DIATOPIC VARIATION AND THE CHANGE PROCESS IN THE SOCIAL VALORIZATION OF PROGRESSIVE PALATALIZATION IN ALAGOAS

ABSTRACT

We investigate the progressive palatalization of alveolar stop consonants preceded by /j/, as mui[ʧ]o (‘very’/‘much’) and doi[ʤ]o (‘crazy’) in Alagoas, based on theoretical-methodological assumptions of variational sociolinguistics. We aim to analyze the diatopic distribution of palatalization and the social and linguistic pressures in the process. We analyzed 4,046 occurrences in interviews of 168 speakers from seven cities in Alagoas, using multilevel logistic regression. We conclude that the process is diatopically conditioned, extending from the west to the east and northeast of the state. It is concluded that the process is inversely proportional to schooling and that the interference of schooling increases as the speaker’s age decreases, revealing a process of change in the social value of palatalization. There is evidence that, although it is significantly more productive in Alagoas, the palatalization should not expand over time due to social pressures. Regarding linguistic variables, the process is favored within the word, although it also occurs at the border of lexical items. The following vowel that favors the process is the vowel /i/, which reveals the interference of the double presence of the [+ coronal] feature in adjacent environments. The absence of voicing in the target consonant and syllable atonicity also favor palatalization.

Variationist Sociolinguistics; Phonological-phonetic variation in Alagoas; Progressive palatalization; Alagoas

RESUMO

Investiga-se a palatalização progressiva das oclusivas alveolares precedidas de /j/, como mui[ʧ]o e doi[ʤ]o em Alagoas, a partir de pressupostos teórico-metodológicos da sociolinguística variacionista. Objetiva-se analisar a distribuição diatópica da palatalização e as pressões sociais e linguísticas no processo. Analisam-se 4.046 ocorrências em entrevistas de 168 falantes de 7 cidades alagoanas, utilizando-se regressão logística multinível. Conclui-se que o processo é diatopicamente condicionado, ampliando-se do oeste para o leste e nordeste do estado. Conclui-se que o processo é inversamente proporcional à escolaridade e que a interferência da escolarização cresce na medida em que diminui a idade do falante, revelando um processo de mudança na valorização social da palatalização. Há indícios de que, apesar de ser significativamente mais produtiva em Alagoas, a palatalização não deve expandir-se ao longo do tempo, devido a pressões sociais. Em relação às variáveis linguísticas, o processo é favorecido no interior da palavra, apesar de também ocorrer na fronteira de itens lexicais. A vogal seguinte favorecedora é a vogal /i/, o que revela a interferência da dupla presença do traço [+coronal] nos ambientes adjacentes. A ausência do vozeamento na consoante alvo e a atonicidade da sílaba também favorecem a palatalização.

Sociolinguística Variacionista; Variação fonético-fonológica em Alagoas; Palatalização progressiva

Introduction

In this work, we investigated, from the perspective of variationist sociolinguistics, the progressive palatalization of alveolar stop consonants preceded by the semivowel /j/ in Alagoas. This type of palatalization finds significant productivity in the Northeast of Brazil, with indications of its practice in almost all the regional states. It is found in the whole Eastern extension from Bahia to Rio Grande do Norte ( SANTOS, 1996 ; MOTA; ROLEMBERG, 1997 ; HENRIQUE; HORA, 2012 ; SOUZA NETO, 2014; OLIVEIRA, A. A., 2017).

The process takes place in examples such as:

1. “Teve [ˈmũɪ̯̃tʃɐ] violência” (muita) - AR18M06, (‘There was much violence’)¹

2. “Pelo [kuɪ̯ˈdʒadʊ]” (cuidado) - SM64M02, (‘For the care’)

3. “Meu pai [tʃãˈbẽɪ̯̃]” (também) - PE51F11, (‘My father also’)

4. “Trabalhei [ˈdʒuɐs] vezes” (duas) - DE51M05, (‘I worked twice’)

The focus of this work is to investigate the diatopic distribution of palatalization in the speech in Alagoas, and the social and linguistic pressures in the process.

This type of palatalization has been more productive in the periphery of large cities (SOUZA NETO, 2014; MOTA; ROLEMBERG, 1997 ; OLIVEIRA; OLIVEIRA; PAULA, 2018), with a negative social value, being avoided by female, young, and better-educated informants.

