24 (3):265-72.

Tajfel, H. (1974). Social identity and intergroup behaviour. Social Science Information, 13, 65-93.

Tarone, E.E. (1979). Interlanguage as a chameleon. Language Learning, 29: 209-31.

Tarone, E.E. . (1982). Systematicity and attention in interlanguage. Language Learning, 32: 69-84.

Tarone, E.E. . (1983). On the variability of interlanguage systems. Applied Linguistics,

4: 142-63.

Tarone, E.E. . (1987). Some influences on the syllable structure of interlanguage. In G. Ioup (Ed.), Interlanguage phonology: The acquisition of a second language sound system (232-47). Newbury House.

Taylor, D. M., & Royer, L. (1980). Group processes affecting anticipated language choice in intergroup relations. In H. Giles, W. P. Robinson, & P. M. Smith (eds.), Social Psychological Perspectives, 185-92. Oxford: Pergamon.

Thompson, I. (1991) Foreign accents revisited: the English pronunciation of Russian Imigrants. Language Learning, 41, 177-204.

Van Lier, L. (1988). The classroom and the language learner: Ethnography and second-language classroom research. London.

Weinberger, S. H. (1994). Theoretical foundations of second language phonology. Doctoral Dissertation, University of Washington.

White, L. (1989) Universal Grammar and Second Language Acquisition. Amsterdam: John Benjamins Publishing Company.

Wolfram, W., & Johnson, R. (1982). Phonological analysis: Focus on American English.             Englewood Cliffs, NJ: Prentice-Hall Inc.

Wolfram, W. (1998). American English. Oxford: Blackwell Publishing.

Wolfson, N. and Elliot, J. (Eds.). (1983). Sociolinguistics and Language Acquisition. Rowley, MA: Newbury House Publishers, Inc.

Yavas, M. (2006). Applied English phonology. Oxford: Blackwell Publishing.

Zampini, M. L. (1994). The role of native language transfer and task formality in the acquisition of Spanish spirantization. Hispania, 77 (3): 470-81.


Bulgarian Phonetics

Bulgarian Phonetics

The Bulgarian language is a member of the Slavic language family, which is a member of the Indo-European family of languages. Other members of the Slavic family include Russian, Czech, Polish, and Bosnian. Bulgarian is classified as being a South Slavic language. The Southern Slavic languages can be divided into two subgroups: Western and Eastern. The Western subgroup is composed of Serbian, Slovene, Croatian and Bosnian, and the Eastern is Bulgarian and Macedonian. Though Bulgarian and Macedonian are two recognizably different languages, Bulgarian and Macedonian are mutually intelligible. Mutual intelligibility is a linguistic term that defines a relationship between languages in which speakers of two different but genetically related languages can readily understand one another without intentional study or extraordinary effort.

The Bulgarian language is the official language of Bulgaria, and it is spoken in the European Union. It is a recognized minority language in Ukraine, Serbia, Romania, Moldova, and among emigrant communities worldwide. In total, the estimated number of people that speak Bulgarian fluently is between 9 and 12 million. It is regulated by the Institute of Bulgarian at the Bulgarian Academy of Sciences (BAS) in Sofia, Bulgaria.

This phonetic data is not representative of the whole Bulgarian language. This data represents the speech of only one consultant, a 52 year old woman from Plovdiv, the Southern Central part of Bulgaria, known as the Thracian valley. She speaks standard Bulgarian and not a specific dialect. The only thing typical of the Plovdiv area is that speakers tend to make some unrounded close-mid front vowels /e/ an unrounded tense close front vowel /i/. Overall, she has spent 14 years in the United States, but she spends 3 months a year in Bulgaria.


In the transcriptions which can be found on appendix A, the syllabic consonant and primary stress marker diacritics didn’t exist in the character viewer on my MacBook Pro. I replaced the nonexistent diacritics with the closest character I could find. So primary stress is marked with this character ( ˡ ) and syllabic consonants have a dot instead of a line under them, e.g. /ṃ/.  Also the small, circular voiceless diachritic is not directly under the consonant to which it is referring. It is to the left of the consonant phoneme, between the phoneme character that comes before it and the voiceless consonant.


The raw data for stress can be found in Appendix A.

Bulgarian is a stress language. Like all stress languages, Bulgarian shows evidence of having one syllable in every word that has primary stress. Most of the words have primary stress that falls on the first syllable. 32 out of 50 words, 64% of the words had primary stress that fell on the initial syllable. Of these 32 words, 17, or 53.1% of them are two syllables. 32, or 64% of the words in the data are two syllables. So, a little more than half, about 53% of two syllable words have primary stress that falls on the word initial syllable. 6 of the 50 words, or 12% had three syllables. Only one word had four syllables: walk: [rəsˡhɤ̥ᴣdṃsej]. This is technically the infinitive, which could be the reason as to why it is longer than all of the other words in the data. Primary stress falls on both the first, second, and third syllables in three syllable words. In the four syllable word [rəsˡhɤ̥ᴣdṃsej], stress falls on the second syllable.

Sounds that occur in Bulgarian, but not in English.

Bulgarian does not use places of articulation that are different from English’s. Both are Indo-European languages, so it’s no surprise. However, Bulgarian has the alveolar trill, which is evident throughout a majority of Indo-European languages.

Some sounds that occur in English but not in Bulgarian

Though Bulgarian has syllabic consonants, it only has syllabic nasals. English has both syllabic nasals and syllabic approximants. Also, English has a lot of aspiration in its words, especially voiceless stops. Throughout the data, aspiration is minimal in Bulgarian. In addition, English makes use of both the voiced and unvoiced interdental fricatives, while Bulgarian does not.


The vowel chart can be found in Appendix B

The Bulgarian language consists of seven primary vowels: the unrounded tense close front vowel /i/, the unrounded close-mid front vowel /e/, the unrounded open-mid front vowel /ɛ/, the unrounded open front vowel /a/, the mid central schwa /ə/, the unrounded close-mid back vowel /ɤ/, and the rounded close-mid back vowel /o/; as well as one diphthong, the unrounded open front onglide /ja/. Bulgarian primary vowels include all four of the unrounded front vowels. There were other vowels recorded, but they appear significantly less frequently than the primary vowels. They are the close-mid /I/, the rounded close-mid central vowel /ɵ/, the front /æ/, the central /ɐ/, the rounded open-mid back vowel /ɔ/, the close back vowel /ʊ/, and the rounded close back vowel /u/. The data also had the diphthongs: the close-mid front onglide /je/, the unrounded close-mid front onglide /je/, the unrounded close-mid front offglide /ej/, the  rounded close-mid central onglide /jɵ/, and the rounded open-mid back offglide /ɔj/. The most common vowels are the /i/, /e/, and /ə/, all three appearing more than twenty times throughout the fifty words in isolation. The schwa unlike in English, is not an unstressed vowel. It is actually a letter sound in the Cyrillic alphabet.

Vowel Length

The raw data for vowel lengths can be found in Appendix C.

I determined the length of vowels in different word positions from only the vowels in the words that are said in isolation. It’s easy to determine word position sounds if the word is said in isolation. In context, it’s not as easy to tell if a vowel is an initial, medial, or final word sound. There were breaks in the speech of the phrases, but that doesn’t necessarily mean that those short pauses in speech were between words.

Like English, vowel lengthening in Bulgarian is most common when a vowel is a word final sound. With every one of the primary vowels and the primary diphthong, the average vowel length for the word final position was longer than the same vowel being a word medial or initial sound. The only vowel sound that appears as a word initial sound is /e/. /ɛ/ is the only vowel to appear as a word medial sound only. The longest average length of a vowel is /a/ at an average length of .1884 seconds. The shortest vowel is /i/, whose average length is .0982 seconds. The shortest vowel is closed and the longest is open. Can I deduce that the longer vowels are going to be when the mouth is more open and the shorter vowels when the mouth is more closed? No, because the rest of the middle vowels are a mixture of different lengths and their hierarchical vowel length does not correspond to where they are produced in the mouth.

Every primary vowel’s average length is longer in words in isolation than in words in context. This is most likely due to my consultant producing all sounds more quickly in the phrases and articulating and fully pronouncing each sound in the words in isolation.

On and Offglides

The raw data for on and offglides can be found in Appendix D.

The only diphthongs that exist in the data set are on and offglides. The most common use of the palatal approximate /j/ is most commonly an onglide instead of an offglide, its most common use in English. The most common diphthong is the unrounded open front  onglide /ja/. It appears more than twice as frequently as any other diphthong. The diphthong is usually a word medial sound, being the second or third sound in the word. However, the diphthong is also uttered as a word final sound, but is never a word initial sound. The only glide that is a word initial sound is the rounded close-mid central onglide /jɵ/. The second most common diphthongs are the close-mid front onglide /je/ and offglide. The offglide is most commonly a word final sound and the onglide is most commonly a word medial sound.


Nasalization of vowels does not occur enough in the data to say that it is necessary to nasalize certain vowels when speaking Bulgarian or that nasalization can determine meaning. Nasalization only occurs twice. Once with the unrounded close-mid front vowel /e/: vegetable: [ˡzɛlẽntʃok] and once with the rounded close-mid back vowel /o/ meat: [ˡmɛsõ]. The /e/ nasalization is most likely due to the alveolar nasal that follows it. However, there are other vowels that precede an Alveolar Nasal /n/ that are not nasalized. Some examples are: mountain: [pɫəˡnina], one: [e̥dˡin], and woman: [̥ᴣejˡna]. Also, vowels that precede a Bilabial Nasal /m/ are not nasalized either. Examples are earth: [̥ˡzImja], eat: [jɵm], time: [vɾɛˡme]. Time for /m/ and mountain for /n/ do not nasalize the proceeding vowel probably because the vowel is the final sound in an unstressed syllable while the nasal that follows is the onset of the stressed syllable. Therefore, the nasalization may not affect the vowel. There’s a similar problem with woman. However, I’m not sure if diphthongs can be nasalized anyway, since they require the movement of the mouth as air passes through it to create the glide.

The nasalized rounded close-mid back vowel /õ/ being a final word sound is intriguing. Reasons as to why this specific vowel was nasalized as a word final sound are unclear since /o/ is a word final sound in other words but isn’t nasalized. Two examples of this non-nasalization are clothing: [ˡɔbleklo] and  grain: [̥zɛrˡno]. I would need more data to examine nasalization more closely since it does not occur very frequently.

Syllabic Consonants

The raw data for syllabic consonants can be found in Appendix E.

The only consonants that are found to be syllabic in the data are the alveolar nasal /n/ and the bilabial nasal /m/. The syllabic alveolar nasal /ṇ/ appears as both a word final and a word medial sound. /ṇ/ in word final position follows both an alveolar flap /ɾ/ and a voiced labiodental fricative /v/. Syllabic bilabial nasal /m/, like syllabic /n/, is both a word medial and word final sound. /ṃ/ follows a voiced velar plosive /g/ and a voiced alveolar plosive /d/ twice. Both of the sounds that /ṃ/ follows are voiced sounds. The syllabic alveolar nasal /ṇ/ follows an alveolar lateral approximant /l/. All of the words with syllabic consonants have what speakers of English would call peculiar consonant clusters. These consonant clusters are how Bulgarian gets the syllabic nasals.


The raw data for rhotic vowels can be found in Appendix F.

There are eleven words and four phrases that contain an Alveolar Flap /ɾ/. Seven of the eleven words and one of the four phrases with an Alveolar Flap are followed by a rhotic vowel, and three of the words that have a /ɾ/ are followed by a non-rhotic vowel. What makes certain vowels rhotic and others not? All the unrounded front vowels are rhotic every time they follow an alveolar flap. However, /ɛ/ is both rhotic in time: [vɾɛme] and  non-rhotic in leader: [liɾɛr].

The most common rhotic vowel is the unrounded open front vowel /a/.  However, there are rhotic vowels that appear significantly less frequently than /a/: the unrounded open-mid front vowel /ɛ/, the unrounded close-mid front vowel /e/ , and the unrounded close front vowel /i/. The rhotic vowels are of all degrees of openness, but they are all articulated at the front of the mouth and are unrounded. The non-rhotic vowels that follow an alveolar flap are the unrounded open-mid front vowel /ɛ/, schwa /ə/, and the unrounded open front onglide /ja/. /ɛ/ is the only vowel that appears both  rhotic and non-rhotic following an alveolar flap /ɾ/. However, if concluded by this data set that all vowels that are front and unrounded are rhotic following an alveolar flap /ɾ/, the fact that /ɛ/ is non-rhotic in the word leader [liɾɛr], could be attributed to its environment. No words that contain a rhotic vowel following /ɾ/ are directly followed by a rhotic alveolar trill /r/.


Aspiration occurs twice in the data. Once in book: [kʰnigʌ] and again in house: [kʰəʃtɤ]. Only the voiceless velar plosive /k/ is aspirated, and only when it is a word initial sound. The aspiration in house could be because a mid central vowel /ə/ follows it, but the aspiration of /k/ in book is interesting since an alveolar nasal, a closed mouth sound, follows it. Neither English or Bulgarian distinguish meaning by aspiration.

Voiceless Consonants

The raw data for devoicing of consonants can be found in Appendix G.

There are five consonants in the data set that are devoiced: the alveolar fricative /z/, the postalveolar fricative /ᴣ/, the labiodental fricative /v/, the velar plosive /g/, and the alveolar plosive /d/. The most common devoicing is of the fricatives /z/ and /ᴣ/. /z/ is devoiced before an unrounded front vowel as in earth: [̥zImja], before an unrounded open-mid front vowel as in grain: [̥zɛrno], and before an unrounded close front vowel as in language: [e̥zik]. It is devoiced in both a word initial position and a word medial. /ᴣ/ is devoiced in word initial, medial, and final sounds, man: [mʊ̥ᴣ], walk: [rəshɤ̥ᴣdmsej], and woman: [̥ᴣejna]. There is no particular pattern for the devoicing of the postalveolar fricative /ᴣ/. /d/ is devoiced as both a word initial and word medial sound as in one: [e̥din] and tree: [̥dʊɾvo] . /v/ and /g/ are the last two consonants that are devoiced. These consonants were devoiced only once each, /v/ in word initial position as in water: [̥voda], and /g/ in word medial position as in  bed: [ˡle̥gɫɤ].

Syllabic Structure

The raw data for syllabic structure can be found in Appendix H.

