1 Heritage Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Geminate Attrition in the Speech of Arabic–English Bilinguals Living in Canada

Anwar Alkhudidi Ryan Stevenson Yasaman Rafat The University of Western Ontario

ABSTRACT This study has three goals. First, it examines phonological attrition of the Arabic geminate (i.e., long sound)-singleton contrast in the speech of native speakers of Arabic who acquired English after puberty. Second, it compares geminate production in late bilinguals to early bilinguals. Third, it investigates whether universal phonetic/acoustic factors have an effect on the degree of attrition across generations. Participants performed a delayed word repetition task, where mean consonant duration was measured and compared across the two bilingual groups and compared with native speakers of Arabic. Results show that a geminate-singleton duration ratio continuum across groups was formed, where monolingual speakers of Arabic had the highest values, followed by late bilinguals, and then early bilinguals. These results are interpreted as evidence that language remains malleable across the lifespan. Moreover, whereas there was no effect of manner of articulation, voiced geminates showed a significantly higher degree of attrition across both bilingual groups. In addition, inherently longer geminates attrited at a significantly higher rate. Implications for models of phonological attrition are discussed.

KEYWORDS: Bilingualism, attrition, contact, phonetics, , geminates, Arabic, English

1. INTRODUCTION (L1) attrition in bilinguals1 refers to the process in which there is a loss of the L1 or a gradual decline in proficiency mainly caused by interference from a second language (L2; Schmid & Köpke, 2007). L1 attrition, for the most part, has been observed in immigrant populations who have had long-term exposure to an L2 and decreased exposure to their native language (e.g., Chang, 2012; de Leeuw, Tusha, & Schmid, 2018). It can be seen in any part of the linguistic (e.g. Seliger & Vago, 1991), although it is generally believed that L1 phonology is not prone to loss (Schmid & Köpke, 2004). As such, most of the literature on L1 attrition has focused on lexical and grammatical aspects of language (Schmid & Köpke, 2004). Less has been done on phonetics (Flege, 1987; Guion, 2003; Major, 1992; Mayr, Price, & Mennen, 2012) and even less on phonological attrition (Celata & Cancila, 2010; de Leeuw, Tusha, & Schmid, 2018; Schmid & Köpke, 2004; Rafat, Mohaghegh, & Stevenson, 2017; among others). To address this gap, first, the current study examines the production of the geminate-singleton phonological contrast in Arabic-English bilinguals living in Ontario, Canada, who learned English in adolescence. In terms of closely related literature, to date, we are only aware of two studies: one that provided evidence of attrition in the perception of geminates in Italian-English bilingual immigrants living in the United States (Celata & Cancila, 2010), and another that provided evidence of geminate attrition in Farsi-English bilingual immigrants living in Canada (Rafat et al., 2017). The current article hopes to shed further light on whether geminate attrition in bilinguals, namely occurring when an L1 with a geminate-singleton contrast (e.g., Arabic) comes in contact with a language in which consonant length is not contrastive (e.g., English), and is not language-

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 2 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

specific but rather universal. The second goal of this study is to investigate whether L1 phonological change in late bilinguals patterns with the L1 phonology of early Arabic-English bilinguals who learned both Arabic and English in childhood. These bilinguals are also considered heritage speakers of Arabic because they learned English, the dominant language of Canadian society, while learning Arabic at home. The third goal of the study is similar to that of Rafat et al. (2017), who examined the role of universal phonetic principles on geminate attrition in Farsi- English bilinguals in Canada. We will examine whether the same universal phonetic principles, such as aerodynamic (Ohala, 1983) and perceptual constraints (e.g., Kawahara, 2007; Podesva, 2002; Taylor, 1985), also affect geminate production in both groups of Arabic-English bilinguals. In other words, we will be testing to see whether there are parallels between sound change, L2 acquisition, and L1 attrition; that is, we are interested in whether the same factors that constrain sound change and L2 speech learning may also constrain L1 speech attrition. Perceptual constraints have been shown to affect geminate change across world languages and geminate acquisition in L2 learners, so we examine their effects on attrition. A noteworthy example to point out is Sorianello (2014), who reported that manner of articulation and voicing determine the degree of geminate acquisition by L2 learners of Italian. Geminates are also typologically considered marked (i.e., rare) sounds (Maddieson, 1984) and may therefore be particularly prone to attrition.

Inspired by the points raised to this juncture, the research questions guiding this study are as follows: 1. Does the geminate production of late and/or early Arabic-English bilinguals residing in Canada undergo attrition? 2. Do phonetic/acoustic factors that affect L2 acquisition also have an effect on the degree of attrition (across generations)? Specifically, do manner of articulation and voicing constrain geminate attrition?

The reminder of this article will be organized as follows: the first half of the article will introduce the phenomenon of gemination cross-linguistically as well as a brief explanation of the nature of geminates in Arabic and English. Next, we will present the Speech Learning Model (Flege, 1995). We will then provide an overview of some of the studies on L1 phonetic and phonological attrition, followed by a review of previous studies on heritage Arabic phonology. This will be followed by the hypotheses, methodology, data analysis, results, and a general discussion.

2. LITERATURE REVIEW 2.1 Geminate Consonants across the Languages of the World Geminates are phonetically long sounds (Esposito & Di Bendetto, 1999; Homma, 1981; Kraehenmann & Lahiri, 2008; Lahiri & Hankamer, 1988; Ohala, 2007). A length contrast between singleton (i.e., short) and geminate consonants is observed in 3.3% of the world's languages (Maddieson, 1984); therefore, gemination is considered a marked phonological phenomenon. Some examples of languages with a binary phonological contrast between short and long consonants include Italian (Celata & Cancila, 2010; Payne, 2005), Arabic (Khattab & Al Tamimi, 2014), Farsi (i.e., Persian; Hansen, 2004; Rafat, 2008, 2010), and some Indonesian languages, such as Buginese, Madurese, and Toba Batak (Cohn, Ham, & Podesva, 1999). Geminates also exist in Inari Sami and Finnic languages, such as Finnish and Estonian, where there is a ternary (i.e., three- way) contrast (Markus, Lippus, Pajusala, & Teras, 2013).

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 3 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Gemination does not affect all consonants equally and is motivated by complex phonetic processes (Blevins, 2004; Podesva, 2002; Steriade, 1982; Taylor, 1985). Both acoustic/perceptual salience and articulatory/aerodynamic difficulty can predict geminate occurrence. For these reasons, geminates are most likely to occur in obstruents, and in particular, in voiceless obstruents. Geminate voiced obstruents are not as common as their voiceless counterparts (Hayes & Steriade, 2004; Jaeger, 1978; Ohala, 1983; Taylor, 1985) because of a phonetic restriction, namely the difficulty in maintaining a sufficient transglottal air pressure drop to produce voicing with a long closure. In general, voicing is a good predictor of degemination, whereby voiced sounds are more likely to be reduced from a geminate to a singleton (Elmedloui, 1993; Kawahara, 2007). Geminate fricatives are more marked than their non-continuant obstruent counterparts (i.e., stops and affricates); there are no languages that have geminate fricatives but lack geminate stops (e.g., Taylor, 1985). Geminate fricatives may be more marked than geminate stops because the singleton–geminate difference is harder to perceive for fricatives than for stops (Kawahara, 2013). It has also been proposed that fricative geminates may involve more articulatory effort than geminate stops (Kirchner, 1998). Sonorant geminates are the most marked group of geminates, which is attributed to sonorance making a geminate-singleton duration contrast harder to perceive (e.g., Kawahara, 2007; Podesva, 2002; Taylor, 1985). Kawahara (2007) proposes that sonorant consonants cannot acoustically signal their duration as well as other sounds do because of blurry formant transitions into and out of flanking vowels.

In L2 production, phonetic universals affect geminate consonant realization. Sorianello (2014) examined the production of Italian geminates by German-, Spanish-, and Chinese-speaking learners of Italian. Specifically, manner of articulation, voicing, and stress affected L2 geminate consonant production, albeit manner of articulation and voicing were better predictors of gemination than stress. Subsequently, Sorianello reported that gemination was more likely to occur with voiceless stops and degemination with sonorant consonants.

Another universal factor that has been examined in the acquisition of geminates more recently is the effect of length. Length is universally considered a salient property of sounds (Bohn, 1995; Cebrian, 2006; Escudero, 2001; Kondaurova & Francis, 2008, 2010; Morrison, 2008; Ruíz-Peña, Sevilla, & Rafat, 2015). Varona, Ruíz-Peña, Sierra, Stevenson, and Rafat (2017) also examined geminate imitation by L2 and second dialect (D2) learners. In particular, they investigated the geminate imitation of Standard Italian (L2) and Havana Cuban Spanish (D2) by native Colombian Spanish-speaking participants. They tested the effect of both language and length. Their results indicated that participants were able to produce geminates from a phonological point of view, though they had difficulty producing them in a target-like manner in both the L2 and D2 contexts. Moreover, length was the main predictor of non-target-like productions, where shorter geminates were easier to imitate. These results led the authors to propose that there might be a trade-off between acoustic salience and target-like productions in L2 and D2 speech learning. Although length was the main predictor of non-target-like productions, the authors also found a correlation between age of arrival (AoA) and degree of geminate duration reduction by the participants.

2.2 Geminates in Arabic and English Arabic is a language that features gemination, where length is phonologically contrastive; for example, /ʕalam/ ‘flag’ contrasts with /ʕalːam/ ‘he taught.’ Gemination in Arabic is represented in its orthography by the diacritic shadda, < ّ >, which is always written above the consonant it ’.ʕalːam/ ‘he taught/ <ع لم> ,follows; for example

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 4 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

All Arabic consonants can be geminated. Geminates in Arabic can be found in word-medial intervocalic position (e.g., /kasːar/ ‘he smashed’ versus /kasar/ ‘he broke’), and word-final position (e.g., /ʕaːm/ ‘year’ versus /ʕaːmː/ ‘public’; Abu-Abbas, Zuraiq, & Abdel-Ghafer, 2011; Al- Tamimi, Abu-Abbas, & Tarawneh, 2010; Mitchell, 1990).