The highest productivity index of this type of palatalization was observed in the city of Maceió, with values between 20% and 25% (OLIVEIRA, A. A., 2017; SANTOS, 1996 ). It raises the assumption that geographic aspects are relevant to the process, and that, in Alagoas, palatalized variants are more frequent.

Some questions guide this study:

1. What is the interference of geographical aspects in palatalization in Alagoas?

2. Is the process sensitive to increased schooling?

3. Is the speech process socially stigmatized in Alagoas?

4. Is there a process of linguistic change in progress?

5. How do internal language factors play a role in the palatalization process?

Methodology

In this study, we adopted the theoretical-methodological proposal of variationist sociolinguistics, presented primarily in Labov (2008) , for the analysis of variation and linguistic change. It predicts the identification of a variable process in a speech community, the selection of informants, the collection and analysis of interviews, and the quantitative analysis of the variation, in the search for factors that interfere in the variation process.

The analyzed data belong to the project PORTAL – ‘Linguistic variation of Portuguese in Alagoas’²(OLIVEIRA, A. J., 2017). We analyzed data from seven cities in Alagoas (Maceió, Arapiraca, Delmiro Gouveia, Santana do Ipanema, Palmeira dos Índios, União dos Palmares, and São Miguel dos Campos); each belonging to a state micro-region (there are 13 micro-regions in Alagoas).

The research sample consisted of 168 participants, 24 per surveyed city. The inclusion criteria were: (1) having been born in the municipality, (2) not having been absent from the municipality for more than ten years, and (3) also having both parents being born in the municipality (preferably). The sampling was non-probabilistic, using the technique called “snowball,” in which the participants were selected by indication of friends or acquaintances. The study approached people who were not known to the interviewer, but were known to someone the interviewer knew. The sample by city was composed of quotas, considering the following social variables: gender (male and female); age group (between 18 and 30 years, between 40 and 55 years, and over 65 years); and education (less than nine years, and more than 11 years).

The research data were collected through “life history” interviews; defined by Chizzotti (2011) as “a retrospective account of an individual’s personal experience, oral or written, regarding facts and events that were significant and constitutive of their lived experience³” ( CHIZZOTTI, 2011 , p. 101, our translation). In addition to this type of report, we also sought the participants’ opinions regarding controversial topics. The intention was to allow interviews that contain different types of narration; ranging from narrative (e.g., “Tell me about an important event from your childhood at school”); to descriptive (e.g., “How was the house where you lived when you were 10 years old?”); and argumentative (e.g., “What do you think about same-sex marriage?”). A pre-defined question script, memorized by the interviewer, was used. Participants were informed that the conversation would cover childhood memories, recent memories, and opinions on controversial topics. To seek a reduction in stylistic monitoring, we informed the interviewee that the questions’ objectives were to verify the participant’s ability to remember facts, from the ancient and recent past, in as much detail as possible, and to know his opinion on certain controversial topics.

A TASCAM voice recorder, model DR-100, was used for recording. The recordings were made in .wav format, with a sampling rate of 24 bits, and a resolution of 48 kHz. An Arcano brand, model WZ-1000, unidirectional cardioid condenser headset microphone was also used. Recordings lasted between 9 and 11 minutes.

Interviews were transcribed according to the standard spelling, using PRAAT software, which enabled the synchronization between audio and transcription. PRAAT intervals were created according to the pauses (silence greater than, or equal to, 200ms). Three tiers were created: for the interviewer’s speech; for the participant’s speech; and for “others”, which registered third party speeches, or other sounds. Collaborators were coded as follows: participant AR25F10 was a resident of the city of Arapiraca (AR), 25 years old, female (F), and completed ten years of schooling.

Automatic selection of data occurrences was achieved using the text editor resources (search and highlighting sequences st, sd, it, id ). All occurrences were analyzed acoustically (with spectrograms and oscillograms analysis) to classify the variants more objectively.

The dependent variable of the study is the alternation between the alveolar stop consonants /t/ and /d/, and the affricated [tʃ] and [dʒ], through the process of progressive palatalization in a context preceded by the glide /j/, e.g. muito (‘very’/‘much’) and doido (‘crazy’).

