1 INTRODUCTION
This paper examines a particular vowel pattern in Louisiana Regional French, /ɛ/ lowering to [æ] before /r/, tracking its use over the last four decades. An example of this process is the word père /pɛr/ ‘father’, which can either be pronounced with the non-lowered variant [pɛr] or with the lowered variant [pær]. The study uses data from recordings made in 1977 and 2010–2011 in a small geographic region, altogether examining 436 tokens from 32 speakers born between 1888 and 1953. It is only by considering sociolinguistic factors not included in existing analyses (birth year and home town) that previously unreported patterns emerge from the data.
Dialectal variation in Louisiana Regional French forms the core of numerous linguistic studies, whether they be in the domain of syntax or phonology. Some studies have also focused on /ɛ/ lowering before /r/. These studies have compared speech from different Louisiana parishes (the Louisiana equivalent of counties) to see if the use of this variable changes based on a person's home parish (Dubois, Reference Dubois, Valdman, Auger and Piston-Hatlen2005; Salmon, Reference Salmon2009), or have reported on a single parish (Guilbeau Reference Guilbeau1950, Reference Guilbeau1958). This study focuses on one area in particular, lower Lafourche Parish, examining variation at the municipal, rather than the parish, level. The motivation for this narrow geographical scope comes as a result of recordings made in lower Lafourche Parish. There, some (but not all) speakers appear to lower /ɛ/ before /r/ much more frequently than is predicted in the literature.
1.1 Language variety
Louisiana Regional French (more commonly known as Cajun French) is an orally-transmitted language, and so the vast majority of native speakers do not read or write in French. It is also a severely endangered language according to the UNESCO nine-factor Language Vitality and Endangerment (LVE) scale (Brenzinger et al., Reference Brenzinger, Dwyer, de Graaf, Grinevald, Krauss, Miyaoka, Ostler, Sakiyama, Villalón, Yamamoto and Zapeda2003), which means that it is no longer being passed on to children, is spoken by a minority of the population in a limited number of settings, and is not necessarily available through education or popular media. Indeed, while the Council for the Development of French in Louisiana (CODOFIL), created in 1968, has worked to promote French language use in Louisiana, the objective has been to teach and use outside varieties of French with higher perceived social status, such as Canadian, Belgian and France French; more recently, the aim has turned to using (‘Standard’) French to attract Francophone tourists to the state.
The present study examines Louisiana Regional French speakers, but it is important to note that this language variety is one of at least three that exist in Louisiana. For instance, Louisiana Plantation Society (‘Colonial’) French (Picone, Reference Picone1998) is more closely related to ‘Standard’ French (SF) and has practically disappeared as its affluent speakers shifted to English (Brasseaux, Reference Brasseaux1992). Also, Louisiana Creole French is linguistically similar to Haitian Creole and Antillean Creole, but developed independently among the African slaves brought to Louisiana in the 1700s (Neumann, Reference Neumann1985). Unfortunately, Louisiana Creole French has not enjoyed the positive status, publicity, and revitalization efforts consecrated to Louisiana Regional French, which is likely due to long-standing ethnic disparities and prejudices in the state (Trépanier, Reference Trépanier1991: 164). As a result, Louisiana Creole French is more severely endangered than Louisiana Regional French. All three varieties have been in heavy contact with English for over a century; this language contact, along with its effects on Louisiana Regional French, is well documented in the academic literature (Picone and Valdman, Reference Picone, Valdman, Valdman, Auger and Piston-Hatlen2005; Brasseaux, Reference Brasseaux2005).
Louisiana Regional French shares some interesting similarities with Jersey Norman French, another endangered variety spoken on one of the Channel Islands between England and France. Known as Jèrriais to native speakers, roughly 3 percent of the island's population still speaks it in private settings, and the majority of speakers are over 60 years of age (Jones, Reference Jones2005: 4). As with Louisiana Regional French, there is widespread dialect variation (Jones, Reference Jones2001: 38), little to no contact with SF (Jones, Reference Jones2012: 201), and no official federal backing for the language variety (Jones, Reference Jones2001: 73).
There are also important differences between Louisiana Regional French and Jersey Norman French. For instance, English has only come into regular and heavy contact with French in Louisiana over the last 150 years, but it has been a part of daily life since the fifteenth century on Jersey (Jones, Reference Jones2005: 4). Also, while Louisiana French speakers cannot read in French, most speakers of Jèrriais can (Jones, Reference Jones2001: 52). Finally, language revitalization efforts in Louisiana are complicated by disagreements regarding which variety of Louisiana Regional French to implement as the norm (Brown, Reference Brown, Valdman, Auger and Piston-Hatlen2005: 399). However, on Jersey, speakers can agree on one standard variety (Jones, Reference Jones2001: 87), and the majority of Jèrriais speakers want the language preserved (Jones, Reference Jones2001: 62) and in schools (Jones, Reference Jones2001: 64).