Consonant clusters are both at the beginning and end of words. The most common amount of consonant sounds in a row is two, though the most in a row in the data is three. Bulgarian’s Maximum Onset Principle (MOP) is the same as English, three. MOP is a “phonotactic principle which determines the number and permissible sequence of consonants in the onsets of syllables for a specific language” (Charkova). However, unlike English, Bulgarian’s MOP consonant clusters can be more than just /stɹ/. Consonant clusters of three include /sgɾ/ in building: [sgɾadə], /str/ in sister: [sIstra], and /gsm/ in smile: [gsmifka]. /sgɾ/ and /gsm/ are both word initial sounds, and /str/ is a word medial sound. According to the data, Bulgarian syllables are usually open, that is they end in a vowel and lack a final consonant sound, known as the coda, which is part of the rhyme. The rhyme is composed of the nucleus, otherwise known as the vowel, and the coda. A Bulgarian syllable’s structure is more likely to have an onset than a coda.

Words that are different in the context of a phrase than when in isolation.

The raw data for Words that are different in the context of a phrase than when in isolation can be found in Appendix I.

The word tall, while in isolation, has a rounded open-mid back vowel /ɔ/, but when said in the context of the sentence How tall is the mountain?, [visɔk] becomes [visokI], a rounded close-mid back vowel. Mountain [pɬənina] possesses an unrounded tense close front vowel /i/ when said in isolation, but when said in the context of the same sentence, has an /I/.

Drink [pjə] has a schwa onglide while said in isolation, but an Unrounded Tense Close Front vowel /i/ when said in the phrase Eat and drink.

In the sentence The red book is on the bed [tʃeɾvɛnətə knig enaleglɤt], the word red [tʃeɾvɛn] has a schwa between the flap and the voiced labiodental fricative /v/. Red also changes the unrounded open-mid front vowel and the alveolar nasal /ɛn/ to a syllabic alveolar nasal /ṇ/. The voiceless velar plosive /k/ in Book [kʰnigʌ], while said in isolation, is aspirated, but said in the context of the sentence, it becomes unaspirated.

Two Birds [ dvɛ ptitʃki] has an unrounded open-mid front vowel /ɛ/, but when said in isolation, two contains an unrounded close-mid front vowel offglide /ej/.

In the sentence I eat meat and vegetables [az̥jam mesɔj ̥zelentʃutsi], eat has an unrounded open front vowel /a/, yet when said in isolation, it contains a rounded close-mid central vowel /ɵ/. The word meat [mescɔj] consists of both an unrounded open-mid front vowel /ɛ/ and a rounded open-mid back vowel offglide /ɔj/ when said in the context of the sentence, but the /ɛ/ changes to a close-mid front vowel and the /ɔj/ changes to a nasalized rounded close-mid back vowel /õ/.

In the sentence The tall woman smiles [̥v̥isɔkətə ̥ᴣənæsjo smik̥va], the voiced labiodental fricative in tall [visɔk] is devoiced, but when said in isolation, the /v/ is voiced. The unrounded front vowel, the ash /æ/ in woman is an unrounded open vowel /a/ when said in isolation.

Said in isolation, the word forest [gɤɾa] contains an Unrounded Close-mid Back vowel /ɤ/, but in the context of the sentence The forest is dark, the vowel becomes rounded in the same position /o/.

In the phrase five white fish [pɛt beli ɾibi], the consonant vowel consonant sequence [eli] in the word white [beli] is an unrounded open front onglide /ja/ followed by an alveolar lateral approximate /l/ when spoken in isolation. The word fish [bɾibi] has a voiced labiodental plosive /b/ as the word-initial sound, but it is missing entirely when fish is said in the context of the phrase. Also, in isolation, the unrounded tense close front vowel /i/ in fish is a schwa /ə/.


Charkova, Krassimira. “Phonology Part Two Above the Segmental Level.” PowerPoint             presentation. Southern Illinois University at Carbondale Linguistics Department, IL. 23 Sept 2009.

Comrie, Bernard. 1990. The World’s Major Languages. Oxford, England: Oxford University Press, Inc..

Appendix A

Bulgarian Tanscription            Stress

baby: [ˡbebe]                                     2 Syllables/1st Syllable

bed: [ˡle̥gɫɤ]                                    2 Syllables/1st Syllable

bird: [ˡptitʃkɐ]                                    2 Syllables/1st Syllable

black: [ˡtʃjeɾṇ]                                    2 Syllables/1st Syllable

book: [ˡkʰnigʌ]                                    2 Syllables/1st Syllable

brother: [bɾat]   rhotic                        1 Syllable/1st Syllable

building: [ˡsgɾadə] rhotic            2 Syllables/1st Syllable

clothing: [ˡɔbleklo]                        3 Syllables/1st Syllable

cry: [ˡbɫatʃjə]                                    2 Syllables/1st Syllable

drink: [pjə]                                    1 Syllable/1st Syllable

earth: [̥ˡzImja]                                    2 Syllables/1st Syllable

eat: [jɵm]                                    1 Syllable/1st Syllable

fish: [ˡbribə] rhotic                        2 Syllables/1st Syllable

five: [pɛt]                                    1 Syllable/1st Syllable

forest: [ˡgɤɾa] rhotic                        2 Syllables/1st Syllable

friend: [pɾjaˡtIl]                        3 Syllables/3rd Syllable

grain: [̥zɛrˡno]                                    2 Syllables/2nd Syllable

head: [gləˡva]                                    2 Syllables/2nd Syllable

house: [kʰəˡʃtɤ]                        2 Syllables/2nd Syllable

human: [tʃoˡveʃki]                        3 Syllables/2nd Syllable

language: [e̥zˡik]                        2 Syllables/2nd Syllable

leader: [ˡliɾɛr]                                    2 Syllables/1st Syllable

man: [mʊ̥ᴣ]                                    1 Syllable/1st Syllable

many: [ˡnɔgo]                                    2 Syllables/1st Syllable

meat: [ˡmɛsõ]                                    2 Syllables/1st Syllable

grain: [̥zɛrˡno]                                    2 Syllables/2nd Syllable

mountain: [pɫəˡnina]                        3 Syllables/2nd Syllable

old: [star]                                    1 Syllable/1st Syllable

one: [e̥dˡin]                                    2 Syllables/2nd Syllable

red: [ˡtʃeɾəvṇ]                                    3 Syllables/1st Syllable

run: [ˡbjagṃ]                                    2 Syllables/1st Syllable

sea: [mɔˡɾe] rhotic                        2 Syllables/2nd Syllable

short: [kəs]                                    1 Syllable/1st Syllable

sister: [sIsˡtra] rhotic                        2 Syllables/2nd Syllable

sit: [ˡsjadṃ]                                    2 Syllables/1st Syllable

sky: [ˡnebe]                                    2 Syllables/1st Syllable

sleep: [spja]                                    1 Syllable/1st Syllable

smile: [gsˡmifka]                        2 Syllables/1st Syllable

sun: [slṇˡtse]                                    2 Syllables/2nd Syllable

tall: [visˡɔk]                                    2 Syllables/2nd Syllable

ten: [dɛˡsIt]                                    2 Syllables/2nd Syllable

time: [vɾɛˡme] rhotic                        2 Syllables/2nd Syllable

tree: [̥dʊɾˡvo]                                     2 Syllables/2nd Syllable

two: [dvej]                                     1 Syllable/1st Syllable

vegetable: [ˡzɛlẽntʃok]                        3 Syllables/1st Syllable

walk: [rəsˡhɤ̥ᴣdṃsej]                        4 Syllables/2nd Syllable

water: [ˡ̥voda]                                    2 Syllables/1st Syllable

white: [bjal]                                    1 Syllable/1st syllable

woman: [̥ᴣejˡna]                        2 Syllables/2nd Syllable

young: [mlat]                                    1 Syllable/1st Syllable

How tall is the mountain?: [koku visokI plenInatə]

Eat and drink: [jaʃ i pi]

What language do you speak?: [kəkə̥vIzi  gvoɾʃ]

The red book is on the bed: [tʃeɾvɛnətə knig enaleglɤt]

Two birds: [d̥vɛ ptitʃki]

A sleeping baby: [spjaʃtʊ bebe]

I eat meat and vegetables: [az̥jam mesɔj ̥zelentʃutsi]

The woman smiles: [v̥isɔkətə ̥ᴣənæsjo smik̥va]

The forest is dark: [goɾatə hetəmnə] rhotic

Five white fish: [pɛt beli ɾibi]

Appendix B



Bilabial Labiodental Dental Alveolar Postalveolar Retroflex Palatal Velar Uvular Pharyngeal Glottal
Plosive p     b t            d k ɡ
Nasal m n
Trill r
Tap/Flap ɾ
Fricative f v s     z ʃ         ᴣ h
LateralFricative ɬ
Approximant j
LateralApproximant l

When symbols appear in pairs, the one to the right represents a voiced consonant.

Appendix C

Vowel Lengths

baby: [bebe]             e (.1146)   e (.2457)

bed: [lekɫɤ]               e (.1245)   ɤ (.2217)

bird: [ptitʃkɐ]            i (.0917)  ɐ (.2246)

black: [tʃjeɾṇ]         je (.2459)

book: [kʰnigʌ]         i (.0917)   ʌ (.2246)

brother: [bɾat]         a (.1692)

building: [sgɾadə]  a (.1892)   ə (.2216)

clothing: [ɔbleklo] ɔ (.1020)   e (.0777)  o (.2419)

cry: [bɫatʃjə]             a (.1420)  je (.1871)

drink: [pjə]              jə (.3941)

earth: [̥zImja]          I (.1350)  ja (.2580)

eat: [jɵm]                jɵ (.1374)

fish: [bribə]              i (.1490)     ə (.1719)

five: [pɛt]                   ɛ (.1789)

forest: [gɤɾa]            ɤ (.1454)     a (.2988)

friend: [pɾja]          ja (.2346)

grain: [̥zɛrno]          ɛ (.1322)      o (.1459)

head: [gləva]           ə (.1333)      a (.1544)

house: [kʰəʃtɤ]         ə (.1110)      ɤ (.1768)

human: [tʃoveʃki]   o (.1269)     e (.0829)   i (.1974)

language: [e̥zik]      e (.1619)      i (.1328)

leader: [liɾɛr]             i (.1062)     ɛ (.1576)

man: [mʊ̥ᴣ]               ʊ (.2072)

many: [nɔgo]            ɔ (.1545)     o (.1624)

meat: [mɛso]             ɛ (.0809)    õ (.2551)

moon: [luna]            u (.1995)     a (.2495)

mountain: [pɫənina]  ə (.1131)   i (.1062)    a (.2447)

old: [star]                     a (.1903)

one: [e̥din]                    e (.1436)   i (.1381)

red: [tʃeɾəvṇ]                 e (.1442)   ə (.0879)

run: [bjagṃ]                ja (.1691)   ə (.0797)

sea: [mɔɾe]                    ɔ (.1580)   e (.2445)

short: [kəs]                   ə (.0959)

sister: [sIstra]              I (.0957)    a (.2340)

sit: [sjadṃ]                  ja (.1878)

sky: [nebe]                     e (.1184)    e (.2330)

sleep: [spja]                 ja (.2717)

smile: [gsmifka]            i (.0868)   a (.2112)

sun: [slṇtse]                    e (.2693)

tall: [visɔk]                      i (.0559)   ɔ (.1635)

ten: [dɛsIt]                       ɛ (.1132)    I (.0842)

time: [vɾɛme]                   ɛ (.1653)   e (.2224)

tree: [̥dʊɾvo]                   ʊ (.0582)   o (.2161)

two: [dvej]                     ej (.2626)

vegetable: [zɛlẽntʃok]   ɛ (.0974)   ẽ (.1000)   o (.1654)

walk: [rəshɤ̥ᴣdmsej]    ə (.0429)    ɤ (.1382)   ej (.2560)

water: [̥voda]                 o (.1302)    a (.1985)

white: [bjal]                 ja (.2301)

woman: [ᴣejna]          ej (.1512)      a (.2421)

young: [mlat]               a (.1495)

How tall is the mountain?: [koku visokI plenInatə]   o (.1471)   u (.0998)   i (.0413)   o (.1374)   I (.1133)

ə (.0566)   I (.0794)   a (.1309)   ə (.1169)

Eat and drink: [jaʃ i pi]                 a (.1608)   i (.1335)   i (.3078)

What language do you speak?: [kəkəv̥Izi  gvoɾʃ]           ə (.0510)   ə (.0575)

The red book is on the bed: [tʃeɾvɛnətə knig enaleglɤt]                   e (.0879)   ɛ (.0686)   ə (.0510)   ə (.0513)

i (.1024)   e (.0879)   a (.0929)   e (.0665)   ɤ (.1227)

Two birds: [d̥vɛ ptitʃki]                  ɛ (.1260)   i (.0634)   i (.0790)

A sleeping baby: [spjaʃtʊ bebe] ja (.1573)   ʊ (.0415)

I eat meat and vegetables: [az̥jam mesɔj ̥zelentʃutsi] a (..2132)   ja (.2560)   e (.0528)   ɔj (.2185)

e (.0812)   e (.0840)   u (.0961)   i (.0503)

The woman smiles: [v̥isɔkətə ̥ᴣənæsjo smikv̥a]           i (.0279)   ɔ (.1176)   ə (.0422)   ə (.0667)   ə (.0644)

æ (.1329)   jo (.1204)   i (.0696)   a (.1423)

The forest is dark: [goɾatə hetəmnə]   o (.1059)   a (.1508)   ə (.2312)   e (.1250)   ə (.1013)   ə (.1009)

Five white fish: [pɛt beli ɾibi]                 ɛ (.1309)   e (.1005)   i (.1049)     i (.1199)    i (.1269)