Investigated social variables were: gender (male or female), age and education (analyzed as continuous variables), and geographical location. Interviewees came from cities in different regions of the state: Arapiraca, Delmiro Gouveia, Maceió, Palmeira dos Índios, São Miguel dos Milagres, Santana do Ipanema, and União dos Palmares.

The investigated linguistic variables were: the following vowel, which refers to the vowel after the stop consonant (muit a (‘many’/‘much’), oit e nta (‘eighty’), noit e (‘night’), Feit o sa (‘Feitosa’), oit o (‘eight’)); the type of stop consonant, related to the sonorous nature of the consonant (cui d ar (‘to take care’) and mui t o (‘very’/‘much’)); the position in the lexical item, which deals with the presence of stop consonants in word-medial position, e.g. mu it o (‘very’/‘much’) and do id o (‘crazy’), or in a borderline position, as in meu pa i t ambém (‘my father also’) and trabalhe i d uas vezes (‘I worked twice’); and tonicity, which refers to the presence or absence of accent on the syllable containing the target consonant: stressed, e.g. ajeitar (‘to fix’), and unstressed, as in oitocentos (‘eight hundred’), and muito (‘very’/‘much’). The lexical item and the individual were also analyzed as variables of a more aggregated level.

For the quantitative analysis, we used inferential methods of statistical analysis (contingency tables, univariate and multivariate tests, and multilevel regression methods). The estimate of the effects associated with the independent variables was made using multilevel logistic regression models, a multivariate model that controls the effects of more aggregated variables. The data analyzed in this work has a hierarchical structure since the observations can be grouped according to the individuals who produced them, and the lexical items. The estimate of how much of the observed variability can be explained by the more aggregated levels (individual and lexical item), was obtained by the intraclass correlation coefficient (ICC).

Two statistical tests were used: the maximum likelihood-ratio test (MLRT) and the Wald test (WT). The MLRT analyzes the statistical significance between independent variables, allowing the identification of statistically significant independent variables, and ranking these variables. The WT analyzes the statistical significance between factors within the independent variables, allowing the identification of factors that present effects that are statistically different from the average of the factors effects in an independent variable (factor weight = 0.50).

The null hypothesis of the MLRT is that the effect of an independent variable in a regression model is equal to 0. The alternative hypothesis is that the effect of such a variable is different from 0. The test significance measures the probability of making an error by denying the null hypothesis, the null hypothesis being true. The lower the significance in the MLRT, the greater the explanatory power of the independent variable, over the dependent variable. In Varbrul and GoldVarb (softwares that are traditionally used in the analysis of linguistic variation), the MLRT is used in step-up and step-down routines. In this work, we chose the selection and hierarchy of the statistically significant variables using a method similar to the step-down. All independent variables were included in the model. The variables were removed one by one, considering the greater significance in the MLRT. The final model contains only variables with significance <0.05. The ranking of the statistically significant variables was made by the statistical significance of each variable included in the final model. The MLRT was also used to test the interaction between social variables.

The null hypothesis of the Wald test is that the effect of a factor, on an independent variable, is equal to the average of the effects of the factors. The alternative hypothesis is that the effect of such a factor is different from the average of the effects of the factors. Likewise, the significance of the WT measures the probability of making an error by denying the null hypothesis, the null hypothesis being true. The smaller the significance in the WT, the greater the difference between the effect of a factor and the average of the effects of the factors. In Varbrul and GoldVarb, the average of the effects of the factors is given by what is traditionally called the neutral effect (factor weight equal to 0.50). The WT allows verifying if the effect of a factor is statistically different from the neutral effect. Such a test is quite useful for factor weights close to 0.50.

In this work, statistical analysis was performed with the aid of the R software, using the ‘gmodel’ packages (to generate contingency tables), ‘lme4’ (for multilevel logistic regression, MLRT, and WT), visreg (for interaction graphics), and DAAG, for the multicollinearity test.

Results and discussion

A total of 4,046 occurrences of contexts conducive to palatalization were identified, of which 844 (20.9%) were palatalized. Mota and Rolemberg (1997) observed 2.4% of this type of palatalization in Salvador-BA; Souza Neto, 6% in Aracaju; Henrique and Hora (2012) , 10.5% in João Pessoa-PB; and Oliveira, A. A. (2017), 20.4% in Maceió, which allows us to state that the progressive palatalization of alveolar stop consonants is more frequent in Alagoas than in other states surveyed in the Northeast.

