1. Blackfoot English
Several researchers, especially in education and speech-language pathology, have investigated a First Nations variety of English distinguishable from mainstream or Standard Canadian English (Ball and Bernhardt Reference Ball and Bernhardt2008; Peltier Reference Peltier2009, Reference Peltier2010, Reference Peltier2011; Fadden and LaFrance Reference Fadden and LaFrance2010; Genee and Stigter Reference Genee and Stigter2010; Thorburn Reference Thorburn2014).Footnote 1 However, as the linguistic details of the variation have not been much discussed in Canada, we are mostly limited to related research conducted in New Zealand (Maclagan et al. Reference Maclagan, King and Jones2003, Reference Maclagan, King and Gillon2008) or the United States (e.g., Wolfram Reference Wolfram1984, Reference Wolfram1996; Leap Reference Leap1993; Coggshall Reference Coggshall2008; Meek Reference Meek2006), much of which is not directly applicable to the Canadian setting. Here we attempt to take a small step forward in the study of First Nations English dialects in Canada, examining the prosodic rhythm of speakers of Blackfoot descent as compared to that of speakers of settler descent in Southern Alberta. Rhythm is the locus of investigation because prosody is one salient aspect used to characterize First Nations English (Leap Reference Leap1993: 50, Ball and Bernhardt Reference Ball and Bernhardt2008: 577), and because similar work has been done in both the U.S. (Coggshall Reference Coggshall2008) and New Zealand (Szakay Reference Szakay, Warren and Watson2006), offering us a point of comparison and a reason to believe that rhythmic differences can carry ethnolinguistic markers.
Many Blackfoot individuals speak a variety of English noticeably different from local mainstream settler-heritage Canadian English (Genee and Stigter Reference Genee and Stigter2010). Local people in the Southern Alberta area, both First Nations and non-First Nations, readily accept that such an ethnolect exists, and First Nations individuals who ‘live in both worlds’ often display and report conscious code-switching behaviours. While its origin may be due to language transfer effects acquired by Blackfoot mother tongue speakers in the colonial education context, it is our contention that it is an intergenerationally transmitted variety of English that is also spoken in families where fluency in Blackfoot is no longer prominent. That is, even past the first generation of English speakers, features of this ethnolect persist, contrary to findings in Hoffman and Walker (Reference Hoffman and Walker2010) for other heritage language speakers. This ethnolect shares features both with other Indigenous Englishes and with other non-standard varieties of English (Penfield Reference Penfield and Leap1977; Stout and Erting Reference Stout, Erting and Leap1977; Leap Reference Leap and Leap1977a, Reference Leap and Leap1977b, Reference Leap, Bartelt, Jasper and Hoffer1982, Reference Leap1993; Mulder Reference Mulder, Bartelt, Jasper and Hoffer1982; Wolfram Reference Wolfram1984, Reference Wolfram1996; Toohey Reference Toohey1985, Reference Toohey1986; Wolfram and Dannenberg Reference Wolfram and Dannenberg1999; Wolfram and Sellers Reference Wolfram and Sellers1999; Ball et al. Reference Ball, Bernhardt and Deby2005; Ball and Bernhardt Reference Ball and Bernhardt2008; Peltier Reference Peltier2009, Reference Peltier2010).