Moreover, in minimal pairs, the vowel preceding a geminate consonant has been reported to be different from the vowel preceding a singleton constant. For example, Al-Tamimi (2004) reported that the duration of the preceding vowel was found to interact with consonant length such that the duration of the preceding vowel in a geminated consonant was shortened to emphasize its production. Other studies, on the other hand, found no such effect of geminates on the duration of the preceding vowel (Ghalib, 1984; Hassan, 2003). Whereas geminates in Arabic are phonemic, gemination in English is phonetic. Geminates in English exist when two identical segments occur at a word boundary (e.g., ‘night time,’ ‘top pick,’ and ‘makes sense’; Abercrombie, 1967; Kaye, 2005; Kreidler, 2004; Oh & Redford, 2012; Trask, 1996). Geminates in English may also exist in child-directed speech (e.g., [sʌnːi] instead of [sʌni] for ‘sunny’). They can also sometimes be found in emphatic speech, although this may be socially constrained (e.g., [həlːoʊ] instead of [həloʊ] ‘hello’), and as such, it may not exist as an allophone of singletons for all English speakers or for Arabic-English bilinguals. In sum, the fact that gemination does not always exist in intervocalic position and/or is not phonemic in English may lead to L1 geminate attrition in Arabic in the same position. 2.3 Speech Learning Model Whereas a number of L2 speech learning models have been proposed (see Brown, 1998; Colantoni & Steele, 2008; Escudero, 2005; Flege, 1987, 1995), only the Speech Learning Model (SLM) (Flege, 1987, 1995) addresses the potential influence of the L2 on L1 categories (i.e., sounds) in addition to the effect of the L1 on L2 category establishment. The SLM models are about advanced learner acquisition of an L2. Because the model is on advanced learners and has considered bi- directional L1 and L2 interactions in bilinguals, we will adopt this model for our current study. Regarding the interaction of L1 and L2 categories in bilinguals, the SLM posits that both sets of categories share the same space, thus L1 and L2 sounds that are linked to the same position may influence one another in the bilingual mind. Influences may occur in two different directions, namely assimilation or dissimilation; that is, L1 and L2 categories that are judged as “similar” may assimilate and become more similar to one another, whereas those that are judged as “different” may dissimilate and become less similar to one another. Similarity between L1 and L2 sounds in the SLM is defined in terms of the phonetic/acoustic distance between the two sounds; the smaller the acoustic/phonetic distance, the higher the possibility that L1 and L2 sounds may be perceived as similar. Based on these hypotheses, we predict that in Arabic-English-bilinguals, Arabic geminates will assimilate to their English singleton counterparts as “similar” sounds, leading to geminate attrition in Arabic.

2.4 L1 Phonetic and Phonological Attrition The literature on phonological attrition has examined different aspects of the L1. For example, in a study on bilinguals, Flege (1987) found evidence of a shift in voice onset time (VOT) values in the L1 of English-French and French-English adult bilinguals resulting from exposure to the L2.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 5 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Another study, by de Leeuw, Mennen, and Scobbie (2012), investigated whether production of the German lateral phoneme /l/ of late German-English bilingual speakers living in Canada underwent attrition. Results showed that the first and second formant (F1 and F2, respectively) values of the German lateral /l/ of the bilinguals differed from those of the native German speakers, showing gradual drift toward L2 (i.e., English) values.

Hrycyna, Lapinskaya, Kochetov, and Nagy (2011) took a sociolinguistic approach to investigating VOT values of the voiceless stops /p, t, k/ in the L1 of successive generations (i.e., first, second, and third generation) of Italian-, Russian-, and Ukrainian-English bilingual communities. They found evidence of a shift in VOT values toward English as well. Differences between language groups were also attributed to social factors, such as the cohesiveness and size of a community, as well as participants’ attitudes toward their L1.

Whereas a number of studies have examined phonetic shift, to the best of our knowledge, fewer studies have examined phonological attrition. Phonetic shift refers to one or more changes in the acoustic-phonetic properties of an L1 sound, where said L1 sound becomes more similar to its nearest equivalent in the other language of the bilingual, from a phonetic point of view. De Leeuw, Tusha, and Schmid (2018) is one of the few studies to have examined phonological attrition in immigrant populations. Specifically, they examined the attrition of the light /l/ and dark /ɫ/ contrast in Albanian in 10 Albanian-English late bilinguals. Impressionistic and acoustic analyses (i.e., measurements of F1 and F2 frequencies) indicated that one late bilingual completely neutralized the phonemic contrast in her native Albanian speech. Moreover, there was evidence of neutralization in coda position in two other bilinguals. Furthermore, there appeared to be a stronger trend for light /l/ to become dark in coda position than for dark /ɫ/ to become light in onset position. Overall, the authors suggested that the restructuring of one’s native language phonology continues to be malleable outside of the critical period. To the best of our knowledge, only two studies have examined geminate change in bilingual speakers. A perceptual study by Celata and Cancila (2010) investigated the discrimination of the geminate-singleton contrast in Italian in three different groups: a monolingual Lucchese-dialect speaking group, a first-generation immigrant group (i.e., those who immigrated to the United States in adulthood), and a second-generation group (i.e., individuals who were born in the U.S. to Lucchese parents), in addition to a native Lucchese Italian group that served as a control group. Two perceptual tasks were administrated. The first task was a real-word discrimination task aimed at examining the performance of the three groups with respect to abilities to discriminate sound contrasts based on their phonological function. A second task involved a non-word discrimination test intended to measure participants’ abilities to discriminate geminate-singleton sounds by tuning in to subtle acoustic differences. Results showed progressive perceptual impairment among the three groups when discriminating between singleton and geminate consonant contrasts; that is, the second- generation bilingual group performed more poorly on both perception tasks compared to the monolingual and first-generation groups. While the performance of the first-generation group was better than that of the second-generation group on the real-word discrimination task, individual differences within the former group emphasized a higher error percentage compared to that of the second-generation group. The authors argued that these individual differences were due to differences in participants’ phonological competence, in addition to external linguistic factors such as education. The performance of the first-generation group on the non-word task, however,

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 6 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

was consistent with that of the control group. As for the performance of the second-generation group, the authors claimed that the poor perceptual behavior of subjects was caused by influence from the American English phonological system, where length is phonologically non-contrastive. We highlight this study because its findings are particularly relevant to our study in that they demonstrate the effect of generation as a predictor of L1 attrition. On the other hand, another perception study, by Mazzaro, Cuza, and Colantoni (2016), examined cross-linguistic effects on the speech of long-term Spanish-speaking immigrants as well as heritage speakers of Spanish residing in the US, while also including a control group. A discrimination task was administrated to assess participants’ abilities to discriminate between Spanish stop consonant contrasts (i.e., voiced versus voiceless) and vowel contrasts (i.e., mid versus high). Interestingly, the performance of the long-term immigrant group differed significantly with regard to error rate and response times from that of the control group; however, the long-term immigrant group’s performance was similar to that of the heritage speakers, suggesting a potential general effect of exposure to English, the societally dominant language.

A recent study by Rafat et al. (2017) examined whether the geminate-singleton consonant length contrast attrites in the speech of Farsi-English bilinguals living in Canada. Similar to Celata and Cancila (2010), the speech of three different groups of participants was examined: a first- generation group who immigrated to Canada in adulthood, a 1.5-generation group that consisted of individuals who immigrated to Canada between the ages of 5 and 14 with their Iranian parents, and a second-generation group of individuals born in Canada. A word-naming task in Farsi revealed that across successive generations, gemination exhibited gradual attrition. Specifically, the following hierarchies were observed, in which geminate duration relative to singleton duration decreased based on age of immigration: homeland > first-generation > 1.5-generation > second- generation. The percentage overlap between singleton and geminate categories increased following age of immigration: homeland < first-generation <1.5- generation < second-generation. There was also a negative correlation between age of arrival and geminate length. Geminate- singleton loss was attributed to contact with English. Moreover, there was evidence suggesting that geminate production was constrained by universal phonetic factors, such as manner of articulation and voicing; that is, sonorants exhibited the shortest durations as opposed to non- continuant obstruents, which were the longest, and voiced geminates were shorter than voiceless ones. Despite these findings, there was no evidence, suggesting that these factors constrained the degree of geminate attrition across generations. The lack of effect found in this study might be attributed to its small sample size (i.e., 11 subjects, including controls).

The present study, similar in design to Rafat et al. (2017), examines geminate attrition in a different group of participants, namely Arabic-English bilinguals living in Canada, whose production data may exhibit a different pattern of geminate attrition because gemination is lexically more frequent in Arabic than in Farsi. This more frequent geminate-singleton contrast in Arabic may surface as a factor that contributes to a different pattern of language attrition in bilinguals.

2.5 Arabic Heritage/Bilingual Phonology Although we are not aware of a large body of literature on Arabic heritage/bilingual phonology, some scholars have examined consonant and vowel production in adult and child bilinguals. Khattab (2000) was one of the pioneering studies on Arabic-English bilingual child phonology. She tested VOT production in three English-Arabic bilingual children and three monolingual

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 7 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

controls from each language. The participants were all between 5 and 10 years old and were tested in England and Lebanon. The study aimed to investigate whether bilinguals acquire separate VOT patterns for each language and whether these patterns are parallel to monolingual ones. Results showed that the three bilinguals did acquire distinct VOT patterns for each language, but the patterns did not always resemble monolingual ones. The feature that was mostly affected was voicing lead in Arabic, as it was often replaced with short lag. On the other hand, similar behavior was found in the young monolingual children. Therefore, the authors attributed some of the bilingual patterns to normal developmental features rather than language interference. The results suggested that learners develop two separate systems and highlighted the importance of input.