Lengths of /i/

.1115 = average length of /i/ in words in isolation

.0844 = average Length of /i/ in words in context

.0982 = total average length of /i/

.1064 = average length of /i/ as an initial word medial sound

.1974 = average length of /i/ as a word final sound

Lengths of /e/

.1679 = average length of /e/ in words in isolation

.0896 = average length of /e/ in words in context

.1381 = total average length of /e/

.1141 = average length of /e/ as a word medial sound

.1527 = average length of /e/ as an word initial sound

.2429 = average length of /e/ as a word final sound

.2429 = average length of /e/ as a word final sound

Lengths of /ɛ/

.1322 = average length of /ɛ/ in words in isolation

.1123 = average Length of /ɛ/ in words in context

.1237  = total average length of /ɛ/

Lengths of /a/

.2102= average length of /a/ in words in isolation

.1484 = average length of /a/ in words in context

.1884 = total average length of /a/

.1745 = average length of /a/ as a word medial sound

.2291 = average length of /a/ as a word final sound

Lengths of /ə/

.1269 = average  length of /ə/ in words in isolation

.0809 = average length of /ə/ in words in context

.1018 =    total average length of /a/

.0948 = average length of /ə/ as a word medial sound

.1967 = average length of /ə/ as a word final sound

Lengths of /ɤ/

.1705 = average length of /ɤ/ in words in isolation

.1227 = average Length of /ɤ/ in words in context

.1609 = total average length of /ɤ/

.1418 = average length of /ɤ/ as a word medial sound

.1993 = average length of /ɤ/ as a word final sound

Lengths of /o/

.1804 = average length of /o/ in words in isolation

.1246 = average length of /o/ in words in context

.1615 = total average length of /o/

.1461 = average length of /o/ as a word medial sound

.1919 = average length of /o/ as a word final sound

Lengths of /ja/

.2207 = average length of /ja/ in words in isolation

.2067 = average length of /ja/ in words in context

.2172 = total average length of /ja/

.1956 = average length of /ja/ as a word medial sound

.2547 = average length of /ja/ as a word final sound

Appendix D

Onglides and Offglides

/ja/ earth: [̥zImja]     Word Final

friend: [pɾjatIl]            Word Medial

run: [bjagṃ]                 Word Medial

sit: [sjadṃ]                   Word Medial

sleep: [spja]                 Word Final

white: [bjaᶥ]                 Word Medial

/je/ black: [tʃjeɾṇ]     Word Medial

cry: [bɫatʃjə]                Word Final

drink: [pjə]                  Word Final

/ej/ two: [dvej]          Word Final

walk: [rəshɤ̥ᴣdmsej]  Word Final

woman: [̥ᴣejna]           Word Medial

/jɵ/ eat: [jɵm]              Word Initial

/ɔj/ I eat meat and vegetables: [ḁzjam mesɔj ̥zelentʃutsi]

Appendix E

Syllabic Consonants

black: [tʃjeɾṇ]                        n            word final

red: [tʃeɾəvṇ]                        n            word final

run: [bjagṃ]                        m            word final

sit: [sjadṃ]                        m            word final

sun: [slṇtse]                        n            word medial

walk: [rəsˡhɤ̥ᴣdṃsej]            m            word medial

Appendix F

Words with an Alveolar Tap followed by a Rhotic Vowel

brother: [bɾat]   rhotic                                    /a/

building: [sgɾadə] rhotic                        /a/

fish: [bribə]             rhotic                                    /i/

forest: [gɤɾa] rhotic                                    /a/

sea: [mɔɾe] rhotic                                    /e/

sister: [sIstra] rhotic                                    /a/

time: [vɾɛme] rhotic                                    /ɛ/

The forest is dark: [goɾatə hetəmnə]             /a/

Words with an Alveolar Tap followed by a Non-Rhotic Vowel

friend: [pɾjatIl]                                    /ja/

leader: [liɾɛr]                                                /ɛ/

red: [tʃeɾəvṇ]                                                /ə/

Five white fish: [pɛt beli ɾibi]                        /i/

Appendix G

Voiceless Consonants

bed: [le̥gɫɤ]                        g

earth: [̥zImja]                        z

grain: [̥zɛrno]                        z

language: [e̥zik]            z

man: [mʊ̥ᴣ]                        ᴣ

one: [e̥din]                        d

tree: [̥dʊɾvo]                         d

walk: [rəshɤ̥ᴣdmsej]            ᴣ

water: [̥voda]                        v

woman: [̥ᴣejna]            ᴣ

Appendix H

Syllable Structure

baby: [ˡbebe] CV‧CV

bed: [ˡle̥gɫɤ] CV‧CCV

bird: [ˡptitʃkɐ] CCV‧CCV

black: [tʃjeɾṇ] CV‧CC

book: [ˡkʰnigʌ] CCV‧CV

brother: [bɾat] CCVC

building: [ˡsgɾadə] CCCV‧CV

clothing: [ˡɔbleklo] V‧CCV‧CCV

cry: [ˡbɫatʃjə] CCV‧CV

drink: [pjə] CV

earth: [ˡ̥zImja] CVC‧V

eat: [jɵm] CVC

fish: [ˡbribə] CV‧CV

five: [pɛt] CVC

forest: [ˡgɤɾa] CV‧CV

friend: [pɾjaˡtIl] CCV‧CVC

grain: [̥zɛrˡno] CVC‧CV

head: [gləva] CCV‧CV

house: [kʰəʃtɤ] CVC‧CV

human: [tʃoveʃki] CV‧CV‧CCV

language: [e̥zik] VC‧VC

leader: [liɾɛɾ] CVC‧VC

man: [mʊ̥ᴣ] CVC

many: [nɔgo] CV‧CV

meat: [mɛso] CV‧CV

moon: [luna] CV‧CV

mountain: [pɫənina] CCV‧CV‧CV

old: [staɾ] CCVC

one: [e̥din] VC‧VC

red: [tʃeɾəvṇ] CV‧CV‧CVC

run: [bjagṃ] CV‧CC

sea: [mɔɾe] CV‧CV

short: [kəs] CVC

sister: [sIstra] CV‧CCCV

sit: [sjadṃ] CV‧CVC

sky: [nebe] CV‧CV

sleep: [spja] CCV

smile: [gsmifka] CCCV‧CCV

sun: [slṇtse] CCVC‧CCV

tall: [visɔk] CV‧CVC

ten: [dɛsIt] CV‧CVC

time: [vɾɛme] CCV‧CV

tree: [dɛɾvo] CV‧CCV

two: [dvej] CCV

vegetable: [zɛlẽntʃok] CV‧CVC‧CVC

to walk: [rəshɤ̥ᴣdmsej] CVC‧VCC‧CCV

water: [̥voda] CV‧CV

white: [bjal] CVC

woman: [ᴣejna] CV‧CV

young: [mlat] CCVC

How tall is the mountai[koku visokI plenInatə] CV‧CV CV‧CV‧CV CCV‧CV‧CV‧CV

Eat and drink: [jaʃ i pi] VC V CV

What language do you speak?: [kəkəv̥Izi  gvoɾʃ] CV‧CV‧CV‧CV CCV‧CC

The red book is on [tʃeɾvɛnətəknig enaleglɤt] CV‧CCV‧CV‧CV‧CCVC VC‧VC‧VC‧CVC

Two birds: [d̥vɛ ptitʃki] CCV CCV‧CCV

A sleeping baby: [spjaʃtʊ bebe] CCV‧CCV CV‧CV

I eat meat and vegetables: [az̥jam mesɔj ̥zelentʃutsi] VC‧VC CV‧CV CV‧CVC‧CV‧CCV

The woman smiles: [v̥isɔkətə ̥ᴣənæsjomikv̥a] CV‧CV‧CV‧CV CV‧CVC‧V‧CVC‧CV

The forest is dark: [goɾatə hetəmnə] CV‧CV‧CV CV‧CVC‧CV

Five white fish: [pɛt beli ɾibi] CVC CV‧CV CV‧CV

Appendix I

Bulgarian Words in Context Versus in Isolation

English Gloss How tall is the mountain?

Context [koku visokI plenInatə]

Isolation [visɔk][pɫənina]

English Gloss Eat and drink

Context [jaʃ i pi]

Isolation [jɵm][pje]

English Gloss What language do you speak?

Context [kəkə̥vzIzi  gvoɾʃ]

Isolation [e̥zik]

English Gloss The red book is on the table.

Context [tʃeɾvɛnətə knig enaleglɤt]

Isolation [tʃjeɾəvṇ]  [kʰnigʌ][lekɫo]

English Gloss Two birds

Context [d̥vɛ ptitʃki]

Isolation [dvej][ptitʃkɐ]

English Gloss A sleeping baby

Context [spjaʃtʊ bebe]

Isolation [spja]   [bebe]

English Gloss I eat meat and vegetables

Context [az̥jam mesɔj ̥zelentʃutsi]

Isolation [jɵm][mɛsõ][zelẽntʃok]

English Gloss The woman smiles

Context [v̥isɔkətə ̥ᴣənæsjo smik̥va]

Isolation [visɔk]   [ᴣejna]  [gsmifkə]

English Gloss The forest is dark

Context [goɾatə hetəmnə]

Isolation [gɤɾa]

English Gloss Five white fish

Context [pɛt beli ɾibi]

Isolation [pɛt bjaᶥ bribə]

Contrastive Dialects: Australian English versus American English

Contrastive Dialects: Australian English versus American English

The accents of four Australian English speech consultants were analyzed both subjectively via listening and objectively according to spectographic data of all four consultants. The consultants were from four different cities in Australia, though they were all concentrated in the south east quadrant of  the continent known as New South Wales. Despite the speakers’ closeness in proximity, some dynamics of their accents, primarily the consonants  differed. Their vowels were a bit more unified though. The consultant’s speech will be analyzed on how their pronunciation differs from American English and from each other.

1.0 Australian Accent Consultants

The consultant’s speech files and biography information were retrieved from the Speech Accent Archive. http://www.accent.gmu.edu. The speakers used as consultants were English 48, English 230, English 294, and English 298. All were male except English 294. The males were ages 45, 35, and 22 respectively. The female was the youngest, age 19. All had acquired English as their native language. English 84 spoke tok pisin in addition to English. English 294 spoke Indonesian French as an L2. English 230 and 298 spoke no other languages besides English.

2.0 Consonants

2.1 Devoicing

Like most speakers of any language, decvoicing of obstruents seems to occur primarily word finally due to the sonority curve of syllables, allowing obstruents that  are syllable initial to be more sonorous than the same obstruent syllable finally. In the data, all consultants devoice, some more so than others. Right from the start, we have devoicing occurring in two of the four speakers in “Please” /pliz/, where the word final /z/ is devoiced to [s] for English 230 and 298, and again in “these” /ðiz/ of line 2, English 230 and English 298 devoice word final /z/ to [s]. In the third line, English 294 devoices /z/ to [s] in “spoons” /spunz/. All four accent consultants devoice word final /v/ to [f] in “five” /fajv/ of line 4. All four consultants also devoice final /b/ of “Bob” /bɑb/ to [p]. In line 7, English 298 devoices word final /g/ to [k] in “frog” /fɹɑg/. English 230 and English 294 both devoice the alveolar plosive /d/ in “kids” /kɪdz/ to [t], causing the normally voiced alveolar fricative /z/ to become devoiced as well. English 298 was the only accent consultant to devoice final /z/ in “things” /θiŋz/ of line 8 to [s]. All four accent consultants devoice word final /z/ to [s] in “bags” /bægz/ of line 8.

Devoicing of final obstruents is occurring with [+Dorsal] and [+Labial] obstruents. However, more often than not, it’s the word final alveolar fricatives that seem to vary most in voicing among consultants. This is probably due to English’s using the alveolar fricative to be the sound that most ends words because of its use as plural, possession, and third-person singular.

Devoicing word initially is rarer linguistically, and definitely rarer in the speech of these four Australian accent consultants. It occurs once in the word “big” /bɪg/ for English 230 and English 294, devoicing word initial /b/ to [p].

2.2 Voicing

Voicing of obstruents is rarer linguistically than devoicing them due to the low sonority rank of these sounds, however, voicing still occurred a few times with these consultants’ speech. In “five” /fajv/ of line 4, English 230 voices word initial labiodental fricative /f/ to [v].

Word medial voicing is rare unless the obstruent is intervocalic. There’s only one instance of word medial voicing with two consultants, but they both do the same thing which could constitute the word medial voicing as word initial voicing. In the phrase “six spoons” /sɪks spunz/ both English 84 and 298 delete word initial /s/ of “spoons” since “six” ends in a voiceless alveolar fricative, and instead of aspirating the now word inital /p/, both consultants voice /p/.

2.3 Flapping

Flapping in Australian English occurs in a similar context as in American English, which is between a stressed and unstressed vowels. In Australian English, it can just occur between two unstressed vowels. The first doesn’t have to be stressed. It can be, but it isn’t a necessity to allow alveolar plosive flapping. In line 2, English 84, 230, and 298 all flap /t/ in “to” /tu/ of the phrase “with her to the store” /wɪθ həɹ tu ðə stɔɹ/. Well, how can this be? It’s occuring between a retroflex liquid /ɹ/ and an unstressed syllable. Rogers (2000) states, rhotic consonants are missing entirely in Australian English pronunciation. I found data to contradict this statement, but Roger’s (2000) observance is certainly true of the rhotic consonant in “her.” More on retroflex liquid deletion can be found in section 2.4. Accent consultants English 84, English 230, and English 298 all delete the retroflex liquid /ɹ/ in “her,” causing their /t/ in “to” to become flapped. English 294 has an apparent /ɹ/ when she speaks “her,” and she doesn’t flap her /t/ in “to.” Instead, she replaces the /t/ with an alveolar nasal /n/. Nasals have an interesting way of replacing alveolar plosives and interdental fricatives in Australian English. More on this replacement in section 2.5.

Flapping occurs again in line 6 with English 230, English 294, and English 298 with the final alveolar plosive /d/ of “need” in the phrase “need a small plastic snake.” In this instance, the flapping occurs after a stressed vowel and before an unstressed syllable. This environment is likely to flap /d/ in American English. Flapping again occurs with English 298 in line 7 with the word final alveolar plosive /d/ in “and” ænd/ in the phrase “and a toy frog” /ænd ə tʰɔj fɹɑg/. The consultant deletes the alveolar nasal, placing the alveolar plosive /t/ between two unstressed vowels. The last instance of flapping that occurs is English 84 with word final alveolar plosive /t/ in “meet” /mit/ in the phrase “go meet her wenesday” /go mit həɹ wɛnzdej/ The consultant deletes the glottal fricative /h/, placing /t/ between two vowels.

2.4 Retroflex Liquid Deletion

Contrary to what Rogers (2000) states, based on the pronunciation of these four consultants’ speech, Australian English is not a totally /ɹ/-less language. The retroflex does delete in one founded primary context: in function words. The only function words that contain a /ɹ/ are “for” and “her,” which occur multiple times throughout the data. This is not a universal phenomenon though, but does happen enough to mention. In line 2, every consultant deletes the retroflex liquid /ɹ/ in “her.” In line 5, English 294 deletes /ɹ/ on “for,” but keeps the liquid when saying “her,” which in line 2, she had previously deleted in the same word. Therefore the phrase “for her” /foɹ həɹ/ sounds like [fə həɹ]. Here, English 294 keeps the voiceless glottal fricative /h/ on “her.” The other three consultants actually delete the word “her” altogether in the phrase “for her” because of the similar sequence of sounds of “for” and “her.” We often do this reduction in American English, so it is not exclusive to just Australian English. To compensate for the “her” deletion, English 84 and English 298 lengthen their vowel, which in this case is /ə/. This reveals a careful pronunciation of the phrase “for her” by English 294. Recall from section 1.0, English 294 was the only female consultant. More on this in section 6.0. English 298 also deletes the retroflex liquid in line 2 in the function word “from,” while all the consultants keep it.