2. Setting the stage: the Blackfoot Nation in Southern Alberta
Southern Alberta is the traditional territory of several First Nations, the most important of which are the Blackfoot. They now live in four reserve communities in Southern Alberta and Montana, but the original Blackfoot territory was much larger and extended north into central and northern Alberta and Saskatchewan, south into Montana and Idaho, west into the Rocky Mountains and east as far the Cypress Hills and into Saskatchewan (A. Hungry Wolf and B. Hungry Wolf Reference Hungry Wolf and Wolf1989, Dwyer and Stout Reference Dwyer and Stout2012: 10–11). First contact with European settlers took place in the late 18th century, but in contrast to other Plains tribes such as the Cree, the Blackfoot did not generally establish direct trade relations with the Hudson's Bay Company, North West Company or other colonial trading outfits; instead they preferred to remain independent and rely on the buffalo hunt for sustenance, rather than trapping for trade furs (Dickason and Newbigging Reference Dickason and Newbigging2010: 118–126), although they did trade in buffalo hides (A. Hungry Wolf and B. Hungry Wolf Reference Hungry Wolf and Wolf1989: 7–14). In the mid-19th century, the area was being colonized by an increasing stream of European and American settlers who brought with them diseases such as smallpox, tuberculosis, and scarlet fever, which severely reduced the population, in particular in the 1870s when several major epidemics broke out. The whiskey trade further weakened the population (Kozak Reference Kozak1971: 5), and around 1880 the bison were extinguished, resulting in the loss of the main source of independence for the Blackfoot and forcing them onto reserves. Treaty 7 was signed in 1877 in Canada, between the Crown and the Siksika (Blackfoot), Piikani (Peigan), Kainai (Blood), Tsuu T'ina (Sarcee), and Stoney (Nakoda) nations (Dickason and Newbigging Reference Dickason and Newbigging2010: 196). As a result of this treaty, separate reserves were established for the nations that were signatories (see map in Figure 1): the Siksika (Blackfoot) reserve east of Calgary, the Kainai (Blood) reserve south-west of Lethbridge, and the Piikani (Peigan) reserve west of Fort Macleod. The linguistically unrelated Tsuu T'ina and Stoney nations also received reserve lands. A tribal territory for the Montana Blackfeet had been established earlier by the Lame Bull Treaty in 1855 (A. Hungry Wolf and B. Hungry Wolf Reference Hungry Wolf and Wolf1989: 11–13) and mission schools had been established there since the 1840s (Still Smoking Reference Still Smoking1997: 11–12).
Blackfoot reserves: A. = Siksika, B. = Piikani (Peigan), C. = Kainai (Blood), D. = Aamsskaapipikani (Piegan Blackfeet, Montana). Map courtesy of Kevin McManigal, University of Montana.
Schools were established on all Blackfoot reserves starting in the 1880s; these were mission schools run by Church of England and Roman Catholic missionaries (Kozak Reference Kozak1971, Miller Reference Miller1996, Anglican Church of Canada 2008.) Initially the schools were day schools, but as time went on, these failed and were replaced with boarding schools, around the turn of the twentieth century. For many Blackfoot students this was where they first encountered the English language and began to learn it. Most of them simultaneously lost fluency in Blackfoot, sometimes through being punished for speaking it, a widespread practice across the country, especially after 1900 (Miller Reference Miller1996: 199–205, 415–417; Truth and Reconciliation Commission 2015: 88–92). While ostensibly intended to provide Blackfoot students with skills and training that could help them integrate into the settler society (Kozak Reference Kozak1971), racially based boarding schools and the reserve system (which included a pass system and required permission for on-reserve individuals to leave the reserve; see Williams Reference Williams2015) in fact served to keep the First Nations population separate from the white settler population and impeded integration.
According to Aboriginal Affairs and Northern Development Canada (AANDC), 23,729 individuals were registered members of one of three Blackfoot First Nations in Alberta in November 2017. In Table 1 we see that 14,887 of these, or 62.7%, indicated that they live on reserve, while 8,842 individuals, or 37.3%, live off reserve.
Registered population on Kainai (Blood), Piikani (Peigan), and Siksika Nations, November 2017. (Aboriginal Affairs and Northern Development 2017).
These figures need to be interpreted carefully, however, since many First Nations individuals move between on-reserve and off-reserve living arrangements depending on their life circumstances.
Aspects of the racial segregation inherent in the early reserve period and perpetuated by the residential school system persist in Southern Alberta to this day. First Nations people can be said to be the original “visible minorities” in the area, and are routinely treated as “other” by the mainstream settler population in a variety of contexts (Fiske et al. Reference Fiske, Belanger and Gregory2010, Granzow Reference Granzow2010, Belanger et al. Reference Belanger, Awosoga and Head2013). Thinly veiled racism is fairly acceptable in “polite society” in various forms, as the authors can attest from personal experience. Many First Nations individuals, even those who live their entire life in urban areas such as Lethbridge, lead lives that are quite separate from mainstream society. We believe that this isolation, coupled with language transfer effects that arose with the transition from Blackfoot to English, has contributed to the dialectal differences we will show here.