For our analysis, the next step was to adjust the regression model, which allowed us to identify the statistically significant variables and the factors that affect regressive palatalization. The model was adjusted using the maximum likelihood ratio test, testing, in a multivariate model of multilevel logistic regression (considering the individual and the lexical item as the variables of more aggregated level), the linguistic independent variables tonicity, position in the lexical item, type of consonant, and following vowel; and the social variables sex/gender, age, education, and city, as well as the possible interactions between variables. The final model was composed by the independent variables tonicity, position in the lexical item, type of consonant, following vowel, city, and by the interaction between age and education, and by the variables of individual aggregate level and lexical item. Solely the sex/gender variable was excluded from the model. The following tables demonstrate the results for the independent variables after adjusting the best model, presented in order of significance in the MLRT.

Table 2 demonstrates that the variable “sex/gender” was not statistically significant and, therefore, palatalization is not significantly different between male and female genders. This result contradicts other research conducted on the progressive palatalization of alveolar stop consonants, which have revealed a greater use of palatalized forms by male members ( HENRIQUE; HORA, 2012 ; SOUZA NETO, 2014; OLIVEIRA, A.A., 2017; OLIVEIRA; OLIVEIRA; PAULA, 2018). We also did not identify any interaction between social variables and cities. This indicates that there is no significant difference in the effects of social variables and their interactions between the different Alagoas cities surveyed in this study.

Table 2
– Independent variable excluded from the final model (without statistical significance)

	Total	% palatalization	Factor Weight	Sig. Wald	Sig. MLRT
Sex/gender					0.745
Female	2126	22.8	*	*
Male	1920	18.8	*	*
Total	4046	20.9

Source: Authors’ elaboration.

The variable with the greatest significance was the “individual” variable ( table 3 ). This indicates that a substantial percentage of the variability (27.7%) can be explained by the variation between individuals, regardless of the city, sex/gender, age, and schooling, all controlled in the model. We can affirm that aspects not controlled in this study, related to individuals, interfere in the variability. However, even without being aware of such aspects, the control of the individual as an aggregate level allows the result of the investigated social variables to be more reliable and better reflect the observed reality.

Table 3
– Aggregate random effect variables included in the final model

	n	Variance	ICC	Sig. MLRT
Individual	168	1.2588	27.7%	2.2e-16
Lexical item	259	0.1618	6.2%	6.0e-05

Source: Authors’ elaboration.

The result for the variable “lexical item” ( table 3 ) allows us to state that a reduced part of the variability, at the linguistic level (6.2%), can be attributed to the variation between the lexical items, and, therefore, our investigated linguistic variables (tonicity, position in the lexical item, following vowel, and type of consonant) explain the process from the standpoint of their linguistic conditioners. Subsequently, we will analyze the linguistic variables, the interaction between age and schooling, and the city variable.

When analyzing the results of the city variable in table 1 , it is possible to observe, based on the significance of the Wald test, three groups of cities that present different behaviors in palatalization. São Miguel dos Milagres and União dos Palmares favor the process (RW=0.68 and 0.62, respectively); Maceió, Santana do Ipanema, Arapiraca, and Palmeira dos Índios have no statistically different effect from the average effect (RW=0.54, 0.53, 0.52, and 0.51, respectively); Delmiro Gouveia favors palatalization (RW=0.16). Let us analyze the following figure:

Table 1
– Independent variables included in the final model (statistically significant)

	Total	% palatalization	Factor Weight	Sig, Wald	Sig. MLRT
City					6.3e-08
São Miguel dos Milagres	546	32.8	0.68	0.002
União dos Palmares	569	25.3	0.62	0.046
Maceió	791	25.9	0.54	0.441
Santana do Ipanema	385	22.6	0.53	0.702
Arapiraca	603	19.2	0.52	0.807
Palmeira dos Índios	440	16.6	0.51	0.827
Delmiro Gouveia	712	5.6	0.16	<0.001
Age * Schooling 4					1.8e-06
Schooling				<0.001
Age				0.084
Schooling* Age				0.033
Position in the lexical item					8.3e-06
internal	3513	23.4	0.68	<0.001
border	533	4.3	0.32	<0.001
Next vowel					3.7e-05
i	413	19.6	0.63	0.026
u	2517	26.1	0.56	0.239
e	178	5.6	0.56	0.481
o	68	14.7	0.46	0.654
a	870	9.8	0.29	<0.001
Consonant type					0.0009
t	3528	23.3	0.63	0.001
d	518	4.2	0.37	0.001
Tonicity					0.039
unstressed	3511	22.8	0.56	0.043
tonic	535	8.4	0.44	0.043
Total	4046	20.9

Source: Authors’ elaboration.