Most other content words that have the liquid in American English keep the sound. Content words where all four consultants keep the liquid sound include “bring” /bɹiŋ/ in line 2, “fresh” /fɹɛʃ/ in line 3, “brother” /bɹʌðəɹ/ in line 5, “frog” /fɹɑg/ in line 7, “three” /θɹi/ and “red” /ɹɛd/ in line 8, and “train” /tɹejn/ in line 9. The one content word where all four consultants delete the retroflex liquid, leaving a rhotic vowel word finally is “store” /stɔɹ/ in line 2. This could be due to the fact that /ɹ/ is word final. In no other content word is /ɹ/ word final except “brother” in line 5. There, all four consultants delete the retroflex liquid and leave a rhotic vowel, just like in “store.”

2.5 Nasal replacements of Interdental Fricatives

The replacement of interdental fricatives with nasal consonants is not universal and does not happen in the same places with every consultant. Really, we believe it is a result of rapid speech, since it happens in areas where the speakers are being particularly careless about the pace of their speech. Much of the time, the nasal replacement seems to be an assimilation process because the word initial replacement follows a word final nasal consonant. For example, English 84 replaces voiced interdental fricative /ð/ with alveolar nasal /n/ following word final velar nasal /ŋ/ in the phrase “bring these.” /bɹiŋ ðiz/ In this case, the only element assimilating is the manner of articulation since the place of the two nasal consonants are in two different areas of the mouth. In this environment, English 294 and English 298 do not replace the interdental with anything. English 230, however, replaces /ð/ with a dental /t/. There is one instance in line 2 where all replace /ð/ with a nasal, but not all the same nasal consonant or in the same environment. In the phrase, “from the store,” /fɹʌm ðə stɔɹ/ English 294 and English 298 both replace /ð/ with an alveolar nasal: [fɹəm nə stoʷ]. English 230 replaces the fricative with an alveolar nasal too, but doesn’t leave much of a vowel following it, therefore replacing the whole word with a syllabic /n/. English 84 does something a little more predictable. He replaces the voiced interdental fricative with a bilabial nasal /m/, a total progressive assimilation process, influenced by the word final /m/ in “from.” Nasal replacement of interdental fricatives occurs nowhere else in the data.

2.6 Other Consonantal Differences

In line 2, the word “with” differs in pronunciation among consultants. English 294 and English 298 both pronounce the coda consonant as a voiced interdental fricative. This differs from American English, since it’s voiceless in American English. English 84 pronounces the coda with an American English pronunciation, cutting off the voicing when the start of the pronunciation of the fricative begins. English 230 actually replaces the fricative with an unreleased voiced alveolar plosive /d/.

All male consultants (English 84, English 230, English 298) delete /s/ when preceded by /ʃ/ in line 3 of the phrase “fresh snow peas” /fɹɛʃ snoʷ piz/. English 294 has a much more careful pronunciation of this phrase and she keeps both the alveolar fricative and the palato-alveolar fricative. English 84 and English 298 both delete the initial /s/ of “spoons” /spunz/ in the phrase “six spoons,” /siks spunz/. Because “six” ends in a final alveolar fricative, both consultants delete it initially and voice the now word initail /p/ to [b]. English 230 also deletes word initial /s/, but instead of voicing /p/, he aspirates it since it is now word initial. English 294 is the only consultant to keep both /s/ sounds.

All the male consultants again do something else that English 294 does not do, something that is common of American English speech, they do not release a plosive when it is followed by another plosive, as evidenced in the phrase “big toy frog,” /bɪg tʰɔj fɹɑg/ of line 7. English 294 is the only consultant careful enough in her pronunciation to release the word final /g/ of “big” before the word inital /t/ of “toy.

English 230 in line 7, instead of ending the word “frog” with a voiced velar plosive /g/ like the orthography entails, he ends the word with a velar nasal /ŋ/. This is very interesting since there is no plausible assimilation process to account for this change, and all the other consultants correctly pronounce the word final plosive.

In line 10 with the word “can” in the phrase “she can scoop,” all four consultants do something differently, as shown in table 1.


She can scoop” /ʃi kən skup/
English 84 [ʃik skuwp]
English 230 [ʃi kə skup]
English 294 [ʃij kən skup]
English 298 [ʃəʔ skʊp]

English 84 simply keeps the word inital /k/ and deletes the nucleus and coda, attaching the velar plosive as the coda to the normally open syllable “she:” [ʃik skuwp]. English 230 only deletes the nasal coda: [ʃi kə skup]. English 294 says it exactly how it would be in narrow transcription, a careful pronunciation. English 298 actually deletes the entire word, probably replacing the velar plosive /k/ with the glottal stop [ʔ]. Also in line 8, English 84 and English 230 both reduce the word “into” /ɪntu/ to /nə/, deleting the initial vowel, and medial plosive /t/ and reducing the rounded back vowel /u/ to unstressed /ə/.

3.0 Vowels

3.1 Vowel Raising

The examples put forth in this section are classified as vowel raising. That is, a vowel in a word of one to all four of the accent consultants is inherently higher than a Standard American English vowel in the same position of the same word. Also, there will not be separate sections for diphthongization or monothongization, they will instead be included in sections 3.1 -3.2.

English 84 produces the same vowel as American English in “store” of line 2, while English 230, English 294, and English 298 produce a higher vowel /o/, [sto]. All four consultants produce /sɪks/ as [siks]. We will classify this as a higher vowel, even though the real distinction between /i/ and /ɪ/ is tenseness. All consultants produce a higher vowel in “fresh” /fɹɛʃ/ of line 2. English 84 produces /ɪ/, [fɹɪʃ], English 230 and English 298 produce a slightly higher vowel /e/, [fɹeʃ]. English 294 diphthongizes the slightly higher /e/, [fɹejʃ]. English 230 produces /æ/ in /snæk/ as a slightly higher /ɛ/, [snɛk]. English 84 realizes an even higher, diphthongized vowel in snack: /ej/ [snejk], which sounds very much like American English “snake.” However, all four consultants produce a low/back to high/front/ diphthong /aj/ for “snake” /snek/, [snajk].

Again, in Standard American speech the lax /ɪ/ is produced as tense /i/, as exemplified by the word “kids” /kɪdz/ in line 7, [kʰits] – as pronounced by English 84, English 230, and English 294. English 298 unstresses the vowel, [kəds]. There is one other case of tense/lax difference in “scoop” /skup/ of line 8. English 298 realizes [skʊp].

3.2 Vowel Lowering

English 84 pronounces “please” /pʰliz/ in line 1 as [pʰlejz]. He lowers and diphthongizes /i/ to /ej/. A word where all consultants produce a different vowel than American English is in “bring” /bɹiŋ/ in line 2. English 84, English 294, and English 298 produce a lower vowel /ɛ/, realizing “bring” as [bɹɛŋ]. English 230 just unstresses the /ɛ/ and produces [bɹəŋ]. All four consultants produce the same vowel in “cheese” /tʃiz/ of line 4. They all produce a lower diphthongized vowel /ej/, [tʃej(s/z)].

All four consultants realize “three” /θɹij/ with a lower diphthong /ej/ [θɹej]. English 84 creates a diphthong for the word “go” /go/ in line 9. He realizes the vowel as diphthong /aʊ/ [gaʊ].

4.0 Creaky Voice

There were only three instances of creaky voice in the data. The first is by English 294 who pronounces the /ɛ/ in “stella” /stɛlə/ of line 1 with creaky voice. The other two were produced by English 230 with the vowel /a/ in “call” from line 2 and /ɑ/ in “frog” from line 7.

5.0 Limitations

The limitations of this contrastive dialect research were three-fold. The first is that not all speakers were from the same city in Australia. They were, however, from the same region, though there still seemed to be quite a bit of variation in their speech. Second is the instrument. It is a good tool for all dialects of English and for speakers with an L1 that is not English to discover pronunciation errors and differences, but now that we’ve been able to identify some of the different linguistic features between American English and Australian English, a new instrument should be created to target more of these specific differences. If this instrument were spoken by a larger group of Australian English speakers, we might be able to perform more in-depth research on specific environments where these linguistic differences occur. The third and final limitation is the American English that was used to compare. The researcher relied on his native perception of how the sounds of American English come together, a more phonemic analysis. For a truly phonetic analysis, the speech of four speakers of American English should have been recorded, transcribed,. and analyzed using the same instrument.

6.0 Areas of Further Research

As stated in section 5.0, an instrument that catered more to the differences of American English and Australian English should be created and used to perform a more in-depth analysis, particularly more function words that contain the retroflex liquid and more content words that that have the liquid as a word final sound.

A second area for possible research would be to analyze and describe how speakers of all different dialects and L1s read the instrument, paying careful attention to the speakers’ pace, his or her intonation, how many times he or she pauses or corrects him or herself, etc. With this kind of analysis many kinds of speaker biography details can be analyzed and compared, such as the speakers gender, fluency, the means of acquistion, age, etc.

7.0 Conclusion

No steadfast conclusions were able to be drawn from this analysis due to the limitations and the fact that the speech of only four consultants were transcribed and considered. However, in short, Australian English differs from American English on a couple accounts. 1) Australian English speakers more readily delete the retroflex liquid /ɹ/ word finally and in function words. 2) Devoicing of word final obstruents is quite common, voicing is less common in Australian English. 3) The voiced and voiceless interdental fricatives are more prone to assimilation than any other fricative in Australian English. 4) Australian English speakers produced slightly higher and slightly more diphthongized vowels than American English Speakers.


Rogers, Henry. (2000). The sounds of language: An introduction to phonetics. Essex: Pearson

Education Limited

Speech Accent Archive: http://www.accent.gmu.edu

Pronunciation of English Interdental Fricatives by French and Spanish L1 Speakers

Pronunciation of English Interdental Fricatives by French and Spanish L1 Speakers

Pronunciation is one of the most complicated aspects of teaching and learning English, and because of this complication, there is much debate in TESOL about the correct instruction of pronunciation for ESL students. ESL teachers, especially those working with oral skills and pronunciation, face a difficult task. Is there a single, correct form or dialect of English that should be taught? Should all English speakers sound like Americans or British? What if EFL students plan to study in Australia or Canada? The question is far more complicated than many English pronunciation instructors admit. Of course, context matters. If international students plan to study at an American university, it behooves them to listen to North American dialects – and make sure that their pronunciation is clear and comprehensible to American listeners. Being audience focused, after all, is part of effective communication and good manners.

Interdental fricatives are notoriously difficult for L2 speakers of English because of the fricatives’ absence in the phonemic inventory of a large majority of other world languages.  I chose Spanish and French speakers for this study. All of the consultants for this research were from either Spain or France. I felt it was important to use only one region for each language. This eliminates speaker variant pronunciation due to their speaking different dialects of the same language. I chose Spanish and French because I have studied both in great detail. I also studied in Pau, France which is less than an hour from the border of Spain. Pau lies in Le Pays Basque or Basque country which extends from southern France to Northern Spain. The number of Spanish to French speakers in this region is considerably close.


In a study to uncover pronunciation similarities and differences of the English interdental fricatives between native French and Spanish speakers, three native French speakers’ and three native Spanish speakers were surveyed.  To find my consultants, I used the website http://accent.gmu.edu/ to find them. The site allows researchers to find speakers of certain language that meet the researcher’s needs. For this study, all four French consultants are native to France, and all of the Spanish consultants are native to Spain.  All of the consultants’ acquisition method is academic, their length of residence in an English speaking country is less than one year, and their ages of English onset is 11 years of age and older.

French Consultants

French consultant 1 is a twenty year old female from St. Laurent D’onay, which is in southeast France, and other than English and French, she speaks Spanish. She was twelve when she began learning English, and her length of residence in the USA is five months. French consultant 2 is a twenty-two year old female from Nice, which is along the French la côté d’azure, in the southeast of France. In addition to English, she also speaks German. She began learning English at the age of eleven, and her length of residence in the USA is two months. French Consultant 3 is a twenty year old male from St. Louis, France, which is the northeast of France near Normandy. He also speaks German and the age of his English onset is twelve.  One month in the U.K. is his length of residence.

Spanish Consultants

Spanish Consultant 1 is a twenty two year old female from Madrid, Spain. In addition to Spanish and English, she also speaks German. Her age of English onset is 14, and her length of residence in the USA is ten months. Spanish Consultant 2 is a twenty-four year old male from Zaragora, Spain which actually lies in le pays basque in northern Spain. His age of English onset is fourteen, and his length of residence in Ireland is a month. Spanish Consultant 3 is a 28 year old male from Cartagena, Spain. Besides Spanish and English, he speaks German. He began learning English at age eleven and his length of residence in the U.K. is 9 months.


I predict that Spanish speakers will pronounce the voiceless interdental fricative /θ/ correctly while the French mispronounce it and replace the sound with a voiceless alveolar fricative /s/. I also think that both the Spanish and French speakers will pronounce the voiced interdental fricative /ð/ incorrectly, since both languages’ phonetic inventory lack the sound.

Data Analysis

The only sound that every speaker pronounced correctly was the voiced interdental fricative /ð/ as a word medial sound in brother. The word brother is the only time that

Cumulatively, the three French speakers pronounced the voiceless interdental fricative /θ/ correctly eight out of fifteen times, or 53.3 % of the time in the data. The three Spanish speakers produced it correctly fourteen out of fifteen times, or 93.3% of the time in the data.  Such high correct pronunciation of the voiceless interdental fricative /θ/ is due to the sound’s presence in the phonetic inventory of Spanish in Spain. However, the voiced interdental fricative /ð/ is not represented in the phonemic inventory. According to Llisterri, the voiceless interdental fricative /θ/ and the voiceless alveolar fricative /s/ have voiced allophones when they are followed by a voiced consonant (2).  The number of times the three Spanish speakers produced the voiced interdental fricative /ð/ correctly is nine out of eighteen times, or sixty percent of the time. The French language lacks both the voiced interdental fricative /ð/ and the voiceless /θ/ in its phonemic inventory. The three French speakers also produced the voiced interdental fricative /ð/ correctly nine out of fifteen times, or sixty percent of the time.