1.2 Region
Lafourche Parish sits at the southwestern edge of the state of Louisiana, running North to South from below the Mississippi river to the Gulf Coast. Three main transportation thoroughfares run down the parish: Highways 1 and 308 for vehicles, and bayou Lafourche for watercraft (a bayou is a slow-moving waterway in the Louisiana landscape). The land itself lies between 0 and 1 metres above sea level, and is increasingly marshland as one travels south. Below the town of Golden Meadow, for instance, the road becomes a ribbon of land completely surrounded by marsh. As a result, traditional ways of life, especially in the southern half of the parish, focused on shrimping and fishing in the summer, and trapping in the winter.
The smaller communities in the southern part of the parish, such as Cut Off, Galliano, and Golden Meadow, exhibit a history of endogamy (as evidenced by the prevalence of certain family names), strong religious ties to Catholicism, and prevalent French use. For instance, the churches of Cut Off and Golden Meadow were the last two French-language Catholic institutions in the state before they switched to English record-keeping in the 1950s (Dubois, Leumas, and Richardson, Reference Dubois, Leumas and Richardson2007: 60; Leumas, Reference Leumas2009: 83). Also, before the oil boom in the 1940s, the lower half of the parish was fairly isolated from English-speaking outsiders (Larouche, Reference Larouche1979).
An isogloss boundary exists between Larose and Cut Off, separating the upper half of the parish from the lower half (Dajko et al., Reference Dajko, Crossland, Fort, French, Goss, Johnson, Hinrichs, Powers, Pulwer, Schneider, Shipps, Snider and Williams2008). This reflects settlement patterns of different French-speaking groups. Acadian settlers stayed mostly in the northern and central parts of bayou Lafourche, making their way as farmers (Brasseaux, Reference Brasseaux and Uzee1985; Smith, Reference Smith1934) and trickling down the bayou in smaller numbers. In contrast, many residents of lower Lafourche Parish, particularly those from Golden Meadow, can trace their ancestry to the fishing village of Chenière Caminada at the southeastern tip of the parish, whose Francophone population hailed mainly from France, not Acadia (Pitre, Reference Pitre, Michot and Doucet1996: 58). Accordingly, lower Lafourche Parish French does not exhibit typically Acadian linguistic traits (Baronian, Reference Baronian2005; Byers, Reference Byers1988; Rottet, Reference Rottet2004), with the exceptions of lexically conditioned affrication (e.g., queue [ʧø] ‘tail’, diable [ʤab] ‘devil’) and stylistic variation between [ʒ] and [h] (e.g., je /ʒə/ ➔ [ʒə], [hə] ‘I’).
Figure 1 gives the geographical situation of Lafourche Parish within Louisiana, while Figure 2 provides a parish-specific map of the towns of Lafourche. Chenière Caminada is located to the northeast of Port Fourchon on Highway 1.
Figure 1. The 64 parishes of Louisiana, including the 22 southern parishes of the French Triangle of Acadiana. Note. From U.S. Census Bureau, Census 2000. Adapted with permission, originally in Blainey (Reference Blainey2013).
Figure 2. Lafourche Parish outline with towns, major roads and waterways. Note. From the Lafourche Parish Government website, retrieved December 15, 2012, http://www.lafourchegov.org/parishimages/ParishBlackWhite.jpg. Reprinted with permission.
1.3 /ɛ/ lowering in North American French
/ɛ/ lowering before a rhotic consonant occurs in many varieties of North American French; in several varieties of Canadian French, the lowered vowel can appear when the rhotic precedes another consonant. Walker (Reference Walker1984: 86) notes that this pronunciation pattern is most common in Montreal and the Outaouais area, and ‘is often considered rural, archaic or otherwise stylistically marked’, giving examples like aubarge [SF auberge] (/obɛrʒ/ ➔ [o.barʒ] ‘hostel’). The same lowering is attested in the Acadian French of the Îles-de-la-Madeleine (Falkert, Reference Falkert2010: 176–177) and New Brunswick (Flikeid, Reference Flikeid1984: 182–185); in Ontario French (Hull, Reference Hull1956); and in Manitoba French (Thogmartin, Reference Thogmartin1974).
In his sociolinguistic study of the history of Parisian French, R. Anthony Lodge explains that [er] lowering to [ar] was a feature of vernacular Parisian French in the thirteenth century (Lodge, Reference Lodge2004: 94), and it also appears in the personal correspondence of sixteenth-century aristocrats and royalty (Lodge, Reference Lodge2004: 145). Salmon (Reference Salmon2009: 64) notes that French grammarians have recorded variable e before r since the sixteenth century, and that this variation is likely due to the fact that the French vowel actually stems from two different Latin vowels, /a/ and /ɛ/. Although the French Revolution of 1789 was ‘politically progressive’, Lodge (Reference Lodge2004: 207) observes that it was ‘linguistically conservative’, so that [ar] endured as a stigmatized pronunciation variant in lower-class Parisian French at least until 1821. However, Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005: 294) remarks that the lowered variant became popular in France at the turn of the eighteenth century, a time when many French citizens immigrated to Louisiana. Since the ancestors of Golden Meadow speakers arrived mainly from France, it is reasonable to assume that they were conscious of the longstanding, stigmatized nature of /ɛ/ lowering before /r/, but also that this process was present in their everyday speech.
