1.1 Background/Literature Review

1.1.1 Geographic Location

Appalachia stands out as a distinctive region, with differences discernible from both MUSE and SAE (Labov et al., 2006; Reed, 2016; Wolfram & Christian, 1976; among others). According to the Appalachian Regional Commission (2024), Appalachia stretches across 13 states from Southern New York to Northern Mississippi, covering over 206,000 square miles. As seen in Figure 1, Labov et al. (2006) divide most of Appalachia into the southern dialect region known as “The South,” with a majority additionally in the sub-region of the “Inland South,” with the upper half of Appalachia being in the “Mid-Atlantic” and “NYC” regions.  

Figure 1

Overall View of North American Dialects (Labov, 2006, p. 148) 

Moreover, Southwest Virginia (SWVA) is an area in Appalachia that is particularly underexplored within the literature, with several studies conducted in surrounding areas, yet none attempted in SWVA (Greene, 2010; Labov et al., 2006; Reed, 2018; Webelhuth & Dannenberg, 2006; Wolfram & Christian, 1976). Additionally, SWVA represents an understudied region within Appalachia that is geographically isolated and has mid-to-low socioeconomic status (SES), compared to other areas of Appalachia, which may influence how /ay/ monophthongization patterns, making it an ideal area to study. 

1.1.2 Social Factors

When compared to other dialectal varieties, AE exhibits a relatively limited amount of literature exploring its linguistic diversity. While some social factors have been explored within dialectal features of AE, gender is rarely mentioned in comparison to other factors like age and rootedness (Greene, 2010; Reed, 2014, 2018). However, gender has been explored in other varieties of English including MUSE, SAE, African American English (AAE), and others (Becknal, 2012; Thomas, 2011). Linguistically, gender groups display varying patterns from one another and can play a significant role in shaping the usage rates of particular linguistic features (Becknal, 2012; Seaver et al., 1991; Thomas, 2011). Seaver et al. (1991) found differences between males and females within the same dialect of New Zealand English. Labov (2001) identifies that gender has a key role in linguistic change and variation, describing the general linguistic conformity of women as having higher rates of standard linguistic variants compared to men (p. 266). Concerning AE, gender can play a significant role in predicting the use of features within the dialect.  

1.1.3 Phonetic Contexts

One stereotypical feature shared by SAE and Appalachian English (AE) is /ay/ monophthongization (Feagin, 2000; Greene, 2010; Kurath & McDavid, 1961; Labov et al., 2006; Reed, 2018; among others). Feagin (2000) discusses this salient feature, writing, “The most notable unchanging element in Southern states pronunciation concerns the ‘long I’” (p. 342). Monophthongization occurs when a diphthong simplifies into a monophthong through the ungliding of the vowel (Hazen, 2004). For /ay/ monophthongization, a word like “ride” would sound more like “rahd.” Additionally, this feature is synonymous with various terms. Wolfram & Christian (1976) write “glide reduction” when referring to monophthongization while Hazen (2004) refers to monophthongization as “ungliding.” It is important to note that a monophthong can be defined as a single, unchanging vowel (e.g., the /I/ in “ship” and /ʊ/ in “put”), while a diphthong can be defined as a vowel that begins as one sound and glides into another (e.g., the /ɔɪ/ in “coin” and /ay/ in “light”). Figure 2 demonstrates where different vowel sounds are made in the mouth with plots on the diagram. The higher the vowels are plotted, the higher the sound is made in the mouth, with the tongue near the roof of the mouth (e.g., the /I/ in “ship”). The red line shows how the diphthong /ay/ is made with the tongue starting low and moving high during the word (e.g., “ride”), while the blue line shows the monophthong and how it stays low (e.g., “rahd”).  

Figure 2

Vowel Diagram Plots Showing /ay/ Diphthong and Monophthong  

Note. The red line shows the diphthong, and the blue line shows the diphthong becoming monophthongized.