The French speakers incorrectly replaced the voiceless interdental fricative /θ/  twice with a voiceless alveolar fricative /s/, twice with a voiced interdental fricative /ð/, twice with a voiceless labiodental fricative /f/, and once with a voiceless alveolar plosive /t/.  The /s/ matches the /θ/ in place and manner of articulation, and their places of articulation are fairly close in the mouth.  The /ð/ matches the /θ/ in place and manner of articulation. The /f/ matches the /θ/ in manner of articulation, though their places of articulation are close as well.

The only error in voiceless interdental fricative /θ/ pronunciation was made by Spanish Consultant 1. She replaced the interdental with a voiceless alveolar plosive /t/. The /t/ doesn’t have /θ/’s  same place or manner of articulation, but they are both voiceless sounds and their places of articulation are close in the mouth. It’s even closer when the interdental is made dental by some American English speakers, which can happen when the voiceless interdental fricative /θ/ is a word initial or word final sound.

The French speakers replaced the voiced interdental fricative /ð/ twice with a voiced alveolar fricative/z/; four times with a voiced alveolar plosive /d/, two of these were devoiced; and once with an alveolar nasal /n/. The Spanish speakers replaced the voiced interdental fricative /ð/ seven times with a voiced alveolar plosive /d/, and twice with an alveolar nasal /n/. The /z/ matches the /ð/ in manner of articulation, though again, alveolar is close in place of articulation to interdental.

French speaker replacement sounds of /θ/

The only instance of the voiceless interdental fricative /θ/ as a word initial sound is in the word thick . French Consultants 2 and 3 followed the correct pronunciation by an unrounded close front /i/, while Spanish Consultant 1 followed it by the lax vowel /ɪ/.

The voiceless alveolar fricative /s/ replaces the voiceless interdental fricative [θ] twice as a word initial sound by the French speakers. Once with French Consultant 1 in the word three , [sri].  The phoneme replacement occurs after the word into, [ɪntu] ending in the nucleus,  a rounded close back /u/, of the word final consonant. This word final vowel is correctly pronounced. The other time /s/ replaces a /θ/ is with French Consultant 2 with the word things, [siŋs]. The phoneme replacement occurs after the word these, [̥diz], the voiced alveolar fricative /z/ is a correct pronunciation. The voiceless alveolar /s/ replacement here could be due to the voiced alveolar /z/; the two sounds are the identical except in respect to voicing.

The voiced interdental fricative /ð/ replaces the voiceless identical sound twice. In the same word with by French Consultant 2 and 3, pronounced [wɪð]. This pronunciation is not necessarily incorrect though. If these two studied English under a British system, this pronunciation would be considered correct. This pronunciation definitely doesn’t affect intelligibility of the word either, but in reference to correct American English pronunciation, it’s incorrect.

The voiceless labiodental fricative /f/’s replacement of /θ/ occurs twice; with the same word things, pronounced [fiŋz]; and by the same speaker, French Consultant 3. The replacement occurs after the word these, [̥diz], a voiced alveolar fricative /z/ as the word final sound.

Spanish speaker replacement sounds of /θ/

Again, the voiceless interdental fricative /θ/ was incorrectly mispronounced only once, by Spanish Consultant 1 in the word things, pronounced [tiŋgs]. The replacement phoneme follows a voiceless alveolar fricative /s/ in the word these, [dɪs].

French speaker replacement sounds of  /ð/

In addition to /ð/ as a word medial sound in brother, the only interdental sound to be correctly pronounced is as a word initial sound in the, which all three consultants pronounce correctly, following the /ð/ by an unstressed schwa /ə/.

The replacement of the /ð/ by French speakers occurs twice with the voiced alveolar fricative /z/ twice by the same speaker, Spanish Consultant 1 in the same word, these, pronounced [zis] both times. The first follows a correct pronunciation of bring, [bɹĩŋ]; velar nasal /ŋ/ is the word final sound. The second /z/ replacement in these follows the word scoop, which she pronounces [skup˺]. The replacement sound occurs after an unreleased voiceless bilabial plosive /p/. This unreleased plosive is an incorrect pronunciation according to Standard American English Standards. The plosive as a word final sound is aspirated /ʰ/.

The voiced alveolar plosive /d/ replaces /θ/ four times in word initial sound position. Once with French Consultant 1 in the word the, pronounced /̥də/. The voiceless allophone /̥d/ of the phoneme /d/ replacement occurs after the technically incorrect pronunciation of from, [fɹʌm]; a bilabial nasal /m/ as the word final sound. The pronunciation is technically incorrect because a general rule for stressed and unstressed syllables of standard American English is that only lexical words like nouns, verbs, pronouns, and adjectives have stressed syllables; and content words, most other parts of speech, are unstressed. So a correct pronunciation would have an unstressed vowel in the preposition from, [fɹəm]. The second replacing of /θ/ with /d/ is with French Consultant 2 in the word these, [diz] after the word bring, pronounced [brĩŋ]. The last two /d/ replacement of /ð/ in the same word, word position, and environment as French Consultant 2. However, French Consultant 3 also mispronounces the /ð/ in the word the which follows a voiceless alveolar plosive /t/ in the word at, pronounced incorrectly as [ɛt].

Spanish speaker replacement sounds of  /ð/

Replacement of  /ð/ with /d/ occurs nine times, five times by the same speaker, Spanish Consultant 1, who uses /d/ to replace /ð/ in every word except for brother. Spanish Consultant 2 replaces /ð/ with /d/ three times, but all for the same word the, pronounced [də]. The follws from, for, and at, pronounced [fɹʌm], [fɔɹ], and [ət] accordingly; so, pronunciation of the as [də] occurs after a bilabial nasal /m/, an alveolar approximant /ɹ/, and a voiceless alveolar plosive /t/.

Spanish consultant 3 is the only Spanish consultant that replaced /ð/ with an alveolar nasal /n/. It occurs as a word initial position in the word the, /nə/ following a bilabial nasal /m/ in the word from, [fɹʌm].


In total the French consultants incorrectly pronounced the voiceless interdental fricative /θ/ seven times out of fifteen. The replacement of /θ/ with a voiceless alveolar fricative /s/, a voiced interdental fricative /ð/, and a voiceless labiodental fricative /f/ all occur two out of seven times or 28.5% of /θ/ pronunciation errors. /ð/ replaces /θ/ only once, or 14.2% of /θ/ pronunciation errors.

Incorrect pronunciation of the voiced /ð/ interdental fricative by the French speakers occurs seven times out of fifteen. The replacement sounds are the voiced alveolar fricative /d/ which occurs five out of seven times or 71.4% of /ð/ pronunciation errors, and the voiced alveolar fricative /z/ two out of seven times or 28.5% of  /ð/ pronunciation errors.

The Spanish consultants had less of a variety in placement sounds than my French consultants. Replacement of the voiceless interdental fricative occurs once as a voiceless alveolar plosive /t/, or 100% of all errors.

The only other sound that the Spanish consultants replaced /ð/ with other than the voiced alveolar plosive /d/ is the alveolar nasal /n/ only once or ten percent of /ð/ pronunciation errors.


According to this data, both of my predictions for how the Spanish speakers would pronounce the interdental fricatives were correct. The most frequent replacement sound of /ð/ by Spanish speakers is the voiced alveolar plosive /d/, just as I hypothesized. However, my hypothesis for how French speakers would pronounce the interdental fricatives was incorrect. The voiceless alveolar fricative /s/, the voiced interdental fricative /ð/, and the voiceless labiodental fricative /f/ all occur at the same frequency for the French speakers’ data. Also they pronounced /ð/ most often as a voiced alveolar plosive /d/, not the voiced alveolar fricative /z/ as I predicted.

Spanish speakers are more likely to replace both interdental sounds with more consistent sounds than French speakers. Also, I found it interesting that the only sound that all speakers produced regularly was the voiceless interdental fricative /ð/ as a word medial sound in brother. If I were to further research in this field, I would like to investigate ELL’s pronunciation of the interdental fricatives as word medial sounds versus the pronunciation of /ð/ as word initial and finial sounds.





The internal reconstruction of the Nahuatl noun stems, and nominative and genitive morphemes, and a morpheme meaning “place of” are observed. This manuscript hypothesizes that Nahuatl speakers insert an /i/ word initially (prothesis) and finally (paragore) to keep a consonant cluster from being anything other than word medial, as this would violate Nahuatl syllabic structure. The original nominative morpheme is hypothesized as *tl, the original genitive morpheme is hypothesized as *no, and the “place of” morpheme is hypothesized as *tlān. A neutralization of word final /m/ also occurs after an apocope of /i/ in noun stems.




In data from Campbell (2004: 246), the nominative and genitive case morphemes along with a suffix morpheme meaning “place of” are observed. This manuscript hypothesizes the original forms of each morpheme as well as the original stem of the nouns given. It then describes the phonetic changes that occur in some word stems and morphemes. Though Campbell (2004) was not clear on whether the first morpheme, original form /tl/, was nominative, dative, or accusative, this manuscript assumes that it is nominative. Note: /tl/ = tˡ

Nowhere in the data are there consonant clusters in initial or final word position. When speakers have a consonant cluster that begins a noun stem, and no prefix morpheme is added, speakers insert an /i/ to the beginning of the word.  Also, when there is a noun stem that ends in a consonant and a suffix morpheme that is only a consonant sound is added, Nahuatl speakers add an /i/ to the end of the word so that it will not end in a consonant cluster. These two epenthesis processes can be established as prothesis and paragoge. The rule for the former is in (1), and the rule for the latter is in (2).

(1) ⍉ –> i/ #__CC

(2) ⍉ –> i/ CC__#

Data for the phonological rule in (1) is in (3) which is the nominative case of each noun on the left and genitive on the right. The /i/ prothesis only applies when the nominative morpheme is attached. We could expect noun stems that are consonant cluster initial would take /i/ prothesis with the “place of” morpheme since it is also a suffix, but this morpheme never attaches to a consonant cluster initial stem in the data.

(3) ikʃi-tl “foot”                        no-kʃi “my foot

ikni-tl “fellow”                        no-kni “my fellow”

isti-tl “fingernail”                        no-sti “my fingernail”

itʃka -tl “cotton”                        no-tʃka “my cotton”

The data for the phonological rule in (2) is in (4). The /i/ paragoge only applies when the nominative morpheme is attached, since attaching it causes a consonant cluster at the end of the word.

(4) tepos-tli “axe”            no-tepos “my axe”            tepos-tlān “place of axes”

mīl-li “cornfield”            no-mīl “my cornfield”            mīl-tlān “place of cornfields”

1.1 The Morphemes

The original form of the nominative morpheme is *tl, and it attaches as a suffix. When the noun stem has a word final /l/, Nahuatl speakers assimilate the suffix-initial tl in the nominative morpheme. Because this /t/ is lateralized, speakers keep the lateralization, making the attachable morpheme /l/. This is a total assimilation process, demonstrated by the data in (5).

(5) kal-li “house”            tʃīmal-li “tortilla griddle”

This rule is generated in (6).

(6) tl –> l/ l+__

For the /i/ paragoge rule in (2) to remain true, we must assume that /ll/ counts as a word final consonant cluster. So, speakers add a word final /i/.

The original form of the morpheme meaning “place of” is *tlān. It attaches to the word stem as a suffix. It is [tlān] following all vowels and all consonants except /l/. When the morpheme attaches to a stem that has a final /l/, much like the nominative morpheme, it undergoes assimilation by deleting the /t/, making the attachable suffix [lān]. This assimilation process is indicated by the data in (7). The same basic assimilation rule in (6) also applies to these data.

(7) mis-tlān “place of cougars”                        mīl-lān “place of cornfields

āma-tlān “place of paper, fig trees”            tʃimal-lān “place of tortilla griddles”

The original form of the genitive morpheme was more difficult to uncover. This manuscript hypothesizes that the original form is *no. Therefore the morpheme is phonetically irregular when it precedes a noun stem that has an initial vowel. Data where the regular genitive prefix attaches to noun stems that are consonant initial are in (9), and data where the genitive morpheme is irregular attaching to vowels is in (8).

(8) n-ol “my rubber”

n-e “my bean”

n-īʃte “my eye”

(9) no-kni “my fellow”

no-tʃka “my cotton”

no-kal “my house”

The data in (8) undergo the phonological rule in (10).

(10) o –> ⍉/ n__+V

2.1 The Noun Stems

Most of the noun stems in the data stay rather phonetically sound, however six stems apocopate word final /i/ which triggers a neutralization process that changes /m/ –> /n/ in three words. The six data that apocopate word final /i/ are in (11) and (12). The data in (12) are the words that undergo neutralization after this apocope occurs.

(11) kaʃi-tl “bowl”                        no-kaʃ “my bowl”

kʷawi-tl “tree, wood”            no-kʷaw “my tree, wood”

māyi-tl “hand”                        no-māy “my hand”

(12) ʃāmi-tl “brick”                        no-ʃān “my brick”

pāmi-tl “flag”                        no-pān “my flag”

kōmi-tl “jug”                        no-kōn “my jug”

The apocope of /i/ occurs when the genitive morpheme attaches to a noun stem ending in /i/ followed by a single consonant sound. If we look through the data, the only time an /i/ is word final is when it is preceded by a consonant cluster. Therefore, this manuscript hypothesizes that /i/ can only occur word finally after a consonant cluster. Hence, the genitive case data in (11) and (12) undergo the phonological rule in (13).

(13) i –> ⍉/ VC__#

The neutralization process occurs because no other words in the data end with an /m/, so to universalize the sounds allowed at the end of a word, Nahuatl speakers neutralize final /m/ to /n/. The rule for this process is in (14).

(14) m –> n/ __#

As for rule ordering of all the phonological rules in this manuscript, the rule in (1) must occur after the rule in (6), since the rule in (6) is about the attachment of the nominative morpheme, to which speakers will always add a word final /i/ so not to have a word final consonant cluster. Also the rule in (13) must come before the rule in (14), since (13) deletes the word final /i/ to allow the neutralization of /m/ to /n/ word finally.


Campbell, Lyle.  2004.  Historical Linguistics:  An Introduction. 2nd ed.  Cambridge: MIT Press..

What Really Happened to English? Examining the Celtic Hypothesis

What Really Happened to English? Examining the Celtic Hypothesis


Despite linguists’ skepticism that Insular Celtic languages had very little impact on English during their language contact of nearly three-hundred years, it can be shown through substratum and superstratum contact features that the Celtic languages of the British Isles had a rather large impact on the English language. Most linguists believe influence beyond lexical borrowing is highly doubtful because few have looked outside the grammatical scope of Indo European. With help from discussion among some of the leading linguists on the Celtic Hypothesis, we can hypothesize that Celtic had a very real influence not only on English words, but also grammatical structure, syntax, and literature.