Alkhudidi, Hakooz, Walker, Stevenson, and Rafat (in press) also examined early bilingual Arabic- English phonology. Specifically, they investigated the production of some of the marked Arabic consonants by five early English-Arabic bilingual children who lived in Ontario, Canada and who were between the ages of 8-13. Specifically, they investigated whether the participants were able to acquire the F2 lowering associated with the target marked sounds. The phonological contrasts they tested consisted of the following: (a) the emphatic-plain contrasts /t̠,/-/t/ (e.g. /ta:ba/ ‘repented’ versus /taba/ ‘healed’), /ḏ/-/d/ (e.g. /ḏa:l/ ‘stray’ versus /dal/ ‘indexical’), /s̠/-/s/ (e.g. /sa:r/ ‘became’ versus /sa:r/ ‘walked’); (b) the pharyngeal-glottal contrast /ħ/-/h/ (e.g. /ħaram/ ‘sanctuary’ versus /haram/ ‘pyramid’); and (c) the uvular-velar contrast /q/-/k/ (e.g. /qalb/ ‘heart’ versus /kalb/ ‘dog’). The results showed that early Arabic-English bilingual children were indeed able to exhibit F2 lowering, however, the degree of F2 lowering was not always target-like/within the expected norm. Moreover, variation in the degree of F2 lowering in the bilingual children in this study was constrained by the type of consonant in that /q/, /ħ/ and /ḏ / were easier than /s̠/ and /t̠,/. Also, similar to Khattab (2000), there was no evidence of age effects. This was attributed to a lower degree of exposure to the Arabic language and possibly the small sample size.

Saadah (2011) examined vowel production by Arabic-English adult bilingual/heritage speakers. She tested the production of Arabic /i: u: a:/ and English /i ɪ u ʊ æ/. Thirty participants were divided into three main groups: native speakers of Arabic, heritage speakers of Arabic, and L2 learners. The heritage group consisted of inexperienced heritage speakers and experienced heritage speakers of Arabic and the L2 group consisted of beginner and advanced learners. The participants were asked to read two lists of Arabic and English words in carrier phrases, with F1 (height), F2 (frontness/backness), and duration being reported. All investigated groups produced distinct Arabic vowels. Heritage speakers aligned with native speakers in F1 and in F2 for /i: i/. For /u:/, heritage speakers performed more similarly to native speakers in F1 (and to L2 learners) but were more similar to L2 learners in F2. For /u/, all three groups exhibited similar F2 values, with heritage speakers producing intermediary F1 values, between those of native speakers and L2 learners. With respect to low vowels, heritage speakers matched L2 learners in F1 but had intermediary values for F2, between those of native speakers and L2 learners. As for duration, there was a consistent pattern across the three groups in that native speakers produced the shortest duration values for short vowels, followed by heritage speakers, and finally L2 learners, who had the longest duration values. Native speakers also had the longest durations for high long /i:/ and /u:/, whereas the heritage speakers produced low /a:/ similarly to native speakers. These findings indicated that native speakers produced a greater durational contrast between long and short vowels, which was shown through shorter short vowels and longer long vowels than those of heritage speakers and L2 learners. In all, based on the F1, F2, and duration results, it was concluded that heritage speakers

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 8 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

did, in fact, pronounce Arabic vowels differently from those of L2 learners, while also having a distinct vowel system from that of native speakers of Arabic. In other words, heritage speakers fell between native speakers and L2 learners on a vowel production spectrum.

Although the review of the above studies show that Arabic-English bilingual speech has received some attention, we are not aware of any previous studies that have examined geminate attrition in Arabic-English bilinguals.

2.6 Hypotheses Based on the research detailed in previous sections, the hypotheses guiding this study are as follows:

H1. Because L1 and L2 'similar' categories may assimilate (Flege, 1995), and because there are no geminates in intervocalic position in English, Arabic geminates will undergo attrition in bilinguals. Specifically, a geminate-singleton duration ratio continuum will be formed, where Arabic monolinguals have the highest values, followed by late learners, and then early bilinguals (Celata & Cancila, 2010; Rafat, et al., 2017). H2. Manner of articulation and/or voicing will impact mean consonant duration. a. Across both groups of bilinguals, sonorant geminates will exhibit a higher degree of consonant reduction (i.e., a decrease in geminate-singleton duration ratios) than fricatives, with the least reduction occurring in stops (Podesva, 2002; Sorianello, 2014; Steriade, 1982; Taylor, 1985; Rafat et al., 2017). b. Voiced geminates will exhibit a higher degree of consonant reduction than voiceless consonants across both groups of bilinguals (Hayes & Steriade, 2004; Jaeger, 1978; Kawahara, 2007; Ohala, 1983; Sorianello, 2014; Taylor, 1985; Rafat et al., 2017). c. The longer the geminates, the higher the impact on the degree of consonant reduction (Varona et al., 2017).

3. METHODOLOGY 3.1 Participants There were 15 participants (see Appendix A) in this study, separated into three groups. The participants were recruited by posters and word of mouth in the community in London, Ontario, Canada. The late bilingual group consisted of five adult-immigrant Arabic-English bilinguals who learned English as an L2 in late adulthood and immigrated to Canada after puberty. There was one male and four females, ranging between the ages of 46 and 60, with a mean age of 53.8. According to their responses to a language background questionnaire (see Appendix B), all five participants had some exposure to English before immigration, mainly in school or work settings; however, Arabic was mostly used, around 80-95% of the time, before they came to Canada. All members of this group have high levels of education (i.e., bachelor’s degree or above). These participants also reported that after immigration, they used English 95% of the time at work and 50-60% of the time in social settings. The early bilingual group consisted of five adult Arabic-English early bilinguals who are heritage speakers of Arabic. These participants consisted of children of Arabic immigrants who were either born in Canada or arrived before the age of five. They are all females and their ages ranged from 24 to 45, with a mean age of 30.2. Participants in this group reported proficiency in both Arabic and English, although they indicated a higher proficiency in English (based on a

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 9 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

scale from 1 ‘beginner’ to 7 ‘native’), their preferred language of communication (i.e., 59-90% of the time at work/school and 80-90% of the time in social settings). The control group consisted of five adult Arabic monolinguals who lived in Saudi Arabia. It was comprised of three males and two females ranging in age from 24 to 60, with a mean age of 42.

3.2 Stimuli The stimuli design from Rafat et al. (2017) was adopted in this study. The stimuli consisted of 158 bi-syllabic Arabic words (see Appendix C). There were 40 minimal pairs (e.g., /ħamːa:m/ ‘bathroom’ vs. /ħama:m/ ‘pigeon’) and 39 near-minimal pairs (e.g., /darːab/ ‘trained’ vs. /darab/ ‘hit’). Both geminate and singleton words included three classes of sounds: stops (/b, d, t, k, t̠, q/), fricatives (/f, ð, ʕ, s /), and sonorants (/m, n, r, l, j/). The stimuli were controlled for position (i.e., intervocalic position) and stress (i.e., on the second syllable). A total of 12 monosyllabic and tri- syllabic distractors were used to divert participants’ attention from any possible common pattern effect. The stimuli were recorded by the first author, who is a female native speaker of Arabic originally from Saudi Arabia with near-native fluency in English, and presented via PowerPoint. 3.3 Procedure A delayed word repetition task was administered. Participants were seated in a quiet room and presented with the auditory and orthographic forms of the target words in Arabic via a PowerPoint presentation on a laptop screen. Each target word was embedded in the carrier phrase /aʔana ʔquːl _ ʔlʔaːn/ ‘I say _ now.’ Participants listened to and saw each phrase simultaneously. The phrase then disappeared from the screen and participants were asked to first count backward from 7 to 1 and then repeat the phrase they had just heard. Backward counting was used to minimize traces of phonological input from memory (Bassetti, 2017). Participants’ production was recorded using an M-Audio Micro-track, 24/96 professional two-channel mobile digital recorder and a lavaliere unidirectional microphone. The recordings were made at a sampling rate of 44.2 kHz and a quantization rate of 16 bits. The audio files containing the extracted tokens were down-sampled at 22.1 kHz and saved in .wav format. Participants completed a language background questionnaire (see Appendix B) upon the completion of all trials. They were recorded individually and each session lasted 60-90 minutes. The first author conducted the experiment with both the bilinguals in Canada and the monolinguals in Saudi Arabia.

3.4 Data Analysis A total of 2,370 tokens were analyzed acoustically by the first author using Praat (Boersma & Weenink, 2012). Consonant duration was measured for each geminate and singleton consonant. Acoustically, this was done by measuring the closure duration of each stop consonant, the onset and offset of frication noise for fricatives, the onset of nasal murmur to the oral closure release for nasals, and the onset and offset of changes in the spectrogram and the shape and amplitude of formant frequencies for liquids (Rafat et al., 2017). Figure 1 illustrates singleton length in the word /safar/ ‘the second month in the Islamic calendar,’ which did not significantly change in a monolingual, late bilingual, and early bilingual participants. Figure 2 illustrates geminate duration in the word /safːar/ ‘whistled’ in a monolingual, late bilingual, and early bilingual participant, where geminate duration decreases across the three generations.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 10 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Figure 1.

Sample Waveforms and Spectrograms of the Word /safar/ ‘the second month in the Islamic

calendar.’

Frequency(Hz)

Time (s)

Note. The area within the two dotted lines indicates the interval corresponding to the duration of the singleton consonant /f/ in a token produced by a participant from: a) the control group (87 ms); b) the late bilingual group (90ms); and c) the early bilingual group (92 ms).

Figure 2. Sample Waveforms and Spectrograms of the Word /saf:ar/ ‘whistled.’

Frequency(Hz) Time (s)

Note. The area within the two dotted lines indicates the interval corresponding to the duration of the geminate consonant /f:/ in a token produced by a participant from: a) the control group (202 ms); b) the late bilingual group (185ms); c) the early bilingual group (168m).

4. RESULTS R 4.1 The Impact of Early versus Late Bilingualism Mean durations were calculated for each individual and for each consonant produced by each individual. Additionally, for each participant the geminate-singleton ratio was calculated for each consonant. Ratio data is dependent upon absolute geminate and singleton durations, but has the effect of normalizing geminate duration to singleton duration to ensure that changes across generations cannot be accounted for by an overall change in duration for all speech productions, but rather are specific to geminates.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 11 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

We used SPSS to conduct all of our statistical analyses. Our initial analysis focused on group and length, and as such, responses were averaged across consonants. A two-way mixed-model ANOVA examined the interaction between group*length with absolute duration as the dependent 2 variable. A main effect of length was found (F(1) = 984.379, p < 0.001, ƞp = 0.988), as well as of 2 group (F(2) = 8.106, p = 0.006, ƞp = 0.575), with geminates being longer than singletons (see Figure 3 dark versus light bars), and with monolinguals, late bilinguals, and heritage speakers decreasing in absolute duration (see Figure 3, all bars from left to right). A significant group*length 2 interaction was also observed (F(2,12) = 17.597, p < 0.001, ƞp = 0.746) such that Arabic monolingual controls showed a greater difference between geminates and singletons than late bilinguals, and late bilinguals a greater difference than early bilinguals. To explore the group*length interaction, a one-way ANOVA was conducted with ratio data across groups (see 2 Figure 3, line). There was a significant effect of group (F(2) = 14.594, p = 0.001, ƞp = 0.709) such that the ratio was ordered, decreasing respectively from monolinguals to late bilinguals to early bilinguals.