In Louisiana Regional French, many words undergo the process whereby /ɛ/ lowers to [æ] before /r/, regardless of syllable structure (/r/ can occupy a coda or an onset position). So, a word like frère (/frɛr/ ‘brother’) can be pronounced [frɛr] or [frær] regardless of whether the word-final consonant is resyllabified with a following vowel-initial word.
The phenomenon of /ɛ/ lowering before /r/ figures in early descriptions of Louisiana Regional French, with Alcée Fortier (Reference Fortier1891: 88) noting its presence in both St. Martin and St. Mary Parish. Citing Fortier (Reference Fortier1891), William Read (Reference Read1931: xx) states that ‘the tense vowel in St.-Fr. faire, etc., becomes lax’ in Louisiana Regional French, without specifying geographic boundaries for the phenomenon. The same author, in his mini-dictionary of Louisiana French, lists a single word with a lowered /ɛ/, gouère ([gwæ:r], ‘blond-haired, blue-eyed person’), stating that the word is used in the parishes of Lafayette, Iberia, Avoyelles, St. Landry and St. Martin (Read, Reference Read1938: 74). Other words in the dictionary either do not have an accompanying phonetic transcription, as in sarbotière [SF sorbétière] (‘ice-cream freezer’) (Read, Reference Read1938: 82), or use the non-lowered variant, as in gaspergou [SF casse-burgot] (‘fresh-water drum’) (Read, Reference Read1938: 72). Finally, in their Louisiana French grammar of Lafayette, St. Landry, and St. Martin Parishes, Marilyn Conwell and Alphonse Juilland (Reference Conwell and Juilland1963: 44) assert that [ɛ] is in free variation with [a] before /r/. The authors cite the works of Trappey (Reference Trappey1940), Hurst (Reference Hurst1936) and Montgomery (Reference Montgomery1941) to note that this also occurs in Iberia, St. Charles, and Vermilion Parishes, respectively.
1.4 /ɛ/ lowering in Lafourche Parish
Four previous studies of Louisiana Regional French examine Lafourche Parish speakers and /ɛ/ lowering before /r/, and make different assertions about how often the process takes place. First, Guilbeau (Reference Guilbeau1950: 44–45) gives an early picture of Lafourche Parish Louisiana Regional French, stating that all transcribed Lafourche speakers in his analysis, born between 1858 and 1917 and recorded in the 1940s, lower /ɛ/ before /r/ word-internally at least some of the time. Demographic information is available for Guilbeau's speakers, but he does not provide specific /ɛ/ lowering frequencies for any of them in his description. However, in a more general description of Lafourche Parish French phonology, Guilbeau (Reference Guilbeau1958: 30) does note that syllable-internal /ɛ/ before /r/ most often surfaces as [æ], with only exceptional cases of [ɛ] or [e].
Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005) analyzes a corpus of 120 speakers recorded in 1997, as well as 17 speakers recorded in 1975. The speakers’ birth years range between 1890 and 1977. Thirty of the speakers come from various towns in Lafourche Parish, but quantitative comparisons occur at the parish, rather than at the municipal, level. In contrast to Guilbeau (Reference Guilbeau1958), Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005: 298) concludes that Lafourche Parish speakers’ pattern of use, with only 26% of all attestations using the lowered variant, does not clearly distinguish them from speakers belonging to Vermilion, St. Landry, and Avoyelles parishes. However, without differentiating between speakers’ towns of origin, the analysis does not explore the possibility of dialectal variation at the municipal level, presenting Lafourche Parish as homogenous with respect to /ɛ/ lowering rates.
Finally, Salmon (Reference Salmon2007, Reference Salmon2009) uses a subset of the recordings from Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005), examining the speech of 29 female speakers born between 1892 and 1947. Six of the women come from Lafourche Parish, hailing from the towns of Galliano (1), Cut Off (4), and Raceland (1) (Salmon, personal communication, January 19, 2013), and their birth years range from 1910 to 1938; all six were recorded in 1997. The small sample size does not allow for any sociolinguistic conclusions to be drawn regarding the influence of age or birthplace on allophone use in Lafourche Parish, but Salmon does note, based on two speakers, that younger speakers in Lafourche Parish rarely use the lowered variant, a (Salmon, Reference Salmon2009: 88). Her results are therefore in keeping with Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005).
Thus, while these studies recognize that this variable may be controlled by sociolinguistic factors, quantitative measures are scarce and require further precision. All three researchers use speakers from throughout Lafourche Parish to make generalizations about /ɛ/ lowering patterns, without considering the possibility that certain locales may differ in their deployment of this process. The current study will examine the speech of informants from Golden Meadow and the immediately surrounding area in lower Lafourche Parish, using recordings from 1977 and 2010–2011. As the present work controls the factors of locale and ethnic identity while contrasting sex, time period, birth year and speech style among 32 speakers, it is in a position to show which of these factors play a decisive role in pronunciation patterns.