3. Prosodic rhythm
In the study of prosodic rhythm, there has been a tradition of language classification as either stress-timed or syllable-timed, terms coined by Pike (Reference Pike1945), and further expounded upon by Abercrombie (Reference Abercrombie1967). Languages with syllable timing are said to have relatively stable syllable duration, while in contrast, in stress-timed languages, all metric feet have the same approximate length. In languages with stress timing, the duration of syllables is variable, but the time between successive stressed syllables is fairly consistent. Ramus et al. (Reference Ramus, Nespor and Mehler1999) suggest that vowel reduction is a prominent aspect of stress-timed languages, such as in Germanic languages like English and German. On the other hand, Romance languages such as French, Spanish, and Portuguese are characterized as syllable-timed. Note, however, that although this rhythmic classification of stress-timed and syllable-timed languages is convenient, Dauer (Reference Dauer1983) argued that a gradient view of relative rhythmic patterns is more accurate. In this view, a language's rhythm lies on a continuum from very stress-timed at one end, to very syllable-timed at the other. Since this work, a number of methods have been developed to mathematically compute an index situating a language along this continuum. The present research does not take a theoretical stand on which of these methods is superior, but rather uses measurement as a comparative descriptive tool for the purpose of quantifying potential rhythmic differences between varieties.
We have used a pairwise measure for measuring rhythm, following Deterding (Reference Deterding2001). Deterding's (Reference Deterding2001) pairwise Variability Index (VI) measures the relative duration of neighbouring syllables, normalizing by average syllable duration in the utterance. Following Deterding (Reference Deterding2001), we also omit the final syllable of each phrase in order to control for phrase-final lengthening effects (see also Deterding Reference Deterding2011, Reference Deterding and Romero-Trillio2012). Deterding's VI has been argued to be less susceptible to uncertainty in the measurement of short syllables, as compared with other pairwise measures (Deterding Reference Deterding2001, Reference Deterding2011, Reference Deterding and Romero-Trillio2012). Similar to other pairwise measures such as Low et al. (Reference Low, Grabe and Nolan2000)’s PVI, a lower measurement on Deterding's VI will also reflect a language that is relatively more syllable-timed (such as Italian, Maori, or Malay), whereas a higher VI measure will reflect a language that is relatively more stress-timed (such as English or German).
4. The study
Impressionistic and anecdotal reports of prosodic differences between the English spoken by speakers of First Nations descent versus those of settler descent have motivated this study. Given our geographic location, we chose to investigate the English spoken by persons of Blackfoot descent (BE), comparing it with the English spoken by persons of settler descent (SE) who reside in the same geographical area.
4.1 Data
Drawing on data from the Southern Alberta Corpus of English (SACE, Rosen and Skriver Reference Rosen and Skriver2015), the sociolinguistic interviews and reading passages of ten speakers were analyzed. To offset the lower power of the small size of the sample, we controlled for other demographic variables such as socio-economic status, sex, age, and rurality. All speakers were therefore women from rural or semi-rural areas in Southern Alberta (i.e., outside Calgary), aged 25–50, and are non-professionals. Crucial to the study, both groups consider English to be their first language. The only demographic difference between speakers is that five are of settler descent and five are of Blackfoot descent. Participants were recruited primarily via social networks and interviewed in their homes or on campus by local student interviewers. The Blackfoot English interviews were conducted by a single interviewer who was also of Blackfoot descent, to obviate any linguistic accommodation to an outsider variety of English. The settler-heritage interviews were likewise conducted by interviewers of settler descent. Interviews lasted approximately an hour and included three parts: a sociolinguistic interview, a word list, and a reading passage. Only the sociolinguistic interview and the reading passage are analyzed here. Twenty phrases containing minimally eleven syllables each were extracted from each of the sociolinguistic interviews. In addition, five phrases from the reading passage “The North Wind and the Sun” were analyzed for each speaker. The result was 3911 syllables measured in the free speech data, and 861 syllables measured from the reading passages, for a total of 4772 syllables measured and analyzed in the study.