Figure 3

The Impact of Group on Geminate Attrition

4.2. The Effect of Consonant Class To explore the possible impact of consonant class, data were then divided into stops, sonorants, and fricatives (see Rafat et al. 2017). To do so, we conducted a three-way mixed-model ANOVA (group*length*class) with absolute duration as the dependent variable (see Figure 4, bars). Main 2 2 effects of group (F(2) = 7.323, p = 0.008, ƞp = 0.550), length (F(1) = 941.394, p < 0.001, ƞp = 2 0.987), and class (F(2) = 39.461, p < 0.001, ƞp = 0.767) were observed, with fricatives having the longest durations and sonorants the shortest. As reported above, a significant group*length effect 2 was observed (F(2,12) = 17.464, p < 0.001, ƞp = 0.744), but neither group*class (F(4,24) = 2.291, 2 2 p = 0.089, ƞp = 0.276) nor length*class (F(2,4) = 0.806, p = 0.458, ƞp = 0.063) interactions were significant. The three-way, group*length*class interaction was also non-significant (F(4,24) = 2 1.494, p = 0.235, ƞp = 0.199).

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 12 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Ratio data were also explored with a repeated-measures ANOVA (group*class; see Figure 4, 2 lines). There was a significant effect of group (F(2) = 15.878, p < 0.001, ƞp = 0.726) such that the ratio decreased across the three groups, and a significant effect of class (F(2) = 26.467, p < 0.001, 2 ƞp = 0.688), with sonorants showing the highest ratio and fricatives the lowest. No significant 2 group*class interaction was observed (F(4,24) = 2.129, p < 0.108, ƞp = 0.262).

Figure 4.

The Impact of Class of Sounds on Geminate Attrition

4.3 The Impact of Voicing To explore possible impacts of voicing (see Rafat et al., 2017), data was then divided into voiced and voiceless consonants, and a three-way mixed-model ANOVA (group*length*voicing) with absolute duration as the dependent variable was conducted (see Figure 5, bars). Main effects of 2 2 group (F(2) = 7.833, p = 0.007, ƞp = 0.566), length (F(1) = 828.864, p < 0.001, ƞp = 0.986) and 2 voicing (F(1) = 127.091, p < 0.001, ƞp = 0.914) were observed, with voiceless consonants having longer durations than voiced ones. As reported above, a significant group*length effect was 2 observed (F(2,12) = 14.864, p = 0.001, ƞp = 0.712), but neither group*voicing (F(2,12) = 1.609, 2 2 p = 0.240, ƞp = 0.212) nor length*voicing (F(1,12) = 0.996, p = 0.338, ƞp = 0.077) were

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 13 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

significant. The three-way group*length*voicing interaction was also non-significant (F(2,12) = 2 0.921, p = 0.424, ƞp = 0.133).

Ratio data were also explored with a repeated-measures ANOVA (group*voicing; see Figure 5, 2 lines). There was a significant effect of group (F(2) = 12.752, p < 0.001, ƞp = 0.680) such that the 2 ratio decreased across the groups, and a significant effect of voicing (F(1) = 75.621, p < 0.001, ƞp = 0.863), with voiced consonants exhibiting a greater ratio than voiceless ones. Interestingly, a 2 significant group*voicing interaction was observed (F(2,12) = 5.903, p = 0.016, ƞp = 0.496); there was a larger decrease across the groups in voiced relative to voiceless consonants.

Figure 5.

The Impact of Voicing on Geminate Change

4.4 The Impact of Individual Consonants Data were also divided into individual consonants (see Varona et al., 2017), and a three-way mixed-model ANOVA (group*length*consonant) with absolute duration as the dependent 2 variable was conducted (see Figure 6, bars). Main effects of group (F(2) = 7.800, p = 0.007, ƞp = 2 0.565), length (F(1) = 942.424, p < 0.001, ƞp = 0.987) and consonant (F(14) = 43.882, p < 0.001, 2 ƞp = 0.785) were observed. As reported above, a significant group*length effect was observed 2 (F(2,12) = 16.599, p < 0.001, ƞp = 0.735), as was a length*consonant interaction (F(14,28) = 2 3.267, p < 0.001, ƞp = 0.214). No significant group*consonant interaction was observed 2 (F(28,168) = 1.304, p = 0.155, ƞp = 0.179). The three-way group*length*consonant interaction 2 was also non-significant (F(28,168) = 1.249, p = 0.283, ƞp = 0.172).

A repeated-measures group*consonant ANOVA was conducted with ratio data across groups and consonants (see Figure 6, lines). There was a significant effect of group (F(2) = 14.357, p = 0.001, 2 ƞp = 0.705) such that the ratio decreased with each successive group, and a significant effect of 2 consonant (F(14) = 18.853 p < 0.001, ƞp = 0.611). Interestingly, a significant group*consonant 2 interaction was observed (F(28,168) = 1.792, p = 0.013, ƞp = 0.230).

In the next step, which explored the group*length*consonant interaction, a repeated-measures two- way ANOVA (group*length) was conducted with absolute durations, followed by a one-way ANOVA across groups with ratio data for each consonant (see Table 1 for detailed statistics). For

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 14 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

all 15 consonants, a significant main effect of length was observed in the two-way ANOVA, with 2 each p < 0.001, F-values ranging from 86.462 to 619.832, and ƞp s ranging from 0.878-0.981. Main effects of group were observed for the stops [b], [d], [t], [k], and [tʕ], and the fricatives [ʕ] and [s]. Interactions between group and length were observed for the stops [b], [d], [t], [k], and [tʕ], the sonorants [r] and [l], and the fricatives [ð], [ʕ], and [s]. ANOVAs with normalized ratio data revealed significant changes in ratio across groups for the stop [d], the sonorants [r], [l], and [j], and the fricatives [ʕ] and [s].

Table 1.

Group*Length ANOVAs for Individual Consonants

Group (ms) Length*Group (ms) Group (ratio values) 2 2 2 Consonant F(2) P ƞp F(2,12) p ƞp F(2) p ƞp [b] 6.554 0.012 0.522 4.892 0.028 0.449 3.146 0.080 0.344 [d] 3.983 0.047 0.399 6.485 0.012 0.519 4.073 0.045 0.404 [t] 4.620 0.033 0.435 5.777 0.017 0.491 3.334 0.071 0.357 [k] 4.857 0.028 0.447 5.042 0.026 0.457 3.195 0.077 0.348 [tʕ] 6.028 0.015 0.501 6.318 0.013 0.513 1.943 0.186 0.245 [q] 3.612 0.059 0.376 1.517 0.259 0.202 0.501 0.618 0.077 [m] 1.3903 0.191 0.241 2.019 0.175 0.252 2.442 0.129 0.289 [n] 0.401 0.678 0.063 3.684 0.057 0.380 3.741 0.055 0.384 [r] 1.706 0.223 0.221 4.491 0.035 0.428 12.648 0.001 0.678 [l] 0.610 0.560 0.092 5.491 0.020 0.478 5.535 0.020 0.480 [j] 0.263 0.773 0.042 4.965 0.27 0.453 7.339 0.008 0.550 [f] 0.738 0.498 0.110 2.969 0.090 0.331 2.401 0.133 0.286 [ð] 2.772 0.102 0.316 5.385 0.021 0.473 3.144 0.080 0.344 [ʕ] 12.585 0.001 0.677 9.863 0.003 0.622 6.619 0.012 0.525 [s] 2.330 0.014 0.280 4.307 0.039 0.418 6.766 0.011 0.530

Figure 6.

The Impact of Type of Consonants on Geminate Attrition

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 15 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

4.5 The Impact of Age of Arrival and Length of Residency To assess the impact of AoA (see Rafat et al., 2017; Varona et al., 2017) and length of residency (LoR), these demographic variables were correlated with mean durations for geminates and singletons collapsed across consonants, as well as with ratios in late bilingual and early bilingual participants (control participants could not be included as they have no AoA or LoR). The variable AoA was significantly correlated with geminate duration such that an early AoA was associated with shorter geminate duration (r(9) = 0.909, p < 0.001), while AoA was not related to singleton duration (r(9) = 0.316, p = 0.374; see Figure 7). This variable also trended towards significantly correlating with ratio (r(9) = 0.615, p = 0.058) such that a younger AoA was associated with smaller geminate-singleton duration ratios. The variable LoR did not significantly correlate with geminate duration (r(9) = -0.446, p = 0.154), singleton duration (r(9) = -0.230, p = 0.522) or ratio (r(9) = -0.318, p = 0.370).

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 16 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Figure 7.

The Impact of AoA on Geminate Change

4.6 The Impact of Consonant Length Although length is considered a salient feature in phonology (see Bohn, 1995; Cebrian, 2006; Escudero, 2001; Kondaurova & Francis, 2008, 2010; Morrison, 2008; Ruíz-Peña, Sevilla, & Rafat, 2015), we are aware of only one study that has examined the effect of inherent length on the acquisition/production of L2 and D2 geminates. Varona et al. (2017) found an inverse relationship between length and the degree of accurate production of L2 and D2 geminates. This study, therefore, would suggest that inherent consonant duration may impact changes to duration such that consonants with longer geminate durations will be impacted to a greater extent than those with shorter geminate durations. To test this in our data, the changes in geminate duration for each consonant across groups (i.e., monolingual versus early bilinguals, monolingual versus late bilinguals, and late bilinguals versus early bilinguals) were correlated with the absolute duration measured in the monolingual group (see Figure 8, left panel). Changes in duration between monolinguals and both late bilinguals (r(9) = 0.540, p = 0.038) and early bilinguals (r(9) = 0.525, p = 0.045) were significantly correlated with initial consonant duration, but the change between late bilinguals and early bilinguals was not (r(9) = 0.163, p = 0.563). A similar analysis was conducted with ratio data (see Figure 8, right panel), where changes in ratio between control and

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 17 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

both late bilinguals (r(9) = 0.055, p = 0.034) and early bilinguals (r(9) = 0.797, p < 0.001) were significantly correlated with geminate ratios of the monolinguals, but the change between late bilinguals and early bilinguals was not (r(9) = 0.377, p = 0.166). In summary, consonants with longer geminate durations showed greater changes in absolute geminate duration and ratio values between the monolingual group and both bilingual groups, but not between late bilinguals and early bilinguals.