Table of Contents

1.1 Introduction

2.1 Substratum Contact Features: Periphrastic Do

2.2 Substratum Contact Features: Verbal Noun, Northern Subject Rule, & Internal Possessor

2.3 Substratum Contact Features: Embedded Inversion

3.1 Superstratum Contact Features

4.1 What Really Happened?

5.1 Mutual Reinforcement Hypothesis

6.1 Conclusion


1.1 Introduction

English is the official language of the United States of America, the United Kingdom, Australia, and about fifty other countries in the world, many of which have other official languages in addition to English. Many English scholars say these other languages that coexist along side English have had very little influence on the language. For example, before European settlers came to North America, all people living on the continent spoke primarily a Native American language. When English appeared there within the last few centuries, the languages were forced to co-exist. Besides a restricted number of nouns (tomahawk, squaw) and place names, Native American languages have not had any influence on English. In most language contact situations, languages influence one another. Sometimes these influences are great such as the Norman French influence on English from the late eleventh century to about 1470, which left behind grammar, phonemes, and a plethora of new words. Other times the influence of one language on another is quite insignificant like the language contact situation between English and Native American languages. Many linguists believe this was true of the language contact situation between Celtic languages and English. Before Anglo Saxons brought their language to the British Isles in the fifth century, Celts lived there, speaking their Brythonic and Goidelic Celtic languages. However, when the Angles, Saxons, and Jutes brought over their Germanic dialect that we call Old English, the languages were forced to coexist on the Isles for nearly three hundred years. Many scholars cite the etymology of English words and discover their root in Celtic languages. For many, belief of linguistic influence stops there. However, there is a select group who believe that the influence was much greater. John McWhorter (2008) argues for the progressive gerund in English having its roots in Celtic languages and Johan Van Der Auwera (2002) both agrees and disagrees with McWhorter’s theories and argues further the Celtic origin of English periphrastic do. Markku Filppula (2000) discovers embedded inversion of modals, nouns, and verbs in some dialects of English have some Celtic syntactical influence. Wolfgang Kuhlwein (1998) believes much Old English poetry was influenced by Celtic Rhetoric. Of course, the influence goes both ways. English influenced Celtic languages by being the power language. Kevin Rottet (2005) argues English’s influence on the Brythonic Celtic language, Welsh and its phrasal verbs. Theo Vennemann (1999) argues for four specific English structural features and discovers these syntactic features are manifestations of Semitic substratum that entered English through Insular Celtic influence. With evidence as overwhelming as these authors’ works, linguists and scholars of English history should force themselves to reexamine Celtic’s rather profound influence on English through extensive language contact.

2.1 Substratum Contact Features: Periphrastic Do

McWhorter (2008) proposes the reason so many linguists have trouble with accepting any kind of theory that says Celtic had a remarkable influence on English is because they are too scientific and unwilling to accept possibilities. Linguists decree, “Where is the textual evidence in Old English? If Celtic languages had any kind of influence on English periphrastic do, why was it not until Middle English, after the Norman invasion in 1066, that English started writing periphrastic do?” And the answer is the installation of do did not happen all at once. McWhorter (2008) hypothesizes that periphrastic dostarted out as vernacular speech. No man was going to write a vernacular with quill and ink on parchment, especially since those who did the writing were educated individuals that most likely spoke a standard form of Old English. We will never be able to see or hear the speech of an illiterate farmer who more than likely used a vernacular form of Old English containing a periphrastic do when he might need to communicate with neighboring Celts. English’s closest relative is Frisian, a relative of Dutch spoken by hundreds of thousands of people in the Netherlands. Bill Bryson (1990) says Angles, one of the first Anglo Saxon invaders, spoke Frisian, which is very similar to Old English. Due to little linguistic alteration in the history of Frisian, most Frisians are able to read Beowulf “almost at sight” (Bryson: 1990: 46). McWhorter (2008) takes such evidence and claims Frisian to be English’s closest relative because it too shed much of Proto-Germanic’s case markings. Frisian had no contact with Celtic languages, so if English hadn’t underwent this Celtic influence so long ago, our question syntax might be similar to Frisian’s in (1).

(1) “Do we eat apples?”

Ite wy appels?

Eat we apples

Van Der Auwera (2002) agrees with McWhorter (2008) and says that of all the Germanic languages, English has had the longest prolonged contact with Celtic languages. She illustrates the appearance of periphrastic do in Early Middle English in (2).

(2) “The bishop of Winchester had them come here.”

ϸe  biscop    of Wincestre … dide heom cumen ϸider

The bishop of Winchester     did  them   come there

Van Der Auwera (2002) discusses periphrastic do‘s use in negative and emphatic assertions in Modern English, but this excerpt shows English used to use periphrastic do in nonemphatic, positive assertions as well. This data is similar to evidence put forth by McWhorter (2008). He shows on page 6 that Welsh uses what he calls “meaningless do” in emphatic, nonemphatic, negative, and positive sentences, and questions:

(3) “Did I open?”            Nes i agor?

“I did not open. Nes i ddim agor.

“I opened. Nes i agor.

Although Welsh has a different word order than English, we can see Welsh uses nes “do” in every type of utterance, affirmative, negative, and question sentences. Middle English was more like Welsh in that sense. Middle English “You wept” would come out something like: Thou dudest wepe.Early Modern English still had this meaningless do in nonemphatic, affirmative statements evidenced in Shakespeare’s plays. The following is Gertrude in Hamlet Act III, scene IV, lines 134-135.

“Alas, how is’t with you

That you do bend your eye on vacancy,

And with th’incorporal air do hold discourse”

Thus, Modern English must have discontinued the use of periphrastic do in affirmative, nonemphatic statements sometime within the last few centuries.

Van Der Auwera (2002) describes other uses of periphrastic do in English besides negative statements and questions. Southwestern American English often uses do in nonemphatic, affirmative statements expressing generic/habitual aspect. She gives the Modern English example “Fred does the washing up.” This data is a case of periphrastic do because it creates a more analytical construction in which the verb following periphrasis is a verbal noun. Van Der Auwera (2002) contends the theory of English’s periphrastic do having Celtic origin would be stronger if there were a cleaner-cut division of periphrasticdo use between Celtic and Germanic languages. She states that colloquial forms of German use their version of periphrastic tun “do” as well as Dutch, which uses its do equivalent “deed.” Germanic languages have use of a verbal noun that is similar to English’s. Modern Dutch expresses “Jan is doing the dishes” as

(4) Jan   doet de     afwas

Jan does the off.washing

Some scholars get distracted because of periphrastic do‘s use in some Germanic language vernacular dialects as well. McWhorter (2008) gives the example in (5).

(5) Er   tut  das schreiben.

He does that write.

However, this do‘s use is entirely optional and adds emphasis to wherever the speaker places it in the sentence. Also Germanic use only sees periphrastic do in the present tense, and we know all too well that English’s do is in the present and past negative and questions, is not optional, always precedes the verb, and doesn’t give emphasis to anything. Some say that English’s use of periphrastic do merely evolved one step further than German’s, but if this were true, why is English’s use of do periphrasis unique among the Germanic languages? With the number of Germanic languages and their many, many dialects, one would think one of the dialects would have sporadically started making do periphrasis mandatory similar to English—none did.

To blur the line between Germanic or Celtic origin of periphrastic do even more, not all Celtic languages have the relevant do. Though, Van der Auwera (2002) does say, “In Celtic ‘do’ periphrasis is very prominent” (288). She lists two subtypes. One where periphrastic do precedes the verbal noun, common in Old Irish, Middle Breton, Middle and Late Cornish, and Middle Welsh whose construction of “I enjoyed myself yesterday” looks like

(5) (mi) wnes i fwynhau ddoe,

(I)  did   I  enjoy   yesterday.

The second type of periphrastic do use in Celtic languages is when the verbal noun comes first proceeded by a relative particle and then periphrastic do. This word order is characteristic of eastern Brythonic (Cornish, Welsh, and Breton) and less common in the Goidelic languages (Gaelic, Manx, Irish).

Van Der Auwera (2002) concedes that there are French examples of periphrastic do as well, however, she says that these uses are “highly specialized” as shown in Do listen to me! Or me faites entender, translating literally as “And yet me do listen” (291). This phrasing can be understood as an emphatic way of telling someone to listen to you. Van Der Auwera (2002) also remarks that causative uses of periphrastic do are not common among Celtic languages, but Latin makes a causative use of facere “to do,” as do its daughter languages.  “My father makes me work” translates to French in (6).

(6) Mon père me  fait travailler

My father me does work.

Van Der Auwera (2002) states there is a large, but not yet complete consensus that periphrastic doin English developed from a causative use. This is pretty damning evidence for the Celtic hypothesis, as “causative uses of ‘do’ verbs in Celtic are rare or nonexistent” (Van Der Auwera: 2002: 290). Early Modern Irish sentence constructions of do + verbal noun were probably prompted by English construction. However, like English, periphrasis is strong in Celtic while it is weaker in non-English Germanic and Romance languages. Causative uses are important in French and Dutch, and Van Der Auwera (2002) claims there has been a contact-instigated influence between French causative doand Dutch’s.

Van Der Auwera (2002) details a certain type of language contact situation that is needed for the development of auxiliaries. In Korean,  words with Chinese and Japanese origins  outnumber native Korean words. A standard strategy for making foreign nouns into Korean predicates is by combining them with hata “do”. The same is true of Hindi do‘s combination with English nominals, as well as Persian with Arabic and Turkish with Dutch. In all language contact situations given, it’s the socially/culturally inferior language that develops this do to verbalize foreign nominals. If a similar situation were true of Celtic and Old English contact, one would not only expect the manifestations of auxiliaries in general, but also for periphrastic do to combine with foreign elements in Celtic. Since this does not help uncover the geneses of English do, it’s safe to say that English periphrastic doresulted from more than just the dynamics of language contact.

Van Der Auwera (2002) discusses the hypothesis that English and Celtic were both givers and receivers. This is what she calls the mutual reinforcement hypothesis, which claims English and Celtic influenced grammatical uses in the other language; therefore, periphrastic do reinforced its use in the other language, i.e., Celtic reinforced English, and English reinforced Celtic periphrasis. She states periphrastic do appears earlier in Welsh texts than in English. English relevant doappeared around the thirteenth century, and Welsh appeared earlier than that, though no date or time span is given. This supports a hypothesis of Welsh having a greater influence on English in the periphrastic do sense. Though Celtic’s earlier use was not used in a systematic way, she notes that do is more established in Middle Welsh than in English of the same time period. And though this does not necessarily mean that Celtic periphrastic do is older, it seems to be a more plausible hypothesis.

Many linguistic scholars believe that English sporadically started using periphrastic do and the progressive. The problem is, however, that few have looked hard at languages outside the Germanic branch. If they did, they would see that few have a periphrastic do, and even fewer have do and a progressive present tense. Those languages that do use do are Japanese, Persian, and Nanai—spoken in Siberia. Of all of these periphrastic uses, none require its use in the past and present tense when in the negative and when asking questions. Mandatory periphrastic do is unique to the languages that shared the British Isles with English. Robert Penhallurick (1996) provides data that illustrates three variants that are matched in present tense: He (goes/do go/’s going) to the cinema every week. He notes the first matches standard English, the second parallels the output in neighboring rural English counties to the south, and the third reflects Welsh structural influence. He shows the same example in Welsh and discovers that Welsh uses “be” plus Welsh’s version of a gerund to express habitual action, shown in (7) (Penhallurick: 1996: 311).

(7) Mae ef yn mynd i’r      sinema bob wythnos.

He     is   going to the cinema every week.

Therefore, it is unrealistic to think that a group of languages that had those two same grammatical elements had no impact on English’s periphrastic do or present participle.

2.2 Substratum Contact Features: Verbal Noun, Northern Subject Rule, & Internal Possessor

Theo Vennemann (1999) develops a similar theory of origin when it comes to English’s progressive construction. “One may object that the progressive . . . may have originated in English independently of its existence in Insular Celtic. However the explanandum remains that among all Germanic languages, only English has undergone [the verbal noun] development” (355). English is the only Germanic language to have developed in a place where the native languages use the progressive as their base present tense, and English is the only Germanic language to use its verbal-noun as its only truly present tense. Go figure.

Vennemann (1999) discusses the influence of Semitic languages on both Celtic and Proto-Germanic. Due to trade relations on the Atlantic coast of the Iberian Peninsula, Hamito-Semitic languages had differing degrees of influence on Celtic and Proto-Germanic. Vennemann (1990) proves that Hamito-Semitic influence had a greater effect on substratum in Celtic and a greater effect on superstratum in Proto-Germanic. That is, Hamito-Semitic had more an influential impact on Celtic grammatical structure and Proto-Germanic lexicality. Germanic has been shown to have many words with Semitic etymologies that are shared by no other Indo-European languages, including Celtic. The Insular Celtic syntax had a radical transformation due to Semitic influence. Despite traces in Old Irish poetry, Insular Celtic syntax no longer shows the Indo-European characteristic of head-final word order. Its word order is quite unlike other Indo-European languages and more similar to the Semitic languages Arabic and Hebrew.

Vennemann (1999) translates Pokorny’s Keltischer Urgescichte und Sprachwissenschaft which says that Insular Celtic’s system of tense and aspect is very much unlike other Indo-European languages (1959: 155). Their construction of the progressive, “be” + preposition + verbal noun is much like the construction in Basque and Egyptian. Vennemann (1999) also notes that aside from the alikeness of the periphrastic progressive construction, there are striking similarities that exist between Insular Celtic and English’s functional range.

Vennemann (1999) also discusses the Northern subject rule that happens to English dialects spoken in Northumberland, Cumberland, Durham, and Westmoreland areas. The rule states that verbs take the typical third-person “s” in all forms except when adjacent to a personal pronoun in the sentence. He illustrates this rule with the following sentences: They peel them and boils themand Birds sings(Vennemann: 1999: 356). Interestingly enough, this Northern subject rule has a similar construction in Brythonic Celtic languages. Their rule is that the third-person-plural form of a verb is used only whennhw “they” is explicitly said. When it isn’t explicitly stated, the third-person-singular form of the verb must be used, and if there is no overt subject, third-person-plural agreement is used. Vennemann (1999) cites (8) from King (1993).