Monophthongization can occur before voiced and voiceless consonants, usually referred to as pre-voiceless or pre-voiced monophthongization (Greene, 2010; Kurath & McDavid, 1961; Labov et al., 2006; Reed, 2016; Wolfram & Christian, 1976; among others). A voiced consonant is produced by vibrating the vocal cords (e.g., “b,” “v,” “d,” “m”), while a voiceless consonant is produced without the vibration (e.g., “p,” “k,” “t”). Additionally, monophthongization can occur in open syllable contexts (e.g., “I,” “my,” “die”), meaning the word ends with a vowel sound. Within SAE, /ay/ monophthongization only occurs in pre-voiced (“ride”) and open-syllable (“rye”) contexts, while in AE it can occur across all contexts, including pre-voiceless environments like “right” (Reed, 2016; Thomas, 2003; Wolfram & Christian, 1976).  

2. Methods

2.1 Research Design

Data utilized for this study was taken from a private corpus containing interviews with 84 speakers. Each interview was collected in SWVA, either on Virginia Tech’s (VT) campus or in the field (specifically in the towns of Abingdon, Lebanon, Big Stone Gap, and Grundy) and transcribed using ELAN transcription software. Recordings were transcribed orthographically by research assistants from SWVA who had extensive exposure to AE. Some were also speakers of AE. The corpus was created in May of 2022, and all interviewees were from SWVA or had spent a significant amount of time in the area. Speaker names were not collected in the original interviews; therefore, they are represented by a number that begins with “#7” and is followed in the order they were recorded (e.g., 709, 710, etc.). During the interview, participants were asked to take part in a task separating town names into groups and then asked follow-up questions that facilitated dialogue (e.g., do you think SWVA is distinct from the rest of the South?). The corpus consists of a wide range of demographics such as age, gender, and geographic location. However, ethnicity lacked diversity with most speakers being white (<75 speakers). All demographics were self-reported by participants. Additionally, education level, SES, and occupation were not gathered. The corpus is securely stored in a research lab on the VT campus and is inaccessible to others.  

My study consisted of 14 speakers (7M; 7F) of AE from SWVA and generated a mixed effects logistic regression model of monophthongization. I controlled for age and only used the interviews of speakers over the age of 50 due to previous research showing /ay/ monophthongization is prevalent in older adults (Reed, 2014). I also controlled for ethnicity, taking only white speakers, due to low non-white speaker numbers. Additionally, I only collected speakers who were born and currently live in SWVA.   

I decided to aurally code my data because I am a speaker of AE with substantial exposure to /ay/ monophthongization. This approach allowed me to utilize my familiarity with regional speech patterns to accurately identify, analyze, and interpret linguistic features (in this case monophthongization) in the data. For each speaker, I took all tokens of the diphthong, whether it was monophthongized or not, and the phonetic context in which it was used. The dataset consisted of a total of 248 tokens. Each measured token had three variables describing the word’s location including if the word was monophthongized, speaker gender, and phonetic contexts. The dependent variable was /ay/ monophthongization, with the independent variables being speaker gender and phonetic contexts. I used Microsoft Excel to create a code book and make lines of tokens with each containing the token number, speaker, the word used, if it was monophthongized, following context, gender, speaker origin, where the data was collected (in the field or on campus), the sentence said, and the time stamp. Some of the variables in the code book (e.g., town, data collection, word) are categorical variables ranging from a binary choice to countless options and are recorded due to the fact there is no way to control them. Table 1 shows an example of what a row of data for each context would look like.

Table 1

Code Book Example in Microsoft Excel  

token	speaker	word	ay mono	voice	gender	data collection	town	sentence	time stamp
1	710	rye	1	0	F	FLD	Galax	My dad loved rye bread.	04:02-04:05
2	710	ride	0	V	F	FLD	Galax	I need a ride.	07:16-07:19
3	710	right	1	VL	F	FLD	Galax	I am right.	11:13-11:15

A number was used when determining if the vowel was a diphthong (“ride”) or monophthong (“rahd”). If the speaker used /ay/ monophthongization, it was recorded under the “ay mono” category as a “1.” If the vowel was a diphthong, it was recorded as a “0.” Phonetic contexts were recorded under the “voice” category as “0” (open), “V” (voiced), and “VL” (voiceless). Data collected in the field was recorded as “FLD,” and data collected on campus was recorded as “VT.” All open syllable words were coded as syllable-final whether they were one syllable (e.g., “die”) or two syllables (e.g., “diverse”). I controlled for word frequency, where the same word could be collected up to three times. I did this to avoid a common word (e.g., “my,” “like,” “I”) taking up the entire data set. This also included each lemma, a morphological term used to describe the canonical form of a word, as the same word. For example, words like “I’ll,” “I’m,” and “I’ve,” were all grouped as one word (“I”), with only three instances taken. I also excluded some phonetic contexts including words where the vowel /ay/ is followed by an “r” or glides (e.g., “tire,” “higher,” “trying,” “wire”) due to the fact the diphthongal vowel has a different syllable structure than the monophthongal vowel, which would make syllabification (i.e., the process of dividing a word into its individual syllables) challenging to differentiate in my code book. 