4.2 Type of speech and syllable length
The sociolinguistic interviews attempted to elicit natural conversation-style speech from participants and were approximately one hour in length. The reading passage was given to each participant, who was instructed to read it at a normal pace, as though they were reading it to a child. Sociolinguistic interviews were transcribed in ELAN (Sloetjes and Wittenburg Reference Sloetjes and Wittenburg2008) and segments were extracted into Praat (Boersma Reference Boersma2001) for analysis. Five of the seven phrases within the reading passage were extracted and their rhythm was analyzed. This allowed for a direct comparison of VI values between speakers. The sociolinguistic interview and reading passage values were used as both covariate and outcome variables. Speech type could then be compared within speakers (i.e., rhythm of free speech vs. read speech), and between speakers (i.e., rhythm of free or read speech in Blackfoot English vs. settler-heritage English).
4.3 Variability Index
In calculating Deterding's (Reference Deterding2001) Variability Index (VI), stress-timed languages display higher VI values than do syllable-timed languages. The Variability Index was calculated for each phrase (a total of 25 times per speaker) according to the following formula:
$${\rm VI\ =} \displaystyle{1 \over {n{\rm -}2}}\sum\limits_{k = 1}^{n{\rm -}2} {\vert {d_{k + 1}{\rm -}d_k} \vert .}$$This formula represents perfect syllable isochrony (i.e., no durational variability) with a value of zero, and the greater the syllable duration variability, the further from zero the VI value will be (Guilbault Reference Guilbault2002). The steps taken to compute the VI for each speaker are presented in chronological order.
First, the measured duration for each syllable was calculated by subtracting the start time of each syllable from the end time of each syllable. Second, the average duration for all the syllables in the phrase was calculated by adding up all the measured durations and dividing by the total number of syllables. Third, in order to analyze similar units across all speakers, we calculate the normalized duration by dividing the measured duration by the average duration. Then the VI was calculated by taking the absolute value of the normalized duration of a syllable subtracted from the normalized duration of the previous syllable. Each VI value was then entered into a spreadsheet for analysis.
4.4 Variables
The data was subjected to statistical analysis in SPSS, with the VI as the dependent variable. Demographic variables were largely controlled for, other than ethnicity and age. Conforming to previous research, linguistic variables tested included the type of speech (free or reading passage) and number of syllables in the segment.
4.5 Statistical results
After examining the descriptive data (see Table 2 for means for all variables in the study), and examining the data for normality, we utilized analysis of variance (ANOVA) to answer our research questions. For each ANOVA, the dependent variable was the Variability Index for each of the 250 phrases spoken by the 10 participants. We concentrated on two different questions, the first methodological, focusing on differences in variability as a function of Type of Speech in order to check how type of speech affects VI; our second question, and the one motivating this research, focused on whether speech was differentially variable for individuals of Blackfoot descent, as well as whether BE differences in variability were maintained after several demographic and other speech variables were controlled for. The results in Table 3 are discussed separately for each research question in sections 4.5.1 and 4.5.2.
Variables studied and overall VI.
Before moving on to discussion of the statistical testing of our research questions, however, we give the distributional differences between the two heritage groups.
At first glance, we observe differences between the mean Variability Index value for all of the BE speakers in this study, in comparison to the mean Variability Index value for all SE speakers in this studyFootnote 4.
Mean Variability Indexes for Blackfoot and settler-heritage speakers by data type.
As we see in Table 3, the BE speakers demonstrate somewhat higher Variability Indexes along the continuum, suggesting a more stress-timed rhythm pattern than that of SE speakers. Note however that this is driven primarily by the interview data. The SE speakers have a larger difference between interview (.469) and reading passage (.621) speech than the BE speakers, who do show a difference (.503 for interview as compared to .597 for reading), but one not as wide as the SE speakers. The interview data (.503 for BE and .469 for SE) can be compared with Deterding's values for Singaporean English at .448 and British English at .54. Here there are two interesting things to note: first, SE is closer to Singaporean than to British English, and second, there is a narrower gap between BE and SE than Deterding found for Singaporean and British English. We argue that this smaller gap is unsurprising given that the two groups Deterding studied were not contiguous speech communities, unlike the case here. In the present study, the two groups live in the same geographic region, and we can expect more interaction between the Southern Albertan groups than those in Singapore as compared with Britain, which in turn would predict closer VIs. It is interesting to note as well that most studies examining different ethnolects of English speakers compare one group with a possibly syllable-timed substrate influence (Deterding Reference Deterding2001 for Singapore English, Coggshall Reference Coggshall2008 for Cherokee English, Shoustermann Reference Shousterman2014 for Puerto Rican English), and in all cases, the substrate language did influence the English in lowering the variability index, bringing it closer to the syllable-timed end of the continuum, though likely still within the range we would consider as stress-timedFootnote 5. In this case, however, the Blackfoot English speakers have a higher VI than the Settler-Heritage speakers, who seem to be closer to syllable-timed rhythmic patternsFootnote 6. In order to see whether this observable difference is significant, we conducted statistical tests, which we outline next.