Figure 8.

The Impact of Inherent/Initial Length on Geminate Change Across Groups

5. DISCUSSION The first aim of this study was to investigate whether geminate production in the speech of Arabic- English bilinguals residing in Canada would undergo attrition. The study investigated geminate attrition across two groups of bilinguals: late bilinguals and early bilinguals. It was hypothesized that geminate attrition would be observed in the speech of these bilinguals. More specifically, based on Celata and Cancila (2010) and Rafat et al., (2017), it was predicted that a continuum in

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 18 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

terms of a geminate-singleton ratio would be formed, where Arabic monolinguals would have the highest values (i.e., clearest distinction) followed by late bilinguals, and then early bilinguals.

The second aim of the study was to test whether manner of articulation and voicing has an impact on the degree of geminate attrition. Based on Podesva (2002), Rafat et al. (2017), Sorianello (2014), Steriade, (1982), and Taylor (1985), it was hypothesized that manner of articulation would impact the degree of change with respect to mean consonant duration; that is, sonorant geminates would exhibit a higher degree of attrition than fricatives, followed by stops, across generations of bilinguals. Moreover, based on Hayes and Steriade (2004), Jaeger (1978), Kawahara (2007), Ohala (1983), Rafat et al. (2017), Sorianello (2014), and Taylor (1985), it was predicted that voiced geminates would exhibit a higher degree of attrition than voiceless ones across these generations. Furthermore, based on Varona et al. (2017), it was predicted that length would impact degree of geminate attrition.

The first hypothesis was confirmed, as the results provided here showed that the L1 geminate- singleton phonemic contrast did show evidence of undergoing attrition in both late Arabic-English bilinguals and early bilinguals (i.e., heritage speakers) of Arabic. Moreover, a continuum in terms of geminate-singleton ratio was formed, whereby the Arabic monolinguals had the highest ratio, followed by the late bilinguals, and then early bilinguals; in other words, geminates underwent attrition in both bilingual groups, albeit they were shorter in early bilinguals. This finding is consistent with Rafat et al. (2017), who reported evidence of geminate attrition in the speech of three successive generations of Farsi-English bilinguals. Although the geminate-singleton contrast is lexically more frequent in Arabic, geminates still exhibited a robust degree of attrition in our Arabic-English bilinguals living in Canada. The present study is also consistent with previous literature (e.g., Celata & Cancila, 2010; Rafat et al., 2017) illustrating the effect of bilingualism on degree of attrition. The evidence for attrition in late bilinguals supports the view that underlying representations are malleable, suggesting that the restructuring of L1 grammar is possible outside of the critical period, consistent with de Leeuw et al. (2018). Degrees of geminate attrition would have to be contrasted with degrees of attrition of other types of sounds in order to verify whether geminates’ markedness makes them more prone to attrition.

As predicted, geminates were shorter in early bilinguals in comparison to late bilinguals (Celata & Cancila, 2010; Hrycyna, Lapinskaya, Kochetov, & Nagy, 2011; Rafat et al., 2017). The effect of AoA, also previously reported in Rafat et al. (2017), was also consistent with the effect of participant group. Speech in the early bilingual group seems more likely to change than in that of the late bilingual group, most likely due to the effect of language dominance, less frequent use of the L1 and/or or insufficient L1 input. All the early bilinguals in this study were English-dominant, and as such, it will be interesting to compare their geminate production with that of L2 learners of Arabic in the future to see whether their reduction in geminate duration patterns more with native speaker productions of geminates or those of L2 learners. A longitudinal study would also be interesting, and would provide more insights on how geminate production develops in early bilinguals, which would, in turn, provide further insight into the degree of brain plasticity over the lifespan and its relationship with language experience.

We have provided evidence for attrition of a phonological contrast in the L1 as a result of contact with English, where this contrast is only phonetic. We have also shown that geminates are shorter

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 19 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

for early bilinguals. It is important to point out that attrition with respect to L1 contrastive categories has also been reported in other cognitive domains, such as color perception. These studies have also shown a shift in the cognitive representation of bilinguals towards the categories of monolinguals of their L1. This has been reported for Greek-English bilinguals (Athanasopoulos, 2009) and Japanese-English bilinguals (Athanasopoulos, Damjanovic, Krajciova, & Sakai, 2011). Unlike English, Greek differentiates the blue region of color space into a lighter shade (ghalazio) and a darker shade (ble). Athanasopoulos (2009) reported a semantic shift of category prototypes with level of bilingualism. Similar to Greek, Japanese also contrasts with English in that it has an additional term (mizuiro) for ‘light blue.’ Athanasopoulos et al. (2011) also showed that Japanese– English bilinguals displayed a cognitive pattern that was “in-between” that of the Japanese and English monolingual groups. Moreover, those bilinguals who used English more frequently distinguished blue and light blue less well than those who used Japanese more frequently, suggesting that bilingual cognition may be dynamic and dependent on language use.

The second hypothesis in this study was that phonetic/acoustic factors, such as manner of articulation and voicing, constrain the degree of attrition across the groups. This hypothesis was partially confirmed; that is, while there was no effect of class of sound on degree of attrition, voicing was found to significantly impact geminate attrition, contrasting with the results from Rafat et al. (2017). More specifically, there was a larger decrease in geminate attrition in voiced consonants relative to voiceless ones. Our results are novel for phonological attrition and pattern with universal geminate typological tendencies, suggesting that the extent of malleability may be determined by universal phonetic principles that affect language change in monolinguals. Voiceless geminates are more common across world languages than voiced ones (Hayes & Steriade, 2004; Jaeger, 1978; Kawahara, 2007; Ohala, 1983; Taylor, 1985), a tendency that is attributed to the fact that it is difficult to maintain voicing during a long closure. Geminate sonorants are also claimed to be more marked than stop and fricative geminates, mainly because the length distinction between geminates and singletons in sonorants is difficult to perceive accurately, thus making it less salient (Kawahara, 2007; Podesva, 2002; Taylor, 1985), and therefore, more susceptible to attrition. Interestingly geminate stops were less likely to be acquired by L2 learners in Sorianello (2014), highlighting the effect of universal phonetic principles on L2 speech learning.

Based on Varona et al. (2017), we had also predicted that consonant length, a salient phonetic feature, would impact geminate reduction. Indeed, we found that consonant length was correlated with rate of change, albeit only the change between the control group and late bilinguals and early bilinguals. This is consistent with the findings of Varona et al. (2017), where Colombian Spanish speakers exhibited more target-like productions when imitating shorter geminates in an L2 (i.e., Italian) and a D2 (i.e., Cuban Spanish). Varona et al. (2017) proposed that although it may be easier to perceive longer consonants, it may be more difficult to maintain length. Our data also suggest that there may be a trade-off between acoustic salience (i.e., length) and accurate production in geminate attrition. This is another instance where we see the same universal factor affecting both L1 attrition and L2 acquisition.

The findings in this study also suggest that attrition of the geminate-singleton length contrast and a shift towards English are gradual processes that may be caused by several factors. Given the evidence in this study for Arabic-English bilinguals, as well as the evidence provided by Celata

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 20 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

and Cancila (2010) for Italian-English bilinguals and Rafat et al. (2017) for Farsi-English bilinguals, we put forth that it is highly probable that geminate attrition occurs universally when an L1 with a geminate-singleton contrast comes in contact with English, which is void of this phonological contrast. Previously, Henriksen (2015) examined Spanish tap-trill attrition across late and early Mexican-Spanish-English bilinguals living in the United States. The Spanish tap and trill are contrastive because, whereas the tap is short (about 20-40ms), the trill is long (80-180ms). Moreover, the tap is characterized by a single occlusion, but the trill may have between 2-5 occlusions. Henriksen’s (2015) results also showed that trills undergo length reduction, although all participants maintained a duration contrast between the tap and trill. Similar to the overall results of our study, the trill durations in Henriksen (2015) were shorter (i.e., 65-80ms) in both the late and early bilinguals than what has been reported for the Spanish trill duration in the literature. Given that more than one feature is used to make the tap and trill contrastive, it might be that, diachronically speaking, the number of closures might be neutralized earlier than duration, when the trill undergoes attrition. Colantoni and Rafat (2013) also found that the length contrast was maintained between the tap and the trill in Argentina when the contrast was undergoing sound change. In our current study, we noticed several instances of complete degemination in the data, not only for the trill but also for the other geminates, where a geminate sound was produced as a singleton sound by early bilinguals; however, we did not analyze them in the current paper due to time constraints. We predict that the process of degemination might be completed, albeit in successive generations or later during the life span of the participants examined in this study or in other immigrant communities. While we believe that contact with English is the main cause of attrition in this study, we have also shown that generation, voicing, and consonant length affected geminate change in our Arabic-English bilinguals. The magnitude of effects, however, may also be modulated by other factors not examined here, especially sociolinguistic factors such as language attitude, language use, and level of contact with the dominant language community. For example, Hrycyna et al. (2011) reported that the negative attitude of Italian first-generation speakers toward their L1, which caused them to place more value on integrating into an English- speaking society, could be one factor contributing to the drift in their VOT values towards English. According to the sociolinguistic information reported by the first-generation bilingual group in the current study, they have a strong preference for speaking Arabic on a daily basis and their use of English was limited to work or during non-Arabic events. Therefore, it is possible that even a more robust degree of geminate duration change may be observed in the first generation of communities in which speakers lose their cultural and social ties to their heritage culture more quickly.