3. Results

3.1 Overview of Data

The frequency of /ay/ monophthongization can be seen in Table 2, with notable variation between speakers ranging from 100% monophthongized to 27.3% monophthongized across all tokens.  

Table 2

Frequency of /ay/ Monophthongization by Speaker, Gender, and Voicing Context   

Speaker	Gender	Number of /ay/ mono / number of tokens				Rate of /ay/ monophthong
		Voiced	Voiceless	Open syllabus	Total
710	F	4/14	1/2	1/6	6/22	27.3%
720	M	12/17	12/12	12/14	36/43	83.7%
721	F	6/7	3/4	8/8	17/19	89.5%
722	F	5/5	1/7	2/4	8/16	50%
726	M	1/2	0/2	1/2	2/6	33.3%
727	M	7/7	5/5	4/7	16/19	84.2%
728	F	2/4	5/5	6/7	13/16	81.3%
730	M	4/4	1/2	2/2	7/8	87.5%
739	M	6/6	5/5	2/2	13/13	100%
752	M	1/1	2/2	3/3	6/6	100%
757	F	3/8	10/10	10/12	23/30	76.7%
764	M	6/9	2/7	1/4	9/20	45%
777	F	7/8	3/5	5/6	15/19	78.9%
784	F	1/1	0/4	5/6	6/11	54.5%

Note. The rate of /ay/ monophthongization is organized by speaker.

3.2 Logistic Regression Model

The data was analyzed via a logistic regression model, as it offers a binary outcome (e.g., was /ay/ monophthongization used?) and allows for a high degree of interpretability. A logistic regression model is a statistical method where the objective is to predict the likelihood of an event occurring based on the relationship between the predictor variables and the outcome variable. In the case of /ay/ monophthongization, this model allows the predictor variables (speaker gender and phonetic contexts) to impact the odds of monophthongization occurring. Each variable was compared against a baseline, which was an open syllable and the speaker gender being female. Table 3 shows the p-values table for each variable. A p-value <0.05 indicates that a factor is a significant predictor of variation. Predictors tested include the voicing of the following segment, the gender of the speaker, and an interaction between the voicing of the following segment and the gender of the speaker. 

Table 3

P-values Table

Variable	P-value
Voiced (Voiced)	0.2610
Voiced (Voiceless)	0.1406
Gender (Male)	0.9380
Male * Voiced	0.1357
Male * Voiceless	0.1569

Note. A high p-value can be observed in the male gender variable. 

In Table 3, we see that none of the variables or interactions rose to the level of significance. This means that none of the variables tested significantly predict the data observed. One possible explanation may be the individual-level variation within the sample, which was accounted for in the random effect of the speaker. That said, there are certain trends that will be described below.  

3.3 Gender Trends

Female speakers contributed a total of 133 tokens, with 88 containing the /ay/ monophthong. Contrastingly, male speakers contributed 118 tokens, with 89 containing the /ay/ monophthong. Although male speakers produced fewer tokens in total, they contributed a higher percentage of the /ay/ monophthong (Table 4).  

Table 4

Frequency of /ay/ Monophthongization by Gender 

Gender	Total number of tokens	Total number of /ay/ monophthong	Percentage of /ay/ monophthong
F	133	88	66.2%
M	115	89	77.4%
Total	248	177	71.4%

When compared side-by-side, this trend can be clearly seen. Figure 3 shows a visual representation of men demonstrating a higher use of /ay/ monophthongization when compared to women.

Figure 3

Frequency of /ay/ Monophthongization Across Gender

When compared, men monophthongized 11% more than women total, with a rate of 77.4% compared to women at 66.2%.  