4.5.1 Variability as a function of speech type
Our first question was whether speech type influenced the VI. Research on prosodic rhythm often measures reading passages for ease of comparison across speakers and across groups (see Low, Grabe and Nolan Reference Low, Grabe and Nolan2000, Grabe and Low Reference Grabe, Low, Gussenhoven and Warner2002), but it is also well known that free speech can pattern differently from read speech (Labov Reference Labov1994). In order to discover whether this is also the case for our data, we used univariate ANOVA to explore group differences in variability of speech type as a function of Type of Speech (free speech versus reading a passage). Results indicated that the VI was significantly higher for speech that was created by reading a passage (M = .62; SD = .08) versus those created during a free speech interview (M = .49; SD = .14); F(1, 248) = 42.62, p < .001. This means that reading passages have higher variability and are more stereotypically stress-timed than free speech for the Southern Alberta speakers as a whole.
4.5.2 Variability as a function of ethnicity
Our overarching research question was whether we could find measurable differences in the prosodic rhythm between the BE and SE speakers, supporting the anecdotal evidence of prosodic differences. In order to answer this question, an ANOVA was run to explore whether the Variability Index differed. Results indicated that individuals who identified as Blackfoot had significantly higher VIs (M = .53; SD = .14) than SE participants (M = .50; SD = .14); F(1, 248) = 4.69, p < .05. We then included several covariates into the ANOVA model to determine whether this finding remained significant after controlling for other relevant variables. The covariates that were added included Speech Type (free speech vs. reading passage), Number of Syllables, and Age of Participant.Footnote 7 As can be seen from the ANOVA results in Table 3, age was not significant, but the three other variables were significant, including whether an individual was of Blackfoot descent (see Figure 2 for estimated means). A third ANOVA was run to explore potential interaction effects of Indigenous status and speech type, but they were not significant, so are not reported here.
Estimated Marginal Means for Variability as a function of Indigenous Status after controlling for speech type, syllables, and age.
5. Discussion
Our results reveal a quantifiable difference between the rhythmic patterns of Blackfoot English and those of settler-heritage English. Blackfoot English speakers have a statistically significant higher average Variability Index, indicating an ethnolectal variety farther along the stress-timed end of the continuum, though both BE and SE values indicate stress-timed treatment of the language. The implications of this research strongly suggest a gradient perspective on stress-timing and syllable-timing.
Given these results, the question arises as to why Blackfoot-descended speakers of English would have a higher VI than speakers of settler extraction. One expected source for this difference might be language transfer effects, which would have occurred during acquisition of an L2 English by Blackfoot native speakers who underwent language shift in the early to mid-twentieth century. This sort of transfer of prosodic features has been shown to exist in Singaporean English (Low et al. Reference Low, Grabe and Nolan2000), New Zealand English (Szakay Reference Szakay, Warren and Watson2006), Spanish Harlem English (Shoustermann Reference Shousterman2014), and Cherokee English (Coggshall Reference Coggshall2008). In each case, a syllable-timed language in close contact with English was shown to result in an ethnolect with a lower PVI, making stress-timed English more syllable-timed. It seems reasonable to predict that Blackfoot would exercise a similar influence over English in our analogous case. Our predicament, however, is that there has been no previous research on Blackfoot as a stress-timed or syllable-timed language. Furthermore, remaining speakers of Blackfoot today are virtually all English-Blackfoot bilinguals, and so it is difficult to uncover the non-bilingual-influenced rhythm of Blackfoot. Rather than analyzing the monolingual Blackfoot speech signal, we turn to typological and phonological evidence. Recall that in section 3 we mentioned that one approach to the study of rhythm is phonological, where rhythm type results from the phonological structure of a given language. We wish now to use the insights this approach provides alongside the pairwise variability approach, to predict where on the rhythmic scale Blackfoot might lie.