As previously mentioned, although there are a number of prominent models of L2 phonological acquisition (Best & Tyler, 2007; Brown, 1998; Colantoni & Steele, 2008; Escudero, 2005; Flege, 1987, 1995), so far, Flege’s SLM (1987, 1995) has been the main (if not the only) model applied to bilingualism. The SLM outlines the potential bidirectional influence of L1 and L2 sounds and considers the effect of the acoustic/phonetic distance between L1 and the L2 sounds. Although there was no evidence to suggest that manner of articulation was a predictor of language change in our bilinguals, we have found evidence to support that both voicing and inherent/initial consonant length are predictors of attrition. Thus, we believe the effect of universal phonetic factors, such as aerodynamic constraints (Ohala, 1983), articulatory difficulty/complexity (Diehl & Lindblom, 2004), articulatory precision (Ladefoged & Maddieson, 1996), and perceptual constraints (e.g., Kawahara, 2007; Podesva, 2002; Taylor, 1985), should be further investigated and incorporated into models of L1 attrition.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 21 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

6. CONCLUSION To summarize, the current study investigated the production of geminate-singleton contrasts in early and late Arabic-English bilinguals, and compared both groups to data from Arabic monolinguals. Similar to Rafat et al. (2017), evidence of geminate attrition was found in both bilingual groups, supporting the claim that language remains somewhat dynamic and malleable across the lifespan. The findings in this study highlight the role of universal phonetic principles and pattern with findings in L2 speech learning. They also point to the possibility that this type of geminate change may happen in the grammar of the speakers of any language with gemination that may come in contact with a language that does not have geminates.

We also believe that the effect of positional factors and prosody (e.g., stress) on geminate attrition merits investigation. Previously, these factors have predicted consonant weakening on a fortis~lenis continuum (see Lavoie, 2015 for more details). Colantoni and Steele (2008) and Rafat (2011) also reported positional asymmetries attributed to acoustic prominence in L2 production. Additionally, Rafat (2011) discussed positional differences in L2 production patterns in lieu of phonological memory. Importantly, when de Leeuw et al. (2018) examined the attrition of the light-dark /l/ contrast in Albanian-English bilingual immigrants, they found that there was a stronger tendency for light /l/ to become dark in coda position than for dark /ɫ/ to become light in onset position. Therefore, it is likely that, consistent with universal patterns, geminates would also exhibit a higher degree of attrition in coda position than in the onset position (e.g., those examined in this study). Future research can also examine the effect of geminates on the preceding vowel (Esposito & Di Benedetto, 1999; Ham, 2001; Hassan, 2003; Homma, 1981; Lahiri & Hankamer, 1988). In addition, future work can also investigate individual differences with respect to geminate attrition. Finally, it would be interesting to examine the effect of social factors such as language attitude (Cherciov, 2013), language identity, and degree of community involvement and contact (Prescher, 2007) on degree of attrition, as well as the role of parental linguistic input (see Rao, 2016).

Overall, this study provides an empirical and theoretical contribution to the growing field of phonological attrition by examining restructuring and change in an understudied bilingual community in Canada. The main limitation of this study is its sample size. Moreover, although it provides evidence for significant geminate length reduction across early and late bilinguals, it does not report on degree of complete degemination. We encourage future researchers to address in more detail cases of geminate reduction and complete degemination, and ideally, include consecutive generations of bilinguals while focusing on other individual and social factors to see when geminate attrition is completed and what other factors may be involved.

REFERENCES Abercrombie, D. (1967). Elements of general phonetics. Edinburgh, Scotland: Edinburgh University Press. Abu-Abbas, K.H. (2011). Geminates and long consonants in Jordanian Arabic. International , 3(1), 3-17.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 22 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Alkhudidi, A., Walker, M., Hakooz, Y., Stevenson, R., & Rafat, Y. (in press). The production of marked Arabic geminates by heritage speakers living in Canada. In Babastoli, E., & Ball, M. J. (Eds.), An anthology of bilingual child phonology. Bristol, UK: Multilingual Matters. Al-Tamimi, F., Abu-Abbas, K., & Tarawnah, R. (2010). Jordanian Arabic final geminates: An experimental clinical phonetic study. Poznan Studies in Contemporary Linguistics, 46(2), 111– 125. Al-Tamimi, F. (2004). An experimental phonetics study of intervocalic singleton and geminate sonorant in Jordanian Arabic. Al-Arabiyya, 37, 37–52. Athanasopoulos, P. (2009). Cognitive representation of colour in bilinguals: The case of Greek blues. Bilingualism: Language and Cognition, 12(1), 83-95. Arvaniti, A., & Tserdanelis, G. (2000). On the phonetics of geminates: Evidence of Cypriot Greek. Proceedings of 6th International Conference on Spoken Language Processing, 2, 559–562. Athanasopoulos, P., Damjanovic, L., Krajciova, A., & Sasaki, M. (2011). Representation of colour concepts in bilingual cognition: The case of Japanese blues. Bilingualism: Language and Cognition, 14(1), 9-17. Bassetti, B. (2017). Orthography affects second language speech: Double letters and geminate production in English. Journal of Experimental Psychology: Learning, Memory, and Cognition, 43(11), 1835-1842. doi:10.1037/xlm0000417. Best, C. T., & Tyler, M. D. (2007). Nonnative and second language speech perception: Commonalities and complementarities. In Munro, M. J., & Bohn, O. S. (Eds.), Second language speech learning: The role of language experience in speech perception and production (pp. 13-34). Amsterdam, The Netherlands: John Benjamins. Boersma, P., & Weenink, D. (2012). Praat: doing phonetics by computer [Computer program]. Version 5.3. 39. Available from http://www.fon.hum.uva.nl/praat/ Bohn, O.-S. (1995). Cross-language speech perception in adults: First language transfer doesn’t tell it all. In Strange, W. (Ed.), Speech perception and linguistic experience: Issues in cross- language research (pp. 279–304). Timonium, MD: York Press. Blevins, J. (2004). Evolutionary phonology. Cambridge, UK: Cambridge University Press. Brown, C. A. (1998). The role of the L1 grammar in the acquisition of segmental structure. Second Language Research, 14, 139–193. Cebrian, J. (2006). Experience and the use of non-native duration in L2 vowel categorization. Journal of Phonetics, 34(3), 372-387. Celata, C., & Cancila, J. (2010). Phonological attrition and the perception of geminate consonants in the Lucchese community of San Francisco (CA). International Journal of Bilingualism, 14, 1–25. Chang, C. E. (2012). Rapid and multifaceted effects of second-language learning on first-language speech production. Journal of Phonetics, 40, 249-268. Cherciov, M. (2013). Investigating the impact of attitude on first language attrition and second language acquisition from a Dynamic Systems Theory perspective. International Journal of Bilingualism, 17(6), 716–733. https://doi.org/10.1177/1367006912454622 Cohn, A.C., Ham, W.C., & Podesva, R.J. (1999). The phonetic realization of singleton-geminate contrasts in three languages of Indonesia. In Ohala, J. J. (Ed.), Proceedings of the 14th International Congress of Phonetic Sciences (pp. 587-590). Berkeley, CA: University of California Press. Colantoni, L., & Steele, J. (2008). Integrating articulatory constraints in models of L2 phonological acquisition. Applied Psycholinguistics, 29, 1–46.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 23 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Colantoni, L., & Rafat, Y. (2013). Las consonantes róticas en el español argentino [Rhotic consonants in Argentine Spanish]. In Colantoni, L., & Rodríguez Louro, C. (Eds.), Perspectivas teóricas y experimentales sobre el español de la Argentina [Theoretical and experimental perspectives on Argentinian Spanish] (pp. 82–98). Madrid, Spain: Iberoamericana/Vervuert. De Leeuw E., Mennen I., & Scobbie J. (2012). Singing a different tune in your native language: L1 attrition of prosody. International Journal of Bilingualism, 16, 101-116 De Leeuw, E., Mennen, I., & Scobbie, J. M. (2013). Dynamic systems, maturational constraints and L1 phonetic attrition. International Journal of Bilingualism, 17, 683–700. De Leeuw, E., Tusha, A., & Schmid, M. (2018). Individual phonological attrition in Albanian– English late bilinguals. Bilingualism: Language and Cognition, 21(2), 278-295. doi:10.1017/s1366728917000025 Diehl, R. L., & Lindblom, B. (2004). Explaining the structure of feature and phoneme inventories: The role of auditory distinctiveness. In Greenberg, S., Ainsworth, W. A., Popper, A., & Fay, R. (Eds.), Speech processing in the auditory system (pp. 101-162). New York, NY: Springer- Verlag. Escudero, P. (2001). The role of the input in the development of L1 and L2 sound contrasts: Language specific cue weighting for vowels. In A. H.-J. Do, L. Domínguez, & A. Johansen (eds.), Proceedings of the 25th Annual Boston University Conference on Language Development (BUCLD) (pp. 250-261). Somerville, MA: Cascadilla. Escudero, P. (2005). Linguistic perception and second language acquisition: Explaining the attainment of optimal phonological categorization. Utrecht, The Netherlands: LOT. Esposito A., & Di Benedetto, M. G. (1999). Acoustical and perceptual study of gemination in Italian stops. Journal of the Acoustical Society of America, 106, 2051-2062. Flege, J. E. (1987). The production of ‘new’ and ‘similar’ phones in a foreign language: Evidence for the effect of equivalence classification. Journal of Phonetics, 15, 47-65. Flege, J. E. (1995). Second language speech learning: Theory, findings, and problems. In W. Strange (Ed.), Speech perception and linguistic experience: Issues in cross-language research (pp. 233–277). Timonium, MD: York Press. Ghalib B. M. (1984). An experimental study of consonant gemination in Iraqi colloquial Arabic (Unpublished doctoral dissertation). Leeds, UK: University of Leeds. Guion, S. G. (2003). The vowel systems of Quichua–Spanish bilinguals: Age of acquisition effects on the mutual influence of the first and second languages. Phonetica, 60, 98-128. Ham, W. (2001). Phonetic and phonological aspects of geminate timing. New York, NY: Routledge. Hansen, B. (2004). Persian geminate stops: Effects of varying speaking rate. In Agwuele, A., Warren, W., & Park, S. H. (Eds.), Proceedings of the 2003 Texas Linguistics Society Conference: Coarticulation in speech production and perception (pp. 86–95). Somerville, MA: Cascadilla Proceedings Project. Hassan, Z.M. (2003). Temporal compensation between vowel and consonant in Swedish and Arabic in sequences of CV:C & CVC: and the word overall duration. PHONUM, 9, 45–48. Hayes, B., & Steriade, D. (2004). Introduction: The phonetic bases of phonological markedness. In Hayes, B., Kirchner, R., & Steriade, D. (Eds.), Phonetically-based phonology (pp. 1-33). Cambridge, UK: Cambridge University Press. Henriksen, N. (2015). Acoustic analysis of the rhotic contrast in Chicagoland Spanish: An intergenerational study. Linguistic Approaches to Bilingualism, 5(3), 285-321.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 24 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Homma, Y. (1981). Durational relations between Japanese stops and vowels. Journal of Phonetics, 9, 273–281. Hrycyna, M., Lapinskaya, N., Kochetov, A., & Nagy, N. (2011). VOT drift in 3 generations of heritage language speakers in Toronto. Canadian Acoustics, 39, 166-167. Jaeger, J. (1978). Speech aerodynamics and phonological universals. Proceedings of the 4th Annual Meeting of the Berkeley Linguistics (pp. 312-329). Retrieved from https://journals.linguisticsociety.org/proceedings/index.php/BLS/article/view/2221/1991 Khattab, G. (2000). VOT production in English and Arabic bilingual and monolingual children. Leeds Working Papers in Linguistics, 8, 95–122. Khattab, G., & Al-Tamimi, J. (2014). Geminate timing in Lebanese Arabic: The relationship between phonetic timing and phonological structure. Laboratory Phonology, 5(2), 231-269. Kawahara, S. (2007). Sonorancy and geminacy. In Bateman, L., Werle, A., O’Keefe, M., & Reilly, E. (Eds.), University of Massachusetts occasional papers in linguistics 32: Papers in Optimality Theory III (pp. 145–186). Amherst, MA: GLSA. Kawahara, S. (2013). Emphatic gemination in Japanese mimetic words: A wug-test with auditory stimuli. Language Sciences, 40, 24-35. doi:10.1016/j.langsci.2013.02.002 Kirchner, R. (1998). An effort-based approach to consonant lenition (Unpublished doctoral dissertation). UCLA, Los Angeles. Retrieved from Rutgers Optimality Archive ROA2, http://ruccs.rutgers.eduJroa.html Kaye, A.S. (2005). Gemination in English. English Today, 21(2), 43-55. Kondaurova, M.V., & Francis, A.L. (2008). The relationship between native allophonic experience with vowel duration and perception of the English tense/lax vowel contrast by Spanish and Russian listeners. Journal of the Acoustical Society of America, 124(6), 3959–3971. Kondaurova, M.V., & Francis, A.L. (2010). The role of selective attention in the acquisition of English tense and lax vowels by native Spanish listeners: Comparison of three training methods. Journal of Phonetics, 38(4), 569–587. Kraehenmann, A., & Lahiri, A. (2008). Duration differences in the articulation and acoustics of Swiss German word-initial geminate and singleton stops. Journal of the Acoustical Society of America, 123, 4446–4455. Kreidler, C.W. (2004). The pronunciation of English. Oxford, UK: Blackwell. Lahiri, A., & Hankamer, J. (1988). The timing of geminate consonants. Journal of Phonetics, 16, 327–338. Lavoie, L.M. (2015). Consonant strength: Phonological patterns and phonetic manifestations. London, UK: Routledge. Maddieson, I. (1984). Patterns of sounds. Cambridge, UK: Cambridge University Press. Major, R.C. (1992). Losing English as a first language. The Modern Language Journal, 76(2), 190-208. Markus, E., Lippus, P., Pajusalu, K., & Teras, P. (2013). Three-way opposition of consonant quantity in Finnic and Saamic languages. In Asu-Garcia, E. L., & Lippus, P. (Eds.), Nordic Prosody: Proceedings of the XIth Conference, Tartu (pp. 225–234). Frankfurt, Germany: Peter Lang. Mayr, R., Price, S., & Mennen, I. (2012). First language attrition in the speech of Dutch–English bilinguals: The case of monozygotic twin sisters. Bilingualism: Language and Cognition, 15, 687–700. doi: https://doi.org/10.1017/S136672891100071X