3.4 Phonetic Contexts Trends

Overall, there was a high percentage (71.4%) of /ay/ monophthongization used by all speakers. Out of 248 tokens, /ay/ monophthong occurred in 177 tokens. Open syllables exhibited the highest percentage of /ay/ monophthong usage compared to voiced and voiceless contexts. Table 5 shows the percentage of /ay/ monophthongization used in each context (open, voiced, and voiceless) and the overall percentage used. 

Table 5

Frequency of /ay/ monophthongization by voicing context 

Voicing context	Total number of tokens	Total number of /ay/ monophthong	Percentage of /ay/ monophthong
Open syllable	83	62	74.7%
Voiced	93	65	69.9%
Voiceless	72	50	69.4%
Total	248	177	71.4%

When comparing voiced and voiceless contexts, /ay/ monophthongization was done at roughly equal rates and around 5% lower rate than open syllables. Figure 4 shows a visual representation of /ay/ monophthongization by category.

Figure 4

Frequency of /ay/ monophthongization across voiced, voiceless, and open syllable contexts  

These results indicate that the open syllable context showed the highest proportion of /ay/ monophthong usage, followed by the voiced context, and then the voiceless context. It should be noted that the rates of /ay/ monophthongization in voiced and voiceless contexts were separated by less than one percent, indicating an almost equal usage. 

3.5 Interaction Effect of Gender and Phonetic Context 

When analyzing each variable independently, distinct trends emerge for both speaker gender and phonetic contexts; however, when examining the interaction between both independent variables, it is not as straightforward. Figure 5 illustrates the interaction effects between both variables. 

Figure 5

Interaction Effect of Gender and Voicing Context on Frequency of /ay/ Monophthongization Usage

Although males demonstrated higher rates of /ay/ monophthongization than women overall, when phonetic contexts are separated women demonstrate higher rates in open syllable contexts than males. A total increase of around 25% occurs in female speakers when comparing rates of /ay/ monophthongization in open syllables versus voiced and voiceless contexts.

4. Discussion

Drawing on the findings presented in the previous section, this research has yielded both anticipated and unforeseen outcomes, shedding light on the use of /ay/ monophthongization in SWVA. This section dives into the interpretation of results, discussing the outcomes and how they compare to previous work done in surrounding areas. This section will be divided into three subsections: influence of phonetic contexts, influence of speaker gender, and impact of intersection effect on gender and phonetic context. 

4.1 Influence of Phonetic Context

Within my dataset, we observe a high usage of /ay/ monophthongization across speakers, which aligns with previous work done in surrounding areas (Greene, 2010; Pederson, 1983). Overall, the speakers showed somewhat typical AE patterning of monophthongization in both voiced and voiceless contexts. The high rates of monophthongization in these contexts appear to support prior work that shows that /ay/ monophthongization can occur across all contexts in AE, contrasting with SAE occurring in limited contexts (Reed, 2016; Thomas, 2003; Wolfram & Christian, 1976). When compared to research done in surrounding areas, rates of monophthongization in a voiceless context (69%) were slightly higher when compared to Pederson (1983), who found a 65% usage rate of /ay/ monophthongization in the voiceless context in East Tennesseans. However, rates are lower when compared to Greene (2010), who found an 89% usage rate of /ay/ monophthongization in the voiceless context in Eastern Kentuckians (p. 59).  

Additionally, results show that monophthongization occurs more frequently in open syllables compared to other contexts at a rate of 74.7%, which is supported by Wolfram & Christian (1976), who describe monophthongization as occurring most likely “…at the end of a word as in pie, sky, or tie” (p.71). Pederson (1983) also found an almost identical rate (75%) of /ay/ monophthongization in open syllable contexts in East Tennesseans, with this context having the highest usage rate out of open, voiced, and voiceless. However, this claim is not supported by other research (Anderson, 2003; Greene, 2010; Reed, 2016; Thomas, 2003; Thomas, 2011) that found similar rates for voiced and open syllable contexts, allowing for these contexts to be grouped. The findings outlined in this study indicate that combining voiced and open syllable contexts in SWVA should be avoided in these communities. Doing so may obscure significant distinctions between these linguistic environments.  