According to Dauer (Reference Dauer1983, Reference Dauer1987), rhythmic diversity results from the combinations of phonological, phonetic, lexical, and syntactic facts associated with different languages, especially syllable structure, the presence or absence of vowel reduction, and word stress. Dasher and Bolinger (Reference Dasher and Bolinger1982) similarly suggest that the rhythm of a language results from the phonological structure of a given language, citing similar features associated with rhythmic types: syllable types, phonological vowel length distinctions, and vowel reduction. We will take up each of these phonological phenomena in turn, to assess where we would expect Blackfoot to fall on the rhythmic spectrum.
Analysis of Variance for Variability Index as a Function of Blackfoot, Speech Type, Number of Syllables, and Age.
5.1 Phonological criteria for prosodic rhythm
Dauer (Reference Dauer1987) sets out a list of criteria which would determine whether a language is likely to have ‘strong stress’, that is, be stress-timed. The relevant criteria for a stress-timed language include: a wide range of permissible syllable structures; no or few long vowels permitted in unaccented syllables; accented syllables correlated with pitch contours; vowels in unaccented syllables being reduced or centralized; consonant neutralization or reduced allophones in unstressed syllables. This means that a hypothetical language with a simple CV syllable structure would tend to be at the syllable-timed end of the rhythmic continuum. In pairwise variability terms, these languages with simple CV syllable structure would have low PVI values. In reviewing the existing literature on Blackfoot phonology, we find that many of the criteria Dauer lays out as predictive of a stress-timed language are reported to exist in Blackfoot, but that the evidence is inconclusive. Here we investigate where Blackfoot is situated along Dauer's phonological criteria for stress-timed languages.
a. wide range of permissible syllable structures
Though Blackfoot has been described as having ‘exceedingly complex’ syllable structure (Taylor Reference Taylor1969), more recent accounts analyze it as allowing only a restricted number of coda consonants once /s/ is excluded (Denzer-King Reference Denzer-King2009, Goad and Shimada Reference Goad, Shimada, Iyer and Kusmer2014, Weber Reference Weber, Macauley, Noodin and Valentine2016). The previous description of a complex Blackfoot syllable structure was due to the fact that /s/ in Blackfoot creates a number of unusual clusters, which are generally now analyzed as involving a moraic /s/ that can serve as the nucleus of a syllable (Derrick Reference Derrick, Fujimori and Reis Silva2006, Reference Derrick2007; Denzer-King Reference Denzer-King2009; Goad and Shimada Reference Goad, Shimada, Iyer and Kusmer2014; Weber Reference Weber, Macauley, Noodin and Valentine2016). That said, even if complex /s/ is removed from consideration, Blackfoot does not have a simple CV structure, as it distinguishes light syllables ((C)V) from heavy syllables ((C)VV and (C)VC) (Elfner Reference Elfner2006, Weber Reference Weber, Macauley, Noodin and Valentine2016), and it contrasts length in both consonants and vowels (Taylor Reference Taylor1969, Elfner Reference Elfner, Mezhevich, Bliss and Dobrovolsky2005 Derrick Reference Derrick2007). The type and number of segments permitted in coda position is restricted, and complex onsets and codas are not permitted. Current analyses, then, might put Blackfoot in the mid-range of permissible syllable structures, with more complex options than simple CV, but not as wide a range as it might seem on the surface. Blackfoot, then, does not easily fit into this first criterion, although it does not seem to be far towards the syllable-timed end of the continuum either.
b. no or few long vowels permitted in unaccented syllables;
Both long and short vowels are permitted in unaccented syllables (e.g., mótookisi ‘kidney’), but note that in unaccented syllables, short vowels are devoiced, while long vowels are partially devoiced (Derrick p.c., Weber Reference Weber, Macauley, Noodin and Valentine2016), as we will see in d) below. Blackfoot then does not strictly fit this criterion, but it is clear that there is phonetic adjustment to long vowels when in unaccented syllables.