(8) maent “[they] are”

maent hwy “they are”

mae ‘r bechgyn “the boys are”

The first line is the no-overt subject rule where the verb has a third-person-plural agreement. The second is when hwy “they” is explicitly said where the verb has a third-person-plural agreement, and the third is when hwy “they” isn’t explicitly stated, but the subject is still third-person-plural, yet a third-person-singular agreement is used.

Vennemann (1999) cites Kemola (2000: 337) calling this type of agreement “extremely rare” (Vennemann:1999:357). The only other languages in which such a subject-verb agreement like this is found is Arabic and Hebrew. Arabic’s lack of grammatical concord is as follows: the verb agrees with the subject in gender and number, but if a plural pronominal subject is explicitly said, it is optional to put the verb in the plural or singular. Hebrew verbs also agree in gender and number with the subject, but if a list of subjects follows a verb that governs them, then the verb must be in the singular.

There is a last contact feature that Vennemann (1999) discusses. It is the internal possessor construction’s replacement of the sympathetic dative. In German, nouns have a case marking, indicating them as the subject (NOM) , indirect (DAT), or indirect (ACC) object. This external possessor construction is illustrated in (9).

(9) “The queen cut off the king’s head.”

Die Königin schlug dem König        den Kopf           ab.

The queen      cut       the king-DAT the   head-ACC off

English dropped this use of case markings when sentence structure became more strict, so English adopted the internal possessor construction. Evidence of the internal possessor construction can be found in Old and Middle English (10) is Old English and (11) is Middle English.

(10) “He cut off their hands and their noses.”

He cearf of heora handa    heora nosa

He cut   off  their  hands and their noses.

(11) “His legs they cut off immediately.”

Hys legges hy corven of anon

His legs   they cut    off immediately

The internal possessor construction began appearing as early as late Old English. The loss of the sympathetic dative can not just be a consequence of English’s loss of case markings because the sympathetic dative could easily have survived just as the directional dative did with verbs of giving. Examples of the directional dative are in (12) and (13).

(12) Josh gave Lindsey the book. Josh gave her the book.

*Josh broke Lindsey’s arm. Josh broke her the arm.

(13) Josh gave the book to Lindsey. Josh gave it to her.

*Josh broke the arm to Lindsey. Josh broke the arm to her.

Other Germanic languages, Swedish and Norwegian, where cases were neutralized like in English, the external possessor construction is not lost and is actually transformed into locative prepositional phrases. Of course, the sympathetic dative is also preserved in Germanic languages that have retained their case distinctions like Icelandic.

This leads us to conclude that English integrated the internal possessor construction through borrowing. And which languages have such a construction? You guessed it—Celtic. Insular Celtic to be exact. Vennemann (1999) calls this a “contact phenomenon” and displays examples of the internal possessor construction in Middle Welsh (13) and Early Modern Irish (14) (361) .

(13) “He had broken his arm.”

Mae e   wedi torri   ei  fraich

is     he ADJ break his arm

(14) “Cut off my head”

ben        mo chend    dim

cut-off  my   head from-me

Vennemann (1999) holds to his claim that the internal construction was a hand-me-down contact feature from Semitic languages by further examining Hebrew and Arabic’s use of external construction which have been highlighted in (15) Hebrew and (16) Arabic (363).

(15) “and he cut off his head with it.”

wajjikråt       -bah       ‘œt   -rōšō

and-cut-off-he with-her ACC head-his

(16) “Mary broke her neck.”

kasarat Mary raqabatahā

broke  Mary neck-her

We can only assume that English’s ousting of the sympathetic dative with the internal possessor construction is exactly what happened when the Celtic languages endured their language contact situation with Semitic languages.

2.3 Substratum Contact Features: Embedded Inversion

Filppula (2000) discusses the influence of Celtic languages on inversion in embedded questions in some regional dialects of English: Tyneside, Hebridean English, Welsh English, Southern-Hiberno-English, and Northern-Hiberno-English. He introduces three different hypotheses for this inversion. The first predicts that embedded inversion was retained from earlier forms of English, possibly from when English had a very robust verb-second property. He says that this hypothesis is undermined by the marginal status of embedded inversion in earlier forms of English. Embedded inversion turns out to be quite rare in Early Modern English, as evidenced Table 1 (Filppula: 2000: 442). The second hypothesis says that this is a simple vernacular inversion caused by simplification. Filppula (2000) agrees that there is evidence to support embedded inversion is to some extent a feature of general vernacular, but in some varieties, embedded inversion is more prominent than in other varieties. The third, and the hypothesis for which Filppula argues most, is that the inversion is caused by an elementary transfer from Celtic languages.

Embedded inversion has been described as a feature that is mostly typical of vernacular or colloquial forms of English. These inversions are typical of vernaculars in the British Isles, but Filppula (2000) points out that it is common enough in other forms of English colloquial speech. Embedded inversion has very close parallels in Celtic substratum languages, which have verb-raising in both root and embedded clauses. Filppula (2000) sites an example of Old Irish which keeps the word order of direct questions in indirect questions. It’s particularly transparent in the case of Yes/No questions. “Were you content?” in Old Irish is An raibh tú sásta? An is the question participle (QP), so a literal translation would be identical to English, besides the question participle an. In an embedded question: “He asked you if you were content,” Old Irish retains this verbal/nominal inversion in (17).

(17)  Chuir sé   ceist        ort    an  raibh  tú sásta

Put    he question on.you QP were you content

Irish doesn’t have the standard English way of stating an embedded question with if or whether. The inversion is retained in word order of Wh-questions as well. The direct Wh-question has the same word order as an English Yes/No question, though it is embedded under a different layer consisting of the relative clause structure. A is Old Irish’s relative particle (RP) in (18) and (19).

(18) Where did you see it?

Cé      an   áit      a bhfaca  tú  é

Where the place RP  saw   you it?

Again, this inversion is retained in an embedded Wh-question.

(19) He asked where you saw it

Chuir sé     ceist        cé    an  áit      a bhfaca tú é

“Put   he question where the place RP saw you it.”

Similar structures exist in other Celtic languages, which are directly parallel to the six regional dialects that Filppula (2000) mentions. The yes/no questions’ correspondence is complete, while the wh-questions in Old Irish require the use of a relative participle signaling the relative clause structure. Filppula (2000) believes this accounts for the greater degree of use in embedded yes/no questions than in embedded wh-questions of the six regional varieties of English.

3.1 Superstratum Contact Features

In his article Celtic Influence on Old English Rhetoric, Wolfgang Kuhlwein (1998) examines the contextualized rhetoric in Old English poetry, contrasting it with Old High German, Icelandic, and Nordic texts and comparing it to Welsh and Cornish texts. He states that the frequency of five rhetoric characteristics have a “full-scale exploitation of the semiotic repertoire of iconic, indexalic, and symbolic relationships between language and the world to which it relates” are mirrored in a large diversity of Old English and Celtic texts and are missing entirely or radically different from Germanic texts (Kuhlwein: 1998: 231). There is a sharp contrast between the way Celtic and Old English texts used these macro- and micro-structures and how Germanic language texts used them.

The first is the interweave of human and supernatural/nonhuman elements. Kuhlwein (1998) sites specific examples from Old English’s Battle of Maldon written in 991 A.D. The poem frequently describes the warriors as wælwulfas “wolves of the blood-drenched battle ground”, which shows an intermixing of human and supernatural elements. As does the Welsh manuscript The Dream of Rhonabwy in which a battle is waged between humans and ravens. In the manuscript the ravens are depicted as having wrath, fierceness, high spirit, and courage. The nature of human and non-human merge in the ravens, as they are equipped with human emotions. Kuhlwein (1998) says that this form of rhetoric prevails over most Celtic poetry. In contrast, he sites the Old Nordic text Völundarkvidha where females from another world called Valkyries marry human mates. This mixing of human with non-human is inherently different than both the Celtic and Old English texts.

The second rhetoric component that Kuhlwein (1998) discusses is the strong emphasis on color. Again, in Old English Battle of Maldon, color plays an important role as the wolves’ gloomy grayness, the silvery shine of the water, and the darkness and blackness of death are all weaved together to contrast the end of the scene where a strong feeling for the color effects of light, luminescence and glow are detailed. In the Welsh text The Dream of Rhonabwy, the description of the youth shows an acute awareness and sensitivity to color. The youth literally have a craving for color and light. Kuhlwein (1998) sites examples such as “a mighty lance, speckled yellow,” “a scarf of blue satin”, and “a brooch of gold.”

A sociocultural evaluation is the third rhetoric microstructure that Kuhlwein highlights. In the Old English poem Hildebrandslied, the poet has a sociocultural evaluation where he equalizes the prospect of fighting and receiving honor for it. In the Welsh text The Book of Aneirin, there is a gathering of strength and energy when the poet rallies to the banner, a sociocultural evaluation. Kuhlwein (1998) states that sociocultural evaluation and ethical values are prevalent through most Germanic texts, but it’s usually from the combatants’ point of view. It is rarely ever an explicit, personal comment from the poet. He sites specific examples from the Germanic language texts Bragarædhur and Gylfaginning where the reader/listener expects to find a personal sociocultural evaluation at  the endings of sagas and tales. Kuhlwein gives two other forms of rhetoric evident in Celtic and Old English texts and lacking in Germanic ones, unity shinning through surface plurality and personal emotion/involvement.

This textual phenomenon can be attributed to the trilingual environment in which these texts were written. Nearly two centuries before the first Anglo Saxon invasion, Roman mercenaries who were of Germanic descent married woman of Celtic descent. They ultimately settled down in Roman forts and raised children in a Latin-Germanic-Celtic environment. Emphasis was given to Germanic and Celtic. It was the Celtic speaking mother who raised the children and instilled the Celtic language and culture. These interlingual families outlived the Anglo Saxon conquest, and by nearly a century later, Irish, a Celtic language, took the lead in Christianizing northern England, excelling in educational and cultural leadership. These mercenaries are to whom many Anglo Saxons flocked to receive their education.

4.1 What Really Happened?

Kuhlwein (1999) also discusses that after the Anglo Saxon invasions, many Celts were enslaved or killed, but those who weren’t, fled to the outskirts of the British Isles to escape the invasions. Many scholars say Celtic languages had very little impact on Old English since the language contact was kept to a minimum, but McWhorter (2008) full-heartedly disagrees. He says that many historians believe that the “Anglo Saxon invaders routed the Celts in more or less a genocide” (11). The genocide of a people expanding an area about the size of New England would only be possible with present-day weaponry. The Anglo Saxon invaders had only swords and spears, so the genocide of an entire people seems highly unlikely. Bryson (1990) compares the writers of this era claiming that there was a total genocide of Celts to historians of the Elizabethan era writing about Elizabethan England based on hearsay evidence. The writing historians during these invasions lacked evidence of a genocide of an entire people because the news they received was unsubstantial and irregular. Modern comparative genetics is able to reconstruct the migration of human beings since the emergence of Homo sapiens by tracing mutations of mitochondrial DNA in women and the Y chromosome in men. These tests show that only four percent of British men and essentially no British woman have DNA that is traceable to migrations across the North Sea. This proves that the Germanic invaders were not men who brought their wives and children with them on the invasions, as bringing them would bulk up the total number of people on the voyages. Thus, McWhorter (2008) corroborates Kuhlwein’s (1998) theory that Germanic invaders settled down with Celtic women, giving birth to bilingual children. McWhorter (2008) furthers the claim that Celts were well integrated into Anglo Saxon society by discussing the laws that were put forth by Ine, the seventh-century king of Wessex, an era before any man considered himself to be king of England as a whole. His laws indicate a Britian in which Celts are in large numbers, referring to them as the Wealhs“Welshmen.” Ine also created legislation on their behalf, proving the integration of Celts in Anglo Saxon society. However, this integration was not exactly full. This name given to Celts was not what they called themselves, which was Cymry. Though Wealhs translates as “Welshmen” to us, Wealhs actually meant “foreigner” with a tacit implication of “slave.” However, Ine created legislation on behalf of these “slaves,” referring to them as respectable landowners and horseman serving the king.

5.1 Mutual Reinforcement Hypothesis

Thus, the genocide story has fallen apart. We must assume that these Celts who were integrated into Anglo Saxon society were forced to speak Old English, sprinkling their Celtic grammar into it. But what kind of influence did English have on the Celtic languages besides just reiterating doperiphrasis, according to Van Der Auwera (2002)’s mutual reinforcement hypothesis? Otto Jespersen (1961) identifies phrasal verbs as “one of the most characteristic traits of the English language (323). Rottet (2005) examines Welsh’s prolonged contact with English and states that it has had a rather profound influence on Welsh speakers use of phrasal verbs. However, the evidence is blurred because of the fact that Brythonic Celtic languages have a similar construction of a phrasal verb: verb + adverb. Rottet (2005) gives quite a few examples of these verb + adverb constructions highlighted in (20) (42).

(20) mynd i lawr “go down”

mynd i mewn “go in”

mynd i ffwrdd “go away”

mynd i bant “go off”

mynd i fyny “go up”

Mynd is the infinitive verb “to go.” The  i “to” in these data attaches to a noun object. i lawr means “down.” lawr is a variation on the word llawr “floor” and ffwrdd is a variant of ffordd “road.” These i+ noun object constructions are inherently directions, thus resembling something quite like English phrasal verbs. In Welsh, pronominal direct objects are placed differently into the sentence than in English. English’s “break it down” looks like (21) in Welsh.

(21) torri   i lawr              ef

break to floor (down) it

Brythonic’s daughter languages include Welsh, Cornish, and Breton; it split into these three distinct, yet extremely close dialects in the sixth century. The latter is the only language not to have had prolonged, intensive language contact with English, which proves to be very helpful in discovering English’s influence on Welsh phrasal verbs. Breton phrasal verbs distinguish between dynamic and static adverbs, Breton d’an nec’h means “up” when motion is implied, but en nec’h is used for “up” when location only is intended. The biggest difference between phrasal verbs in Welsh and Breton is their frequency of use. Because Breton has quite a few single word verbs that express the same thing as many of its potential phrasal verbs, a limited number of verbs of motion enter into a phrasal verb construction in Breton, while Welsh “constructs new combinations quite productively” (Rottet: 2005: 47).

Rottet (2005) gives quite a few patterns that lend themselves to prove that English’s phrasal verbs did have a large impact on Cornish and Welsh phrasal verbs. The first is when Welsh speakers add an adverb to a verb that customarily does not take a particle. Rottet (2005) calls these additions as “pleonastic claquing of particle” (53). Welsh similarly has a lot of one-word verb phrases that are semantically identical to phrasal verbs in the verb + i + adverb construction, but often times, a Welsh speaker will choose the use of the phrasal verb over the one-word verb. And then there is complete lexical borrowing on behalf of Welsh, such as in the example in (22) from Owen (1993: 65).