Figure 1
– Effect of the city variable on the palatalization of alveolar stop consonants in Alagoas5

From this result, we can infer three regions of palatalization in Alagoas: the west, the northeast, and a region that aggregates the center and the middle east, which we named the center-east. To test the effect of these regions, we created a new region variable with three categories: west (Delmiro Gouveia), center-east (Maceió, Santana do Ipanema, Arapiraca, and Palmeira dos Índios), and northeast (São Miguel dos Milagres and União dos Palmares). Then, using the MLRT, we tested the difference between the city and region variables in the adjusted models. The result was a significance of 0.949, which means that the models are practically identical and, therefore, the model that best explains the variability is the simplest, the one that contains the region variable. The following table demonstrates the results of this variable:

Table 4
– Effect of the region variable on the palatalization of alveolar stop consonants in Alagoas

	Total	% palatalization	Factor Weight	Sig. Wald	Sig. MLRT
Region in Alagoas					3.4e-10
Northeast	1,115	29.0	0.72	<0.001
Center-East	2,219	21.7	0.60	0.007
West	712	5.6	0.21	<0.001
TOTAL	4,046	20.9

Source: Authors’ elaboration.

Considering the results of the analysis, we conclude that there is a diatopic effect on palatalization in Alagoas, increasing from west to east and amplifying in the northeast region of the state, as we can see in the following map:

Map 1
– Distribution of palatalization of alveolar stop consonants in Alagoas

Our results confirm the initial hypothesis that palatalization is geographically conditioned and has a direction (from west to east) in the state of Alagoas.

Subsequently, we analyze the effect of age and schooling on the palatalization of alveolar stop consonants.

Figure 2
– Effect of the interaction between age and schooling on the palatalization of alveolar stop consonants in Alagoas

The lines in the Figure above represent the effect of schooling on different age groups. We observe that, at all ages, there is an inversely proportional effect of schooling, which indicates that the higher the level of schooling, the lesser the use of palatalization of stop consonants. We also verify that, with increasing age, the slope of the schooling straight line decreases, which indicates that schooling has its effect reduced with increasing age.

It is reasonable to assume that the linguistic behavior of speakers indicates that palatalization has acquired, over the years, a negative social value and that the school is the primary social environment for the dissemination of such value. This result is related to that obtained in Oliveira, Oliveira and Paula (2018) in Maceió, in which it became clear that the effect of schooling on the speaker was different depending on his/her age, since the most educated favored the process the most, when young; while the primary inhibitors were the elderly. Now, we have observed that this effect extends to other Alagoas speeches.

Sociolinguistic research that have investigated the correlation between schooling and variable linguistic processes are generally faced with double social information, the influence of the level of schooling on the interviewee linguistic choice, and evidence of their social class, since the trend is that the level of the interviewee’s social status follows their school development.

Because schools use the linguistic structures, authority patterns, and curricula of the dominant culture (i.e. that of the middle and upper classes), there is a natural alignment between middle-class families and the culture of the school ( CREGAN, 2008 , p. 12).

When investigating the relationship between schooling and the realization of linguistic variants, we can verify the influence of social valuation on the linguistic speakers’ choices, since the educational environment contributes to the promotion of cultured forms of language, as it stigmatizes other competing forms, attributing negative values to them.

Although we have identified an effect of age on palatalization, this effect does not appear to be related to a linguistic change in progress, at least for the time being. What we have is a strong influence of the school in reducing the palatal variant among younger speakers. We can affirm that the progressive palatalization of alveolar stop consonants preceded by /j/ suffers negative social pressures and is becoming increasingly resistant in educational environments, affecting primarily the young.