c. accented syllables correlated with pitch contours;
Blackfoot is normally described as a pitch-accent language, and Taylor (Reference Taylor1969: 48–51) outlines in detail how pitch contours align with accented syllables. The system is rather complex and interested readers are referred to the relevant section in Taylor (Reference Taylor1969). Prominence in Blackfoot has different phonetic realizations depending on the location of the syllable in the word (Taylor Reference Taylor1969, Stacy Reference Stacy2004, Frantz Reference Frantz2009, Miyashita and Fish Reference Miyashita and Fish2015). The important insight is that pitch contours are in fact correlated with accented syllables in Blackfoot; there are rises and falls associated with Blackfoot stressed syllables, following the criteria laid out in Dauer (Reference Dauer1987).
d. vowels in unaccented syllables are reduced or centralized
This property contrasts with the situation in syllable-timed languages, where vowel reduction is rarely found. Although Blackfoot appears not to have the English-type unstressed vowel centralization and reduction attested in stress-timed languages, it does appear to reduce vowels in its own way. Blackfoot short vowels are widely attested to be devoiced or deleted word-finally both auditorily (Taylor Reference Taylor1969, Frantz Reference Frantz2009) and acoustically (Van Der Mark Reference Van Der Mark2003, Gick et al. Reference Gick, Bliss, Michelson and Radanov2012) in unstressed position. This devoicing can be extreme to the point that even a previous sonorant is deleted or devoiced (Stacy Reference Stacy2004). Consider the examples in (1), where the relevant devoiced segments are shown in boldface.
Blackfoot (Stacy Reference Stacy2004: 13-14)
- (1)
a. ó’kaattsoohsiwa
[óʔka:t:suxsiw̥ḁ]
'he/she overworked him/herself
b. ni'táa'siwa
[niʔtáaʔsi]
'it is/was one mile'
Furthermore, Taylor (Reference Taylor1969: 77) states that short unstressed /ɪ/ is regularly deleted between most obstruents and /s/, yielding the complex clusters described above. To sum up, this pattern of devoicing and deletion of vowels in Blackfoot in a number of unstressed contexts is one of Dauer's indicators of a stress-timed language.
e. consonant neutralization or reduced allophones in unstressed syllables.
The phonological descriptions of Blackfoot include contrastive consonant length both intervocalically and preconsonantally (Elfner Reference Elfner2006: 38), and there is no known consonant neutralization in unstressed syllables. Weber (p.c.) reports that while there are cases of consonant lengthening and shortening, these are rarely reflected in the orthography, and their origins remain poorly understood.
Taken together, Blackfoot does fit many of the criteria that would place it in the category of stress-timed languages, and we might therefore expect it to have relatively high Variability Index values. That said, it is not a clear-cut case, and more research is needed in order to learn about the rhythmic patterns of Blackfoot. Although we cannot test the VI of contemporary monolingual Blackfoot speakers, it is reasonable to hypothesize that Blackfoot-speaking learners of English would transfer rhythmic patterns from their native Blackfoot to their learned English, and that these patterns may have been passed on to their children today, given other examples of language transfer to a first generation of English speakers (Hoffman and Walker Reference Hoffman and Walker2010, Rosen et al. Reference Rosen, Nicole, Lanlan and Sky2015), and especially of prosodic transfer (Low et al. Reference Low, Grabe and Nolan2000, Coggshall Reference Coggshall2008, Shoustermann Reference Shousterman2014). Given this evidence, we posit that the prosodic differences reflected in the VI stem from language transfer from Blackfoot that has persisted even in the monolingual English children of native Blackfoot speakers. The strong community ties reinforced by outside institutionalized racism and prejudice mentioned in section 2 have contributed to the development of ethnolectal features. These features, in turn, are reinforced at the local level.
6. Conclusion
We have shown a quantifiable difference in prosodic rhythm between the English spoken by the Blackfoot and that spoken by those of settler descent, where speakers of Blackfoot English in Southern Alberta demonstrate higher Variability Index values, suggesting that their speech is more stress-timed than the speech of SE speakers. Because we were not able to test the rhythmic patterns of contemporary monolingual Blackfoot speakers, we instead related this difference in rhythmic patterning to phonological accounts of rhythm, hypothesizing that the Blackfoot language falls along the continuum with highly stress-timed languages, and that the relatively higher VI values reflect language transfer effects which have persisted through to the generation of monolingual English speakers of Blackfoot descent. We hope to confirm this in future work with existing older recordings of fluent, if not monolingual, speakers of Blackfoot.
Appendix Mean VI values for each speaker for interview and reading passage data