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 25 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Mazzaro, N., Cuza, A., & Colantoni, L. (2016). Age effects and the discrimination of consonantal and vocalic contrasts in heritage and native Spanish. In Tortora, C., den Dikken, M., Montoya, I., & O’Neill, T. (Eds.), Romance linguistics 2013: Selected proceedings of the 43th Linguistic Symposium on Romance Languages (pp. 277-300). Amsterdam, The Netherlands: John Benjamins. Mitchell, T. (1990). Pronouncing Arabic 1. Oxford, UK: Oxford University Press. Morrison, G. S. (2008). L1-Spanish speakers’ acquisition of the English /i/-/ɪ/ contrast: Duration- based perception is not the initial developmental stage. Language and Speech, 51(4), 285–315. Oh, G.E., & Redford, M.A. (2012). The production and phonetic representation of fake geminates in English. Journal of Phonetics, 40(1), 82-91. doi: 10.1016/j.wocn.2011.08.003 Ohala, J. (1983). The origin of sound patterns in vocal tract constraints. In MacNeilage, P. (Ed.), The production of speech (pp. 189–216). New York, NY: Springer Verlag. Ohala, M. (2007). Experimental methods in the study of Hindi geminate consonants. In Solé, M. J., Beddor, P., & Ohala, M. Experimental approaches to phonology (pp. 351–368). Oxford, UK: Oxford University Press. Payne, E. (2005). Phonetic variation in Italian consonant gemination. Journal of the International Phonetic Association, 35(2), 153-181. Prescher, P. (2007). Identity, immigration and first language attrition. In Köpke, B., Schmid, M. S., Keijzer, M., & Dostert, S. (Eds.), First language attrition: Theoretical perspectives (pp. 189- 205). Amsterdam, The Netherlands: John Benjamins. Podesva, R. (2002). Segmental contrasts on geminates and their implications for typology. Presented at the LSA Annual Meeting, San Francisco, California. Rafat, Y. (2010). A socio-phonetic investigation of rhotics in Persian. Iranian Studies, 43, 667- 682. Rafat, Y. (2011). Orthography-induced transfer in the production of novice adult English- speaking learners of Spanish (Unpublished doctoral dissertation). University of Toronto, Canada. Rafat, Y., Mohaghegh, M., & Stevenson, R. (2017). Geminate attrition across three generations of Farsi-English bilinguals living in Canada: An acoustic study. Ilha do Desterro Investigating Second Language Speech Learning, 71(3), 151-168. Retrieved from https://periodicos.ufsc.br/index.php/desterro/artic le/view/2175-8026.2017v70n3p151 Rao, R. (2016). On the nuclear intonational phonology of heritage speakers of Spanish. In Pascual y Cabo, D. (Ed.), Advances in Spanish as a heritage language (pp. 51-80). Amsterdam, The Netherlands: John Benjamins. Ruíz-Peña, E., Sevilla, D., & Rafat, Y. (2015). Second dialect imitation: The production of Ecuadorian Spanish assibilated rhotics by Andalusian speakers of Spanish. In Babatsouli, E., & Ingram, D. (Eds.), Proceedings of the International Symposium on Monolingual and Bilingual Speech 2015 (pp. 288-300). Retrieved from http://ismbs.eu/publications Saadah, E. (2011). The production of Arabic vowels by English L2 learners and heritage speakers of Arabic (Unpublished doctoral dissertation). University of Illinois at Urbana-Champaign. Schmid, M. S., & Köpke, B. (2004) Language attrition: The next phase. In Schmid, M. S., B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First language attrition: Interdisciplinary perspectives on methodological issues (pp. 1-43). Amsterdam, The Netherlands: John Benjamins.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 26 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Schmid, M.S., & Köpke, B. (2007). Bilingualism and attrition. In Köpke, B., Schmid, M. S., Keijzer, M., & Dostert, S. (Eds.), Language attrition (pp. 1–7). Amsterdam, The Netherlands: John Benjamins. Seliger, H. W., & Vago, R. M. (Eds.). (1991). First language attrition. Cambridge, UK: Cambridge University Press. Sorianello, P. (2014). Italian geminate consonants in L2 acquisition. In Costamagna, L., & C. Celata, C. (eds.), Consonant gemination in first and second language acquisition (pp. 25-46). Pisa, Italy: Pacini editore. Steriade, D. (1982). Greek prosodies and the nature of syllabification (Unpublished doctoral dissertation). MIT, Cambridge, MA, USA. Trask, R.L. (1996). A dictionary of phonetics and phonology. London, UK: Routledge. Taylor, M. (1985). Some patterns of geminate consonants. The University of Chicago Working Papers in Linguistics, 1, 120–129. Varona, D., Ruíz-Peña, E., Sierra, E.D., Stevenson, R., & Rafat, Y. (2017). Second dialect and second language imitation of geminates by Colombian Spanish speakers. Proceedings of the International Symposium on Monolingual and Bilingual Speech 2015 (pp. 99-106). Retrieved from http://www.ismbs.eu/publications-2018/

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 27 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

APPENDIX A

Generation Participant Age Gender Country of birth Dialect of origin AOA LOR #1 53 Female Syria Syrian 21 32 years #2 56 Female Syria Syrian 48 8 years Late bilinguals #3 60 Female Syria Syrian 38 22 years #4 54 Male Syria Syrian 18 36 years #5 46 Female Syria Syrian 34 12 years #1 24 Female Saudi Arabia Syrian 3 years 21 years #2 45 Female Canada Syrian N/A N/A Heritage #3 28 Female Canada Syrian N/A N/A speakers #4 24 Female Canada Syrian N/A N/A #5 30 Female Saudi Arabia Syrian 1 year 29 years #1 39 Female Syria Syrian N/A N/A #2 47 Female Syria Syrian N/A N/A Homeland #3 60 Male Syria Syrian N/A N/A #4 51 Male Syria Syrian N/A N/A #5 48 Female Syria Syrian N/A N/A AOA= age of arrival LOR= length of residency

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 28 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

APPENDIX B

Language Background Questionnaire Part 1. Personal Information 1. Participant Number: ______2. Gender:  Male.  Female. Part 2. General Language Questions 3. Please specify the known languages of: Your Father:  English  Arabic (which dialect) ……………….  Other (please specify): ……………… Your Mother:  English  Arabic (which dialect) ………………  Other (please specify): ……………… 4. What was your main language of instruction?