4.2 Influence of Gender Speaker 

Overall, results indicate males and females exhibit differing rates of /ay/ monophthongization usage, with males having overall higher rates of the feature. This supports previous work that found gender groups patterning differently when using linguistic features within the same dialect (Becknal, 2012; Seaver et al, 1991; Thomas, 2011). Results in female usage of /ay/ monophthongization are 15% lower than males in both voiced and voiceless contexts, with usage being 59.6% in voiced context and 62.7% in voiceless. While males had slightly higher rates of voiced (80.4%) contexts than voiceless (77.1%), the opposite is observed in female speakers.  

4.3 Impact of Intersection Effect of Gender and Phonetic Context 

As stated in the results section, analyzing each variable independently shows distinct trends for both speaker gender (males having higher usage) and phonetic contexts (open syllables having higher usage). However, upon closer examination of the interaction between both independent variables, results reveal an unexpected atypical pattern. When separated by phonetic contexts, open syllables had the highest overall usage, but when separated by gender, open syllable context had the lowest usage rate by male speakers overall. Although female speakers had lower rates of /ay/ monophthongization than their male counterparts, they exhibited slightly higher rates of monophthongization in open contexts than male speakers. This result is unusual because previous research indicates that open syllable contexts and voiced contexts pattern at similar rates when examining different social factors besides gender (Anderson, 2003; Greene, 2010; Reed, 2016; Thomas, 2003).

Several potential explanations could provide insight into the reasons that we see this uncommon pattern of women demonstrating higher rates of monophthongization in open syllable contexts than male speakers, as well as in both voiced and voiceless contexts. Perhaps it is due to a lot of open syllable contexts being pronouns (e.g., “I,” “my”) and occurring more naturally and frequently in conversation. Most participants had at least six tokens of open syllable contexts each because they had at least three or more instances of “I” and “my” compared to other words with the /ay/ diphthong that occur less often in other contexts (e.g., “ride,” “bike”).  However, only 36% of total tokens taken from female speakers occurred in open syllable contexts, and for both genders combined, only 33% of all tokens occurred in open syllable contexts. 

Another explanation for this atypical pattern could be due to the unconscious acceptability of open syllable contexts compared to other phonetic contexts. As stated above, when describing the role gender plays in predicting monophthongization, women often tend to have lower rates of the dialectal linguistic feature compared to men (Seaver et al., 1991; Thomas, 2011). Labov (2001) describes linguistic conformity of women as having higher rates of standard linguistic variants, stating, “…women show a lower rate of stigmatized variants and a higher rate of prestige variants than men” (p. 266). It is important to note that research on /ay/ monophthongization occurring in open syllable contexts can be found at high rates in speakers born before the 1920s compared to rates of speakers born after that period (Bowie, 2001). It very well could be the possibility that /ay/ monophthongization that occurs in open syllable contexts is less stigmatized than in voiced and voiceless contexts, making it more acceptable to use. This acceptability could explain why women have a higher usage rate in this context.

Additionally, it is possible that as women age, they become less standard in and increase their usage of their dialectal features, beginning with open syllables. Paunonen (1994) found that women become less normative in the use of a linguistic feature from middle age to old age. He discusses the change and attributes it to a higher position in society and presumably more freedom (Paunonen, 1994). Labov (2001) somewhat supports these claims by addressing socioeconomic and social class factors, with the interpretation that the jobs/lifestyles of women might not automatically change linguistic behavior, but it is possible. If SES, education level, and gender were to be recorded in future research, the linguistic patterning could support this possibility. This would combine the above explanation of unconscious acceptability and women’s perception of holding a more comfortable place in society; therefore, explaining the high usage of /ay/ monophthongization in open syllable contexts. 

All hypothetical explanations stated above could be possible; however, due to the low token count, it is hard to determine which is more likely. As a speaker of AE from SWVA, I could allude to the fact that the unconscious acceptability of open syllables seems to align more with native speaker intuition. When talking to other AE speakers from home, many suggest that words like “I” and “my” are easier to get away with saying in an accent than words like “ride” and “like.” These assumptions are backed by claims that open syllables can “hide between words easier” and can be “said faster” than that of their counterparts.