The analysis of the linguistic independent variables investigated in this study is presented below.

The variable position, on the lexical item, refers to the position of the process in relation to the lexical items; if it is internal to the item and, therefore, /j/ belongs to the same item (e.g. va id oso (‘vain’) and ace it a (‘accepted’); or if it is at the border, with /j/ being the final segment of the previous lexical item (e.g. pegue i d uas (‘I got two’) e fo i t omar (‘went to have a drink’)). The results presented in table 1 demonstrate that the border context is highly unfavorable to the progressive palatalization (4.3% and RW=0.32); unlike the internal context (23.4% and RW=0.68).

Figure 3
– Effect of the variable position on the lexical item in the palatalization of alveolar stop consonants in Alagoas

The variable position on the lexical item revealed that although palatalization is possible in the limit of lexical words, this probability decreases when the stop consonants are in a border beyond the limit of phonological words. It can be said, then, that the process is highly favored in the domain of the phonological word, but it is not blocked in domains above it.

The variable following vowel refers to the vowel following consonants t/d: a (muit a ) (‘many’/‘much’), e (d e le (‘his’), i (jeit i nho (‘subterfuge’), o (aceit o u (‘accepted’)), u (t u do (‘everything’). The results expressed in Table 1 and Figure 4 demonstrate that palatalization is favored by the following vowel i (RW=0,63), and disfavored by the following vowel a (RW=0,29.) The other vowels present no statistical difference concerning the average effect.

Figure 4
– Effect of the variable following vowel on the palatalization of alveolar stop consonants in Alagoas

Since the [+coronal] feature is the universal trigger in the palatalization process ( NEUSCHRANK; MATZENAUER, 2012 ), it is reasonable to think that its double presence, both in the previous context, with the semivowel /j/, and in the following, with the vowel /i/, causes a double trigger for the process, indicating that in this context there is a significantly higher probability of palatalization.

The consonant type variable refers to the target consonant of the process, /t/ or /d/ (mui t o (‘very’/‘much’), doi d o (‘crazy’). The results presented in table 1 and Figure 5 demonstrate that the probability of occurrence of palatalization is much higher when the consonant is /t/ (23.3% and RW =0.63) than when it is /d/ (4.2% and RW=0.37).

Figure 5
– Effect of the variable type of consonant on the palatalization of alveolar stop consonants in Alagoas

The observed result corroborates other studies that demonstrate the favoring of /t/ in the process ( MOTA; ROLEMBERG, 1997 ; SANTOS, 1996 ; HENRIQUE; HORA, 2012 ; SOUZA NETO, 2014; OLIVEIRA, A. A., 2017). It is possible that the favoring of /t/ is associated with the absence of vocal fold vibration, which causes this consonant to be articulated with less energy. Such argument is highlighted by Hora (1990) , Abaurre and Pagotto (2002) , and Battisti and Guzzo (2009) .

The tonicity variable refers to the presence or absence of stress on the syllable containing the target consonant t/d, e.g. mui to (‘very’/‘much’) and oi to centos (‘eight hundred’) (unstressed), and prefei tu ra (‘municipality’) (stressed). The results presented in Table 1 and Figure 6 demonstrate that palatalization is more likely to occur in unstressed syllables (22.8% and RW=0.56) than in stressed syllables (8.4% and RW=0.44).

Figure 6
– Variable tonicity and palatalization of alveolar stop consonants in Alagoas

This result confirms the hypothesis that the unstressed syllable favors the process. We follow the affirmation by Bisol (1991) , that the salience criterion attests that innovative forms, in less prominent syllables, tend to be more successful than innovations in strong syllables.

The incipient rule shows preference for sensitive manifestations that go unnoticed, that is, for less complex positions in the syllable structure, or for less prosodic strength, or for less complex phonological segments and less abstract underlying representations. Lessening the external effect, it thus prevents criticism of unusual linguistic behavior, most harmful for the expansion of a rule and its consequent generalization ( BISOL, 1991 , p. 117, our translation)⁴

Thus, it is possible to explain how the process of progressive palatalization of alveolar stop consonants presents a different tendency for realization, depending on the tone of the syllable in which the target consonants appear, and why the stressed syllable decreases the probability of occurrence of the palatalization process.