2 HYPOTHESES
Results from the existing academic literature on /ɛ/ lowering before /r/ lead to three hypotheses about the process in Lafourche Parish Louisiana Regional French. The first is that /ɛ/ will not lower to [æ] frequently for speakers interviewed in the 1970s or 2000s, given the results in Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005) and Salmon (Reference Salmon2007, Reference Salmon2009), but it will generally lower to [æ] for speakers interviewed in the 1940s (Guilbeau, Reference Guilbeau1958). The second hypothesis is therefore that the point in time at which an interview takes place will play a role in a speaker's rate of /ɛ/ lowering. The third hypothesis comes from sociolinguistic research on Lafourche Parish Francophones (e.g., Carmichael, Reference Carmichael2007, Reference Carmichael2008; Dajko, Reference Dajko2009; Larouche, Reference Larouche1979, Reference Larouche1980; Rottet, Reference Rottet1995): sociolinguistic factors such as sex, age, and speech register will influence pronunciation patterns in Lafourche Parish Louisiana Regional French.
3 METHOD
3.1 Speakers
This analysis draws on two sets of recordings. The first set of interviews took place in 1977, when Alain Larouche spoke with residents of Golden Meadow and the surrounding area for his master's research (Larouche, Reference Larouche1979, Reference Larouche1980). His purpose was to uncover the social fabric of Golden Meadow, and so the interviews focus on local pastimes, family trees and the history of the area. Since the interviews were not conducted as part of a linguistic study, there was less of a focus on sound quality, speakers’ language use or language tasks, and only one speech register is available for each speaker. The interviews were part of a Canadian multi-university documentation effort, the Louisiana Project (Projet Louisiane, Breton and Louder, Reference Breton and Louder1979; Louder and Waddell, Reference Louder and Waddell1979). First made on cassette tape, the sound files have since been digitized at the University of Louisiana at Lafayette's Center for Cultural and Eco-Tourism (CCeT) (http://ccet.louisiana.edu/tourism/cultural/archives.html). Twelve interview files (seven men, five women) have been selected from the collection because they are sufficiently clear for phonetic and phonological analysis.
Another set of digital recordings were made in 2010 and 2011, when Darcie Blainey interviewed residents of the Golden Meadow area for her doctoral research (Blainey, Reference Blainey2013). The objective was to restrict the geographical area as much as possible in order to control for any pronunciation variation stemming from diatopic factors. Twenty interviews, evenly divided by the interviewee's sex, have been selected for analysis. Twelve of the recordings form part of the Phonology of Contemporary French project (la Phonologie du Français Contemporain [PFC], Durand, Laks and Lyche, Reference Durand, Laks, Lyche, Pusch and Raible2002, Reference Durand, Laks, Lyche and Williams2005), an international linguistic initiative to document and understand French as it is spoken worldwide.
Since the vast majority of Louisiana Regional French speakers do not read or write in their native tongue, the PFC-Louisiana protocol eliminates any text-based tasks in favor of conversational tasks: guided (formal) conversation, free (informal) conversation, sentence translation, and word translation (Klingler and Lafleur, Reference Klingler, Lafleur, Detey and Nouveau2007). Interviewees carry on the formal conversation with a community outsider and the informal conversation with a community insider in order to maximize the difference between conversational speech registers (Labov, Reference Labov, Baugh and Sherzer1984). The current analysis examines data from the conversational portions of the interviews in order to compare them with conversational data in the Larouche corpus. Also, the code name for each speaker consists of two letters followed by a number as the speaker identifier, followed by the speaker's birth year and sex. For example, DC1-1888-M is a male speaker born in 1888.
Eleven of the 12 interviewees recorded in 1977 were born between 1888 and 1939 in the area surrounding Golden Meadow, and all 12 were living in either Chenière Caminada, Golden Meadow or Galliano when they were interviewed. JB1-1923-M represents the only exception among the 1977 speakers, because he was born in the state of Arkansas. However, he learned French upon moving to Golden Meadow, married local native speaker MB1-1924-F, and does not display non-native characteristics in his Louisiana Regional French. JB1-1923-M is therefore included in the analysis.
As for the 2010–2011 interviewees, they were all born and living in either Golden Meadow, Galliano or (more rarely) Cut Off, and their birth years range between 1921 and 1953. All but one (GG1-1953-M) were retired when they volunteered to be interviewed, and their level of education varies between university degrees (GG1-1953-M, JA1-1926-M) and no schooling (CG1-1928-M, GG2-1929-F).
3.2 Speech tokens
The study uses five minutes of phonemically transcribed speech for each speaker and speech style, resulting in 260 minutes of speech overall (12 speakers x 1 speech style x 5 minutes; 20 speakers x 2 speech styles x 5 minutes). This database of over 70,000 sounds yields 436 tokens of potential /ɛ/ lowering for analysis.
The two possible surface forms for /ɛ/ before /r/ in the corpus are [ɛ] and [æ]. The following steps were taken to determine the correct surface form for each speech token. First, the author listened to each potential site of /ɛ/ lowering and identified the surface variant based on perceptual cues. Second, any cases of uncertainty were verified with a native speaker of lower Lafourche Parish.