English Only Arabic only Other Primary/Elementary School Middle School High School University Other:______

5. How much Arabic instructions did you receive in school at the levels listed below? Less than one year 1-2 years More than 2 years Primary/Elementary School Middle School High School University Other:______

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 29 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

6. Please list all the languages you know IN ORDER OF DOMINANCE. 1- ………………….. Most dominant 2- …………………. 3- …………………. Least dominant

7. Please list all the languages you know IN ORDER OF ACQUISITION. 1- …………………… First acquired 2- ………………….. 3- …………………... Last acquired

8. Please specify the age when you... • …began acquiring English. ______• …became fluent in English. ______• …began reading in English. ______• …became fluent in reading English. ______

9. Please specify the age when you… • …began acquiring Arabic. ______• …became fluent in Arabic. ______• …began reading in Arabic. ______• …became fluent in reading Arabic. ______

10. Please list what percentage of time you are CURRENTLY and ON AVERAGE exposed to each language (percentages should add up to 100). English: ……./%100 Arabic: ……. /%100 Other: ……. %100

11. When choosing to read a text available in all your languages, in what percentage of cases would you choose to read it in each of your languages? Assume that the original was written in another language which is unknown to you (percentages should add up to 100). English: ……./%100 Arabic: ……. /%100 Other: ……. /%100

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 30 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

12. When choosing a language to speak with a person who is equally fluent in all your languages, what percentage of time would you choose to speak each language (percentages should add up to 100)? English: ……./%100 Arabic: ……. /%100 Other: ……. /%100

13. Please rate the amount of each language that you use at HOME. 1 2 3 4 5 6 7 None Strong majority English

Arabic

Other

14. Please rate the amount of each language that you use at WORK. 1 2 3 4 5 6 7 None Strong majority English

Arabic

Other

15. Please rate the amount of each language that you use in SOCIAL SITUATIONS. 1 2 3 4 5 6 7 None Strong majority English

Arabic:

Other:

16. Please rate your linguistic ability in ENGLISH according to the following scale.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 31 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

1 2 3 4 5 6 7 Beginner Intermediate Advanced Native Speaking Ability Reading Ability Writing ability Translating Ability Listening comprehension Pronunciation Fluency Vocabulary Grammatical Ability Overall competence

17. Please rate your linguistic ability in Arabic according to the following scale. 1 2 3 4 5 6 7 Beginner Intermediate Advanced Native Speaking Ability Reading Ability Writing ability Translating ability Listening comprehension Pronunciation Fluency Vocabulary Grammatical ability Overall competence

18. Please rate how much the following factors contributed to you learning. 1 2 3 4 5 6 7

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 32 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Beginner Intermediate Advanced Native Speaking ability Reading ability Writing ability Translating ability Listening comprehension Pronunciation Fluency Vocabulary Grammatical ability Overall competence

19. Please rate how much the following factors contributed to you learning English.

1 2 3 4 5 6 7 None A lot Interacting with friends Interacting with family Work

Reading Language tapes/self instruction Watching TV / listening to the radio Formal courses

20. Please rate how much the following factors contributed to you learning Arabic. 1 2 3 4 5 6 7 None A lot

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 33 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Interacting with friends Interacting with family Work

Reading Language tapes/self instruction Watching TV / listening to the radio Formal courses

Part 3. Health & Medication Questions 21. Have you ever had: YES NO A vision problem? A hearing impairment? A language disability? A learning disability? A stroke? A neurological disorder? A psychiatric disorder? Attention problems? Sleeping problems?

If you responded yes to any of the preceding questions, please specify:…………………………………………………………………………………………… …………………

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 34 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

APPENDIX C

Tokens with Stop Consonants Singletons Geminates Stops b - ب /t̠ab:aq/ ١- طبّق /t̠abaq/ طبق /s̠ ab:aːr/ ٢- صبّار /s̠ abaħ/ صباح /kab:ar/ ٣- كبّر /s̠ abar/ صبر /jab:ar/ ٤- جبر /jabal/ جبر /ʔab:ar ٥- عبّر /ʕabar/ عبر t - ت /ħat:a/ ١- حتى /mata/ متى /ʃat:a/ ٢- شتى /ʃatam/ شتم /fat:an/ ٣- فتّن /fatan/ فتن /sat:ar/ ٤- ستّر /satar/ ستر /χat:am/ ٥- ختّم χatam/ ختم d - د /ʕad:al/ ١- عدّل /ʕadas/ عدس /s̠ ad:aq/ ٢- صدّق /s̠ adaq/ صدق /qad:am/ ٣- قدّم /qadam/ قدم /s̠ ad:ar/ ٤- صدّر /s̠ adam/ صدر /had:am/ ٥- بدّل /badan/ بدل t̠ - ط /sat̠:aħ/ ١- س ّطح /nat̠aħ/ نطح /ħat̠:am/ ٢- ح ّطم /ħat̠ab/ حطب /sat̠:ar/ ٣- س ّطر /sat̠r/ سطر /mat̠:aːt̠/ ٤- مطاط /mat̠aːr/ مطار /qt̠:ar/ ٥- ق ّطر /qat̠ar/ قطر q - ق /naq:aːl/ ١- ن ّقال /maqaːl/ مقال /waq:aʕ/ ٢- و ّقع /waqaʕ/ وقع /naq:ab/ ٣- ن ّقب /naqal/ نَ َقل /ʕaq:ad/ ٤- ع ّقد /ʕaqad/ عقد /raq:am/ ٥- ر ّقم /raqam/ رقم

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 35 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

k - ك /fak:ar/ ١- ف ّكر /makar/ مكر /sak:ar/ ٢- س ّكر /sakan/ سكن /ðak:ar/ ٣- ذ ّكر /ðakr/ ذكر /rak:ab/ ٤- ر ّكب /rakab/ ركب /mak:aːr/ ٥- م ّكار /makaːn/ مكان

Tokens with Fricative Consonants Singletons Geminates Fricatives

ð - ذ /kað:b/ ١- كذّب /kaðab/ كذب /ʔð:an/ ٢-أذّن /ʔðan/ أذن /bað:ar/ ٣- بذّر /baðal/ بذل /ʔað:ab/ ٤- عذّب /ʕaðb عذب /hað:ab/ ٥-هذّب /haðaː/ هذى ʕ - ع /s̠ aʕ:ab/ ١- صعّب /s̠ aʕb/ صعب /saʕ:ar/ ٢- سعّر /saʕal/ سعل /ʃaʕ:ab/ ٣- شعّب /ʃaʕar/ شعر /naʕ:am ٤- نعّم /naʕam/ نعم /baʕ:ad/ ٥- بعّد /baʕd/ بعد

f - ف /t̠af:a/ ١- ط ّفى /ɣafa/ غفى /χaf:af/ ٢- خ ّفف /χafaq/ خفق /naf:ar/ ٣- ص ّفر /nafar/ سفر /s̠ af:ar/ ٤- ن ّفر /safar/ نفر /s̠ af:a/ ٥- ص ّفى /safa/ صفى s - س /kas:ar/ ١- ك ّسر /kasar/ كسر /mas:aħ/ ٢- م ّسح /masaħ/ مسح /mas:aːħah/ ٣- م ّساحه /masaːħah/ مساحة /nas:aq/ ٤- ن ّسق /nasab/ نسب /qas:am/ ٥- ق ّسم /qasam/ قسم

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 36 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

Tokens with Sonorant Consonants Singletons geminates Sonorants

r - ر /ħar:an/ ١- حران /ħaram/ حرام /jar:ab/ ٢- ج ّرب /jarab/ جرب /ʃar:aħ/ ٣- ش ّرح /ʃaraħ/ شرح /dar:as/ ٤- د ّرس /daras/ درس /dar:ab/ ٥- د ّرب /d̠ arab/ ضرب l - ل /ʕal:am/ ١- ع ّلم /ʕalam/ علم /ħal:aq/ ٢- ح ّلق /ħalaq/ حلق /sal:am/ ٣- س ّلم /salab/ سلب /fal:aːħ/ ٤- ف ّالح /falaːħ/ فالح /sal:a/ ٥- ص ّلي /ʕala/ صلي m - م /rum:aːn/ ١- ر ّمان /ʕumaːn/ عمان /ħam:aːm/ ٢- ح ّمام /ħamam/ حمام /ʃam:aːm/ ٣- ش ّمام /ʃamaːl/ شمال /jam:aʕ/ ٤- ج ّمع /jamal/ جمع /zam:aːr/ ٥- ز ّمار /zamaːn/ زمان n - ن /fan:aːn/ ١- ف ّنان /ħanaːn/ حنان /qan:aːs/ ٢- ق ّناص /qanaːh/ قناة /ban:aːʔ/ ٣- ب ّناء /banaːt/ بنات /aln:aːs/ ٤- ال ّناس /ʔnaːnaːs/ اناناس /man:aːn/ ٥- م ّنان /manaːl/ منال y - ي /χay:aːl/ ١- ّخيال /χayaːl/ خيال /bay:aːʕ/ ٢- بيّاع /bayaːn/ بيان /t̠ay:aːr/ ٣- طيار /χayaːr/ خيار /ħay:aːt/ ٤- حيّات /ħayat/ حياه

Downloaded from Brill.com10/06/2021 03:36:42PM via free access 37 Heritage Language Journal, 17(1) https://doi.org/10.46538/hlj.17.1.1 April, 2020

NOTES

1. We are using attrition as a synonym for “change across generations”; that is, we do not mean to say that bilingual populations speak worse than monolingual speakers, but rather that they speak differently.

Downloaded from Brill.com10/06/2021 03:36:42PM via free access