(22) If he were in the chapel, he would be kicked out at once.”

Dase fo yn     y capel,            mi fase yn cael y kick out yn syth

If     he were in the chapel,    would be        kicked out  at once.

However, nominal phrasal verbs are the most frequent direct borrowings into Welsh, such as in the example in (23) (Rottet: 2005: 60).

(23)   criw  o dwyllwyr, comiwynyddion a dropowts.

“crew of deceivers, communists,  and dropouts.”

The rendering of English phrasal verbs in Welsh is only one kind of linguistic element that English has had influence on in Modern Welsh according to Rottet (2005).

Of course, we couldn’t conclude a manuscript about the influence of Celtic on English without surveying a sample of the many, many authors that argue for lexical borrowing only. They are linguistic researchers who choose not to buy into the overwhelming amount of evidence that Insular Celtic languages had a profound effect on English. Stalmaszczyk (1997) believes that the amount of Celtic influence on English was minimal and applies only to a rather expansive list of vocabulary, and only a handful of these words exist in the language today. Loreto (2000) documents a few scholars saying that Latin words borrowed into English during this time were most likely from Latin-speaking Celts. It’s strange if these scholars truly believe that Anglo Saxons would borrow Latin words from Celts but not Celtic words. Loreto (2000) believes that many of the words that scholars say are of Celtic origin have multiple etymologies due to the number of languages spoken in the British Isles during this time. Lovis (2001) finds that Celtic loanwords come from three identifiable sources: the continent—words associated with battle and conflict, loans taken over after settlement—usually place names, and words associated with the Christianization of the British Isles—usually of Irish origin in particular.

6.1 Conclusion

The hypothesis of Celtic contact influence on English is more widely accepted among non-British scholars. One would think that acceptance of a theory among scholars would not depend on the nationality, institutional environment, or native language of each scholar, but in the case of Celtic influence on English it apparently does. The work and textual analysis presented in this manuscript are all only possible theories in the eyes of most scholars of the English language. The difference in acceptance of such theories lies in the difference of opinion of how much coexistence English and Celtic really had. Many historians believe that the Anglo Saxons only terrorized the Celts, never really integrating with them. This seige, according to some historians, made the Celts flee to the outskirts of the Isles, keeping language contact at a minimum. However, historians and linguists who believe the genocide of an entire people is highly unlikely and that the Anglo Saxons had to have had some integration with the Celts, believe in greater influence of Celtic languages on English. Within the last thirty years though, Linguists have been able to develop research methods that bring them closer to being able to judge some theories more plausible than others. There will, however, probably never be a true consensus.


Bryson, Bill. 1990. “The Mother Tongue: English and how it got that way”. Harper Collins: New             York.

Filppula, Markku. 2000. “Inversion in Embedded Questions in Some Regional Varieties             of English”. Generative Theory and Corpus Studies ed. by Ricardo Bermúdez, David             Denison, Richard M. Hogg, C. B. McCully, 439-453. Berlin, Germany: Mouton de             Gruyter.

Jerpersen, Otto. 1961. A Modern English grammar on historical principles. Vol 2. London:             Allen and Unwin.

Kemola, Juhani. 2000 “The Origins of the Northern Subject Rule – A Case of Early contact?”

Tristram 2000.329-346

King, G. 1993. Modern Welsh: A comprehensive grammar. London: Routledge.

Kuhlwein, Wolfgang. 1998. “Celtic Influence on Old English Rhetoric-A Case Study of the             Interface between Diachronic Contrastive Rhetoric and History of Art”. Studia Anglica             Posnaniensia 33.213-243.

Loreto, Todd. 2000. “Where Have All the Celtic Words Gone?”. English Today 3:63. 6-10.

Lovis, Claire. 2001. “Celtic Influence on the English Language”.  Unpublished ms. Retrieved             March 15, 2010, from the University of Toronto’s website:             http://www.chass.utoronto.ca/~cpercy/courses/6361Lovis.htm

McWhorter, John. 2008. Our Magnificent Bastard Tongue: the Untold Story of English. New             York: Penguin Group.

Owen, Daniel. 1993. Rhys Lewis. Cardiff: Hughes

Penhallurick, Robert. 1996. “The Grammar of Northern Welsh English: Progressive Verb             Phrases“. Speech Past and Present: Studies in English Dialectology ed. by Juhani             Klemola, Merja Kytö, Matti Rissanen, 308-342.

Pokorny, Julius. 1959. “Keltische Urgescgichte und Spracjwissenschaft.” Die Sprache 5.152-164

Rottet, Kevin J. 2005. “Phrasal Verbs and English Influence in Welsh”. Word 56:1. 39-70.

Shakespeare, William. Hamlet

Stalmaszczyk, Piotr. 1997. “Celtic Elements in English Vocabulary-A Critical Reassessment”.            Studia Anglica Posnaniensia, 32. 77-87.

Van Der Auwera (2002), Johan & Inge Genee. 2002. “English do: On the Convergence of             Languages and Linguists“. English Language and Linguistics 6:2. 283-307.

Vennemann, Theo. 1999. “Atlantis Semitica: Structural Contact Features in Celtic and             English”.Historical Linguistics 1999 ed. by Laurel J. Brinton, 351-369. Amsterdam:             John Benjamins.


Racial Dynamics: Toni Morrison’s “Recitatif”

*Published in The Fogdog Review Winter 2010

The concluding line to Toni Morrison’s only published short story “Recitatif,”  ‘‘What the hell happened to Maggie?” is a curious one indeed (Morrison 2698).  Instead of bringing about a resolving conclusion, this question advances the reader’s curiosity to the nostalgic memory of Maggie, a mute woman who worked in the kitchen at the orphanage in which the story’s two main characters, Roberta and Twyla, were raised. The ambiguous childhood memory of Maggie figuratively and literally becomes the central conflict between Twyla and Roberta’s friendship. Neither of the girls ever truly knew Maggie well and neither saw Maggie after leaving St. Bonny’s orphanage, but the memory of the orchard in which the bowlegged Maggie fell reoccurs every time the girls reencounter one another in adulthood. Maggie is not an active character in the story; she is a mere memory for Twyla and Roberta. Instead, Maggie’s character works as a symbol for both Twyla and Roberta’s companionship and conflict and their similarities and differences.

Twyla, the story’s narrator, tells the story from a first person point of view. However, her narration is challenged as Morrison explores the memory of Maggie in the orchard, making Twyla reevaluate this incident every time she encounters Roberta. Twyla and Roberta disagree about the incident in the orchard, and this disagreement forces Twyla to wonder if her memory is deceiving her. The orchard becomes an important setting, as it is the place where Twyla and Roberta become both victims and victimizers. When she first mentions the orchard, Twyla says, “I don’t know why I dreamt about the orchard so much. Nothing really happened there. Nothing all that important, I mean,” (Morrison 2686).  Here, Twyla’s remark is a bit confusing and is indeed something to look at. As David Goldstein-Shirley says in his article “Race and Response: Toni Morrison’s ‘Recitatif’:”

This comment is most problematic. The entire narrative comprises Twyla’s recollections of past events. If, as readers naturally assume when reading a first-person account, the narrator is speaking to them in the present, then why would Twyla say that nothing really happened in the orchard? Telling a retrospective story, she ought to know that the incident with Maggie, which obsesses her throughout the story, is not only significant but crucial (Goldstein-Shirley).

Yes, as the audience progresses further and further into the story, they find that important events did take place in the orchard, and that Maggie’s humiliation, as well as their own, become a central theme to the story. When first discussing the orchard, Twyla describes it as having hundreds of apple trees and being “fat with flowers,” (Morrison 2686). They would watch the older, intimidating girls dance to the radio there. Twyla then describes a particular day when Maggie hurries through the orchard to catch her bus but falls, and all the older girls laugh at her. Twyla feels some sting of guilt as she ventures to say, “We should have helped her up, I know, but we were scared of those girls,” (Morrison 2686). Twyla realizes, here, the hierarchal power at St. Bonny’s and her own position in it. Twyla and Roberta were intimidated by and frightened of the older girls. Looking back on it, however, Twyla realizes the older girls were merely acting tough even though they were “poor little girls [who] fought their uncles off,” (Morrison 2686). At St. Bonny’s, the older girls group together to victimize Roberta, Twyla, and Maggie. Twyla and Roberta then victimize Maggie by calling her names because they fear being voiceless and powerless like Maggie. They identify themselves with the older, victimizing girls instead of coming to Maggie’s aid.

Twyla and Roberta are able to be victimizers because they have each other. They share the fact that their mothers have left them at St. Bonny’s, Twyla’s because she “danced all night and Roberta’s [because she] was sick” (Morrison 2685). Although each girl is lucky that her parents are not dead, both feel the pain of abandonment and this feeling becomes a key reason as to why the girls become so close despite their racial differences. Although the girls were very close at St. Bonny’s, Twyla describes their meeting again at a diner called, “Howard Johnson’s” as being much less warm. Their racial separation seems to be a much larger issue. When Roberta and Twyla later discuss their militant meeting, Roberta says, ‘‘Oh, Twyla, you know how it was in those days: black-white,” (Morrison 2693). In Newburgh, not only do the two women live in separate sections of a racially segregated community, but also they are of different socio-economic classes. Roberta married a wealthy man that worked with “computers and stuff,” and Twyla married a fireman (Morrison 2693).

When the town becomes divided because of the bussing controversy, the racial and economic differences between Twyla and Roberta become even more apparent. Twyla becomes an advocate for bussing while Roberta vehemently opposes it. Their disagreement over what exactly happened to Maggie is reflected in this public confrontation. Twyla begins to argue with Roberta while Roberta is picketing, and a large mass of people surround Twyla’s car and begin throwing rocks at it. Twyla responds to this aggressive action by saying:

Automatically I reached for Roberta, like in the old days in the orchard when [the older girls] saw us watching them and we had to get out of there, and if one of us fell the other pulled her up and if one of us was caught the other stayed to kick and scratch, and neither would leave the other behind. My arm shot out of the car window but no receiving hand was there (Morrison 2695).

At this point, Twyla identifies with the helpless feelings that Maggie had in the orchard that day. At St. Bonny’s, Twyla and Roberta were close friends because of their feelings of abandonment, but here, Roberta abandons Twyla. Because she gropes for Roberta’s hand only to find it not there, Twyla relates to the same humiliation that Maggie had in the orchard. Roberta then compares Twyla to one of “the big girls on the second floor” by telling Twyla that she is “the same little state kid who kicked a poor old black lady when she was down on the ground,” (Morrison 2686, 2695).

The hierarchies that existed back in the orchard at St. Bonny’s now exist in this historical, public scene. These hierarchies, however, are different. The power hierarchies in the orchard concern intimidation and powerlessness, but in the bussing scene, they concern race. Twyla had previously described Maggie as “sandy-colored,” so when Roberta described Maggie as being black, Twyla at once rejects this idea (Morrison 2686). This, however, can lead readers to question the validity of Twyla’s memory. “The problematic accusation also calls into question the completeness of Twyla’s storytelling,” (Goldstein-Shirley). Upon later thought, Twyla admits, “I actually couldn’t be certain. She wasn’t pitch-black, I knew, or I would have remembered that. What I remember was the kiddie hat, and the semicircle legs,” (Morrison 2697). Because Twyla was unsure of which race Maggie actually belonged to, it is obvious that race was not a factor in the St. Bonny’s orchard. “During their childhood, this bond conceal[ed] complications of race and class,” but now with where each of the women are in this tightly-knit community, race takes on a whole new meaning (Androne 136). Elizabeth Abel even says, “Roberta is skeptical about racial harmony.” Is it her overbearing mother’s fault for her abrupt discrimination, or was it her uprising in social class? Regardless, Maggie’s humiliating fall in the orchard begins to take on racial significance because of it.

At Twyla and Roberta’s last meeting at the diner where they run into each other, Roberta confesses to having lied about the two of them kicking Maggie when she was already down on the ground. Twyla had already concluded that she had not kicked Maggie like Roberta said but admits to having wanted to. Her wanting to kick Maggie when she was already down is due to the fact that Twyla connects Maggie with her mother. “Maggie was my dancing mother,” (Morrison 2697). “Twyla and Roberta revise their memories of Maggie in order to transfer their anxieties and anger toward their mothers onto her,” (Androne 134). Twyla connects her repressed hostility toward her irresponsible mother with the pity she had for Maggie. At the diner, Roberta concedes that she had also wanted to kick Maggie. She too identifies her feelings of abandonment with Maggie, comparing Maggie with her mother as well even though Roberta’s mother  is the opposite of Twyla’s mother, Mary. Moreover, both Twyla and Roberta identify themselves with Maggie. “I knew she wouldn’t scream, couldn’t—just like me—and I was glad about that,” says Twyla as she compares her own feelings of helplessness to Maggie’s (Morrison 2697).

The way in which Toni Morrison creates an inactive character such as Maggie, and turns her into a central symbol brings about interesting questions regarding the racial fissure that divides Roberta and Twyla. Through Maggie, the woman begin to release their tumultuous and repressed feelings of St. Bonny’s because of Mrs. Itkin, Twyla and Roberta’s caretaker, nicknamed Big Bozo, and their mothers’ neglect. By connecting themselves with Maggie, a woman they formerly suppressed, they gain a greater understanding of themselves, each other, and their racial differences. However, “What the hell happened to Maggie?” is still a lingering question on their minds (Morrison 2698). It is one that will never be answered, and the women will never be as complete as they would have been if they had not abandoned Maggie and eventually each other.

Works Cited

Abel, Elizabeth. ‘‘Black Writing, White Reading: Race and the Politics of Feminist Interpretation,’’ inCritical Inquiry, Spring, 1993, pp. 471–98.

Androne, Helane Adams. MELUS, Summer2007, Vol. 32 Issue 2, p133-150, 18

Revised Memories and Colliding Identities: Absence and Presence in Morrison’sRecitatif” and Viramontes’s “Tears on My Pillow.”.

Goldstein-Shirley, David. ‘‘Race and Response: Toni Morrison’s ‘Recitatif’,’’ in Short Story, Vol. 5, No. 1, Spring, 1997, pp. 77–86.

Morrison, Toni. “Recitatif.” The Norton Anthology of American Literature Volume E. 7th ed. Ed. Nina Baym. New York: Norton & Company, 2007. 2684-98.

Tag Cloud