4 RESULTS
4.1 1977 interviews
In the 1977 corpus, there are 77 tokens for analysis. It should be noted that some speakers contribute only a few tokens to the data set; nevertheless, a consistent pattern still emerges. Specifically, there is a sharp, age-based divide in the rate of /ɛ/ lowering: speakers born between 1888 and 1920 lower /ɛ/ between 0% and 27% of the time, and speakers born after 1920 lower /ɛ/ between 50% and 100% of the time. Table 1 summarizes this information.
Table 1. /ɛ/ Lowering to [æ] before /r/ for 1977 Speakers. * = Birth year within 2010–2011 speakers’ birth year range. CC = Chenière Caminada; CF = Cut Off; EL = Elliot, Arkansas; GL = Galliano; GM = Golden Meadow; LV = Leeville.
Indeed, there is a statistically significant positive correlation between a speaker's birth year and his or her rate of /ɛ/ lowering to [æ] before /r/, rs = .66, N = 12, p = .02. This means that as a speaker's age decreases, the rate of [æ] before /r/ generally increases, and this is true for both male and female speakers. A visual representation of this relationship is given in Figure 3.
Figure 3. Rate of /ɛ/ Lowering to [æ] before /r/ for 1977 Speakers by Birth Year. Striped symbols represent speakers born north of Golden Meadow.
The regression lines for male and female speakers in Figure 3 indicate that there is a cross-over interaction between birth year and sex. The regression lines also show that there is potentially a main effect of sex. Indeed, if the data are organized by the variable of each speaker's sex, the distribution attains statistical significance, with women lowering /ɛ/ more often than men, χ2 (1, N = 77) = 8.24, p = .04. However, it is clear from Figure 3 that only one speaker born before 1921 in the 1977 corpus is a woman, and among the younger speakers, men and women pattern similarly regarding the rate of /ɛ/ lowering. Therefore, without more female speakers born before 1920 in the sample, it is not possible to make conclusive statements about differences between male and female speakers’ /ɛ/ lowering rates in the 1977 corpus.
Regarding the variables of birthplace or place of residence, which contrast speakers of Lafourche Parish from north of Golden Meadow to those from the town of Golden Meadow or further south, chi-square tests of independence indicate that neither is able to explain differences in /ɛ/ lowering in 1977 speech. Specifically, speakers born north of Golden Meadow do not lower /ɛ/ significantly more than speakers born south of or in Golden Meadow, χ2 (1, N = 77) = 3.05, p = .07. The same is true for a speaker's place of residence at the time of the interview, because speakers living north of Golden Meadow do not lower /ɛ/ significantly more often than speakers living in or south of the town, χ2 (1, N = 77) = 1.30, p = .21. It should be noted that only three speakers were born north of Golden Meadow (JB1-1923-M, AG1-1928-M, ET1-1931-F), and only two speakers were living north of Golden Meadow at the time of the interview (AG1-1928-M, ET1-1931-F). Just the same, the tests of significance suggest that birthplace and place of residence are not factors that can reliably explain rates of /ɛ/ lowering in 1977 Louisiana Regional French, and this is supported by the finding that year of birth is positively correlated with the rate of /ɛ/ lowering in this time period.
Thus, in 1977 Louisiana Regional French, the rate of /ɛ/ lowering before /r/ forms an important relationship with a speaker's age, so that speakers born before 1921 rarely or never lower /ɛ/ before the rhotic consonant, whereas speakers born after 1921 do so between 50% and 100% of the time. Statistical tests have confirmed that this pattern is not related to geography in the 1977 corpus. For instance, ES2-1904-M was born in Golden Meadow and was still living there at the time of the interview, but produced a single [æ] in place of [ɛ] out of nine cases (a rate of 11%), while CS1-1939-M, another informant born in Golden Meadow and still living there in 1977, used [æ] in place of [ɛ] in six out of seven instances (a rate of 86%). Also, AG1-1928-M, who was born in Cut Off and living in Galliano at the time of the interview, uses [æ] exclusively (4 of 4 tokens). The next two sections will examine the data from the more formal and less formal speech registers in the 2010–2011 interviews.
4.2 2010–2011 formal speech
When speaking to a community outsider, the 20 interviewees from the 2010–2011 corpus produce a total of 168 tokens for analysis. As with the 1977 interviewees, some speakers produce only a few tokens; however, the data available for these speakers pattern similarly to the rest of the group, and so are included in the analysis. Unlike the data from 1977, the rate of /ɛ/ lowering cannot be ordered by a speaker's age, as evidenced by a non-significant rank correlation between birth year and the rate of /ɛ/ lowering, rs = .02, N = 20, p = .93. Since all speakers in the 2010–2011 corpus are born after 1920, this result appears to conform to the generational divide seen in the 1977 data. There are, however, some speakers who never lower /ɛ/, and this cannot be explained by their age alone. Sex is not an active variable, either, with men and women lowering /ɛ/ at roughly the same rate, χ2 (1, N = 168) = .40, p = .32. An overview of the data is presented in Table 2.
Table 2. /ɛ/ Lowering to [æ] before /r/ in 2010–2011 Formal Speech. * = Birth year within 1977 speakers’ birth year range. CF = Cut Off; GL = Galliano; GM = Golden Meadow.