5. Conclusion

Results from this study show SWVA patterns somewhat differently, but not entirely, from previous work in surrounding areas (Greene, 2010; Labov et al. 2006; Reed, 2018; Wolfram & Christian, 1976). The study’s scope may be constrained by the low token count, potentially limiting its depth of analysis. This warrants further research in SWVA to truly understand why it is turning away from prior work. Expanding upon this research could involve implementing an elicitation task, where word frequency is controlled, potentially leading to clearer and more precise results. One approach could involve creating a word list and sentences for participants to read, ensuring an identical token count across speakers for consistency. Additionally, further research could investigate other social factors that might exhibit different patterns, such as SES, education, the interplay between age and gender, and so forth. Overall, SWVA is a ripe region for further investigation of /ay/ monophthongization.  

References

About the Appalachian region. (2024a). Retrieved from https://www.arc.gov  

Anderson, B. L. (2003). An acoustic study of Southeastern Michigan Appalachian and  African-American Southern migrant vowel systems. University of Michigan. 

Becknal, R. S. (2012). Normative nasalance patterns in male and female speakers of Southern  American English native to Texas (Doctoral dissertation, Texas Christian University). 

Bowie, D. (2001). The Diphthongization of/ay/ Abandoning a Southern Norm in Southern Maryland. Journal of English linguistics, 29(4), 329-345. 

Feagin, C. (2000). Sound change in the South. American Speech, 75(4), 342-344. 

Greene, R. D. (2010). Language, ideology and identity in rural eastern Kentucky. Stanford  University. 

Hazen, K. (2004). Some cases for the syllable in Southern English. Southern Journal of  Linguistics, 28(1), 64-80. 

Kurath, H., & McDavid, R. I. (1961). The pronunciation of English in the Atlantic States: Based  upon the collections of the linguistic atlas of the Eastern United States. (No Title).  

Labov, W. (2001). The Gender Paradox. Principles of linguistic change: Social Factors (pp.  261-293) Blackwell.  

Labov, W., Ash, S., & Boberg, C. (2006). The Dialects of North American English. The atlas of  North American English: Phonetics, phonology and sound change (p. 119-151) 

McConnell-Ginet, S., & Eckert, P. (2003). Language and gender (pp. 75-99). Cambridge:  Cambridge University Press.  

Moreton, E., & Thomas, E. R. (2007). Origins of Canadian Raising in voiceless-coda effects: a  case study in phonologization. Laboratory phonology, 9, 37-64. 

Paunonen, H. (1994). Language change in apparent time and in real time.  

Plichta, B., & Preston, D. R. (2005). The /ay/s have it: The perception of /ay/ as a north-south  stereotype in United States English. Acta Linguistica Hafniensia, 37(1), 107-130. 

Pederson, L. (1983). East Tennessee folk speech: a synopsis. Lang. 

Reed, P. E. (2014). Inter-and intra-generational /aɪ/ monophthongization, indexicality, and  Southern Appalachian identity. Southern Journal of Linguistics, 38(1), 159-194. 

Reed, P. E. (2016). Sounding Appalachian: /ai/ monophthongization, rising pitch accents, and  rootedness (Doctoral dissertation, University of South Carolina). 

Reed, P. E. (2018). The importance of Appalachian identity: A case study in rootedness.  American Speech: A Quarterly of Linguistic Usage, 93(3-4), 409-424. 

Renwick, M.E.L. (2020). Linguistic features of Southern American English. https://mrenwick.franklinresearch.uga.edu/southern_speech/linguistic_features

Seaver, E. J., Dalston, R. M., Leeper, H. A., & Adams, L. E. (1991). A study of nasometric  values for normal nasal resonance. Journal of Speech, Language, and Hearing Research, 34(4), 715-721. 

Thomas, E. R. (2001). An acoustic analysis of vowel variation in New World English. Published  by Duke University Press for the American Dialect Society 

Thomas, E. R. (2003). Secrets revealed by Southern vowel shifting. American Speech, 78(2),  150-170. 

Thomas, J. (2011, June). Gender and/ai/monophthongization in African American English. In  Annual Meeting of the Berkeley Linguistics Society (pp. 449-463). 

Thomason, S. (2008). Social and linguistic factors as predictors of contact-induced change.  Journal of language contact, 2(1), 42-56. 

Webelhuth, G., & Dannenberg, C. J. (2006). Southern American English personal datives: The  theoretical significance of dialectal variation. American speech, 81(1), 31-55. 

Wolfram, W., & Christian, D. (1976). Appalachian speech.