The pattern becomes more apparent if speakers are grouped by birthplace. Four of the 20 speakers were born north of Golden Meadow. Chi-square tests of independence show that speakers born in or south of Golden Meadow lower /ɛ/ significantly more often (N = 108/140, 77.1%) than speakers born north of Golden Meadow (N = 0/28, 0%), χ2 (1, N = 168) = 60.48, p = .00. Furthermore, even though the difference is not as great, the town in which a speaker was living at the time of the interview is also a significant variable, with speakers in or south of Golden Meadow lowering /ɛ/ significantly more often (N = 74/97, 76.3%) than speakers north of Golden Meadow (N = 34/71, 47.9%), χ2 (1, N = 168) = 14.40, p = .00.
Figure 4 provides a visual representation of the rate of /ɛ/ lowering by speaker in the formal speech style. In keeping with the statistical test results, the regression lines show that there is no clear relationship between a speaker's year of birth or sex and his or her rate of /ɛ/ lowering; although there is a cross-over interaction between birth year and sex, there are no main effects present in the data. The clusters of tokens do clearly show that a speaker's birthplace effectively separates the data into those who lower /ɛ/ and those who do not, since the four speakers born north of Golden Meadow are the only speakers never to lower /ɛ/.
Figure 4. Rate of /ɛ/ Lowering to [æ] before /r/ for 2010–2011 Speakers by Birth Year, Formal Conversation. Striped symbols represent speakers born north of Golden Meadow.
4.3 2010–2011 informal speech
In the less formal 2010–2011 speech register, speakers are conversing with a Golden Meadow native; this community insider is often also a family member or friend of the person being interviewed. Only one speaker, KC1-1941-M, does not have any tokens available for analysis. Altogether, the informal speech register contributes 191 tokens to the analysis. Table 3 summarizes the lowering rate for each speaker, maintaining the same groupings as were made for Table 2.
Table 3. /ɛ/ Lowering to [æ] before /r/ for 2010–2011 Informal Speech. ↓ = significant decrease in /ɛ/ lowering compared to the 2010-F Speech data. * = Birth year within 1977 speakers’ birth year range. CF = Cut Off; GL = Galliano; GM = Golden Meadow.
Starting with non-significant variables, recall that sex was not an active variable in explaining /ɛ/ lowering patterns in the more formal speech style. A chi-square test of independence confirms that this variable cannot help to explain variation in less formal conversation either, since men do not lower /ɛ/ significantly more often than women, χ2 (1, N = 191) = .35, p = .33. The variable of birth year provides another set of results that are similar to those for the 2010–2011 formal speech register. In particular, there is non-significant rank correlation between a speaker's birth year and his or her rate of /ɛ/ lowering when speaking to a community insider, rs = .15, N = 19, p = .54. This suggests that birth year is not an active variable in explaining different rates of /ɛ/ lowering in 2010–2011 speech.
In contrast to the more formal speech register, a person's place of residence does not have a significant effect on his or her /ɛ/ lowering behaviour in less formal conversation, with people living in or south of Golden Meadow only lowering /ɛ/ slightly more often than those living north of the town, χ2 (1, N = 191) = .37, p = .32.
The only significant factor in 2010–2011 informal speech is a speaker's birthplace. Specifically, speakers born in or south of Golden Meadow lower /ɛ/ 65.2% of the time (N = 105/161), while speakers born north of Golden Meadow never lower /ɛ/ (N = 0/30). A chi-square test of independence shows this to be highly significant, χ2 (1, N = 191) = 43.45, p = .00. This pattern is clearly visible in Figure 5. The regression lines in Figure 5 show the same cross-over interaction between sex and age that has been seen in the other data sets, but again, there are no main effects.
Figure 5. Rate of /ɛ/ Lowering to [æ] before /r/ for 2010–2011 Speakers by Birth Year, Informal Conversation. Striped symbols represent speakers born north of Golden Meadow.
4.5 Comparison of data sets
A chi-square test of independence shows that the 1977 data set does not differ significantly from the 2010–2011 formal speech style, χ2 (1, N = 245) = 2.70, p = .07. The 1977 data are even more similar to the 2010–2011 informal speech style, with lowered /ɛ/ appearing 53.2% of the time (N = 41/77) in 1977 and 55.0% of the time (N = 105/191) in 2010–2011 informal conversation; there is no statistical difference between the two data sets, χ2 (1, N = 268) = .07, p = .45. However, there is a statistically significant difference between the two speech registers of the 2010–2011 interviews, with speakers lowering /ɛ/ significantly more often in careful speech (N = 108/168, 64.3%) than in less careful speech (N = 105/191, 55.0%), χ2 (1, N = 359) = 3.21, p > .05. Interestingly, after performing a two-tailed Fischer's exact test for each speaker in the 2010–2011 corpus, only one speaker, JR1-1947-M, had statistically significant differences in /ɛ/ lowering between the two speech styles (χ, df = 1; p = .02). Thus, while the group as a whole lowers /ɛ/ less often when speaking to a community insider, suggesting a style-based contrast, this is generally not statistically significant at the individual level.
5 DISCUSSION
5.1 1977 trends
In the 1977 recordings, speakers born before 1921 lower /ɛ/ to [æ] before /r/ between 0% and 27% of the time; speakers born during or after 1921 lower /ɛ/ between 50% and 100% of the time. A potential explanation for the significant generational divide arises when we consider the history of Golden Meadow, and the natural disasters that led to the formation of its core population.
As mentioned in the /ɛ/ lowering in North American French section, lowering /ɛ/ to [æ] before /r/ was seen as low-class and rural in nineteenth-century France, a time at which many speakers left France for Louisiana. The cosmopolitan fishing settlement of Chenière Caminada, located on the barrier islands at the base of bayou Lafourche, included many French families. These people suffered through two major natural disasters. First, in 1893, a devastating hurricane obliterated Chenière Caminada. Any surviving residents resettled further north on the bayou in the towns of Leeville and Golden Meadow (Falls, Reference Falls1893). Second, in 1915, another hurricane virtually destroyed Leeville (Roth, Reference Roth2010: 30), forcing survivors to move north again to Golden Meadow. This substantial population increase allowed for the establishment of many new Golden Meadow businesses and institutions in the 1920s, such as a Catholic church, shrimp and ice factories, as well as a grocery store, bank, and movie theatre (Rome, Reference Rome, Michot and Doucet1996: 72). The 1930s saw even more development with the discovery of oil wells in the parish; electricity, the petroleum industry and education were not long in following.
The hurricanes represented major ruptures in everyday life for people living south of Golden Meadow, and the business developments in the 1920s and 1930s firmly established Golden Meadow as a community on bayou Lafourche. People born in the 1920s in lower Lafourche Parish thus experienced a much different everyday reality growing up than the generation before. It is possible that this younger generation sought to distinguish itself from older speakers by using a salient phonetic feature such as /ɛ/ lowering at a much higher rate. These events and social changes help to explain why there exists a generational divide in /ɛ/ lowering patterns among those speakers interviewed in 1977.
5.2 1977 to 2010–2011 trends
This study includes a subset of speakers born during the same time period (1921–1939) but interviewed nearly forty years apart (in 1977 and in 2010–2011). Isolating the /ɛ/ lowering rates for these speakers allows a comparison of the data sets based solely on the point in time at which a speaker was interviewed. There are seven speakers from 1977 who fall into this age range (three men and four women), while 13 speakers from 2010–2011 (six male and seven female) are born in the same period of time. Figure 6 combines the /ɛ/ lowering rates for a subset of speakers interviewed in 1977 and in 2010–2011 in the less formal speech setting, since these two data sets are not significantly different from one another, and highlights speakers born north of Golden Meadow by representing them with striped symbols.
Figure 6. Rate of /ɛ/ Lowering to [æ] before /r/ for 1977 and 2010–2011 Speakers (Informal Conversation) by Birth Year (1921–1939). Striped symbols represent speakers born north of Golden Meadow.
Figure 6 shows that for speakers within this age range who were interviewed in 1977, the same pattern of /ɛ/ lowering applies regardless of birthplace. For instance, AG1-1928-M was born and living north of Golden Meadow when he was interviewed, but always lowers /ɛ/; ET1-1931-F was also born and living north of the town when she was recorded, but lowers /ɛ/ 50% of the time. In 1977, then, residents of the towns of Galliano and Cut Off did not use this pronunciation pattern to distinguish themselves from the residents of Golden Meadow.
It is clear that this pattern is different in 2010–2011 for speakers born during the same time period in the same tight geographical area. Speakers interviewed in 2010–2011 employ the process of /ɛ/ lowering to make a strong divide between Golden Meadow and two otherwise comparable settlements in south Lafourche Parish. In 2010–2011, /ɛ/ lowering has therefore come to characterize the speech of people born in or south of Golden Meadow, and is no longer present among people born in Cut Off or Galliano. While not all speakers are necessarily conscious of this isogloss boundary, the community insider performing the 2010–2011 informal interviews delighted in eliciting this shibboleth from Galliano and Cut Off natives in order to show how their French differed from that of Golden Meadow. There is, therefore, at least some awareness within the lower Lafourche Parish Francophone communities that this pronunciation divide exists, and that it represents a strong town-based identity for residents of Golden Meadow.
5.3 Comparison with previous research
It is possible to compare the 2010–2011 data from the present study to a subset of Salmon's (2007: 187–190) /ɛ/ lowering data. In particular, six of Salmon's female speakers come from Lafourche Parish. These women were born in Galliano, Cut Off, and Raceland between 1910 and 1938. It is not possible to consider the effect of time period in the comparison, because all of Salmon's Lafourche speakers were interviewed in 1997. However, her speech sample represents an intermediate point between the 1977 and 2010–2011 recordings in the present study.
The results in Salmon (Reference Salmon2007: 187–190) indicate that Lafourche Parish Louisiana Regional French speakers lower /ɛ/ between 0% and 31.7% of the time in more careful speech with a community outsider, and between 0% and 8.6% of the time when speaking to a native Louisiana Regional French speaker. Naturally, these rates of /ɛ/ lowering are much lower than those seen in the present study, because in 2010–2011, /ɛ/ lowering has been shown to be an active identity marker for people born in or south of Golden Meadow, and all interviewees in Salmon (Reference Salmon2007, Reference Salmon2009) were born north of Golden Meadow. Also, Salmon (Reference Salmon2007, Reference Salmon2009) notes the same statistically significant decrease in /ɛ/ lowering when comparing careful and less careful speech, which is in accordance with the current study's findings.
Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005: 298) reports that Lafourche Parish speakers lower /ɛ/ 26% of the time. Although it is not possible to directly compare the current study's findings with those of Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005), it is reasonable to expect that the differences in /ɛ/ lowering rates between the two investigations stem from the same demographic difference, since Salmon's (Reference Salmon2007, Reference Salmon2009) speakers come from the same corpus used in the larger Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005) study. The current study's findings are thus not incompatible with these previous investigations.
Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005) and Salmon (Reference Salmon2007, Reference Salmon2009) divide their speakers by generation, basing birth year ranges on language learning experiences. One such generational division separates the ‘elders’ (doyens), born between 1905 and 1915, from the ‘old generation’ (aînés), born between 1919 and 1933 (Dubois, Noetzel and Salmon, Reference Dubois, Noetzel, Salmon, Brasseur and Falkert2005: 29). The current study has shown that in 1977 Louisiana Regional French, speakers born before 1921 rarely or never lower /ɛ/ before /r/, while speakers born after 1921 do so between 50 and 100% of the time. This trend appears to correspond to the doyen – aîné generational divide in Dubois (Reference Dubois, Valdman, Auger and Piston-Hatlen2005) and Salmon (Reference Salmon2007, Reference Salmon2009). In the current investigation, there are no other generation-based differences in /ɛ/ lowering. Speakers born after 1933, termed the ‘middle-aged generation’ (cadets, 1934–1951) and ‘young generation’ (benjamins, 1953–1978) in previous studies (Dubois and Noetzel, Reference Dubois and Noetzel2005: 133; Dubois, Reference Dubois, Valdman, Auger and Piston-Hatlen2005: 288–290), do not lower /ɛ/ any more or less than speakers born before 1933.
Guilbeau's (Reference Guilbeau1958) lower Lafourche Parish informants were interviewed in the 1940s, and would therefore be expected to lower /ɛ/ between 50% and 100% of the time according to the present analysis. While Guilbeau (Reference Guilbeau1958) does not give quantitative measures of /ɛ/ lowering in his article, he does remark that this process happens most of the time for speakers in the lower half of the parish. This matches the current study's findings exactly. Therefore, some previous sociolinguistic research regarding /ɛ/ lowering in Lafourche Parish French presents different findings from the present study because of demographic differences in the population samples. Otherwise, earlier descriptions of this process in lower Lafourche Parish are compatible with the patterns found in the current analysis.
6 CONCLUSION
This sociolinguistic study has used 12 recordings from 1977 and 20 recordings from 2010–2011 in order to explore the process of /ɛ/ lowering before /r/ in lower Lafourche Parish. By taking birthplace and place of residence into account, this analysis has succeeded in showing a previously undocumented isogloss boundary between Golden Meadow and Galliano. The historical aspect of the examination has revealed the social roots of the /ɛ/ lowering process, as well as a change in the factors controlling /ɛ/ lowering rates over the past 40 years.
With reference to previous studies, this paper laid out three new hypotheses regarding /ɛ/ lowering in lower Lafourche Parish. The first, that /ɛ/ would lower to [æ] more frequently for speakers interviewed in the 1940s than for speakers interviewed in the 1970s and 2000s, was not able to be tested by the present analysis and awaits future examination. The second hypothesis, that the point in time at which speakers were interviewed would play a role in the process of /ɛ/ lowering before /r/, proved to be correct: the factors determining the rate of /ɛ/ lowering changed between the 1977 and 2010–2011 recordings. The third hypothesis was that sociolinguistic factors would play an active role in the rate of /ɛ/ lowering. This hypothesis was also correct, but different sociolinguistic factors were statistically significant for different subsets of the data.
Thus, what was a generational marker in 1977 has become a geographical marker in 2010–2011, demonstrating that /ɛ/ lowering to [æ] before /r/ has changed in meaning over time. Results from the 2010–2011 data also suggest that the formality of the speech event has a significant effect on the rate of /ɛ/ lowering, with speakers generally using the lowered variant [æ] less often in less careful speech, despite the severely endangered status of Louisiana Regional French. This may be due to the fact that in the informal speech register, performance of town affiliation identity is not as crucial, because speakers are confident that the interviewer is already aware of their community membership. Finally, the current study's results refine earlier statements regarding /ɛ/ lowering in Lafourche Parish Louisiana Regional French, showing that in the southern half of the parish, this variable represents an isogloss boundary between Golden Meadow and Galliano.
Louisiana Regional French speakers, like speakers of Jèrriais, ‘are less likely to be influenced by any prescriptive norms or intuitive tendencies associated with the standard than speakers of the dialects of mainland France’ (Jones, Reference Jones2012: 201). The exploration and analysis of seriously endangered varieties of French, especially those varieties whose speakers have little contact with a written code, allows for interesting insights into sound pattern changes, innovation and variability over time.