Conclusion

Two conclusions are worth highlighting in this work: (1) There are significant regional differences in progressive palatalization in Alagoas, given the increased incidence of the process going from west to east, and expanding in the northeast of the state; (2) The process occurs less as schooling increases, and, more interestingly, schooling interference grows as the speaker’s age decreases, which reveals a process of change in the social appreciation of palatalization.

There are still other advances in the study. The control of the aggregate variables individual and lexical item, through the use of multilevel regression models, allowed greater reliability in the identification of the effects associated with the independent linguistic and social variables, isolating eventual idiosyncrasies related to elements of such aggregated variables.

Although the process is significantly more productive in Alagoas (compared to other regions in northeastern Brazil), it is likely that, due to social pressures, we will not observe an increase in its prevalence over time.

Regarding linguistic variables, we conclude that the process has a higher probability of occurring in word-medial context, although it also occurs at the border of lexical items. The following vowel that favors the process is /i/, which reveals the increased probability of palatalization due to the double presence of the [+coronal] feature in adjacent environments. The absence of voicing in the target consonant and the unstressed nature of the syllable also contributes significantly to the incidence of palatalization.

REFERÊNCIAS

• ABAURRE, M. B. M.; PAGOTTO, E. G. A palatalização das oclusivas dentais no português do Brasil. In: NEVES, M. H. de M. (org.). Gramática do português falado VII: novos estudos descritivos. Campinas: Ed. da Unicamp, 2002. p. 557-602.
• BATTISTI, E.; GUZZO, N. B. Palatalização das oclusivas alveolares: o caso de Chapecó. In: BISOL, L.; COLLISCHONN, G. (org.). Português no sul do Brasil: variação fonológica. Porto Alegre: EDIPUCRS, 2009. p.114-140.
• BISOL, L. Palatalization and its variable restriction. International Journal of Sociology of Language , Mouton, n. 89, p. 107-124, 1991.
• CHIZZOTTI, A. Pesquisa em ciências humanas e sociais . 4 ed. São Paulo: Vozes, 2011.
• CREGAN, A. Sociolinguistic Perspectives on the Context of Schooling in Ireland: parent perceptions. Working Paper Series 08/04, Combat Poverty Agency, v.II, August 2008.
• HENRIQUE, P.; HORA, D. Um olhar sobre a palatalização das oclusivas dentais no vernáculo pessoense. In: JORNADA NACIONAL DO GRUPO DE ESTUDOS LINGUÍSTICOS DO NORDESTE, 24., 2012, Natal, RN. Anais [...] Natal: EDUFRN, 2012. p. 150-161.
• HORA, D. A palatalização das oclusivas dentais: variação e representação não-linear. 1990. Tese (Doutorado em Letras) – Instituto de Letras, PUC-RS, Porto Alegre, 1990.
• LABOV, W. Padrões Sociolinguísticos . Trad. Marcos Bagno. São Paulo: Parábola, 2008. Original de 1972.
• MOTA, J.; ROLEMBERG, V. Variantes africadas palatais em Salvador. In: HORA, D. (org.). Diversidade Linguística no Brasil . João Pessoa: Ideia, 1997. p. 131-140.
• NEUSCHRANK, A.; MATZENAUER, C. L. B. A palatalização na diacronia do PB: o surgimento dos segmentos palatais à luz de teoria fonológica. Linguística, Montevidéu, v. 27, p.18-46, jun. 2012. Disponível em: http://www.scielo.edu.uy/scielo.php?script=sci_arttext&pid=S2079-312X2012000200003 . Acesso em: 22 jun. 2021.
  » http://www.scielo.edu.uy/scielo.php?script=sci_arttext&pid=S2079-312X2012000200003
• OLIVEIRA, A. A. Processos de Palatalização das oclusivas alveolares em Maceió . 2017. Tese (Doutorado em Letras e Linguística) - Programa de Pós-Graduação em Letras e Linguística, Universidade Federal de Alagoas, Maceió, 2017.
• OLIVEIRA, A. J. Projeto PORTAL: variação linguística no português alagoano. Disponível em: http://www.portuguesalagoano.com.br . Acesso em: 22 jun. 2017.
  » http://www.portuguesalagoano.com.br
• OLIVEIRA, A. A.; OLIVEIRA, A. J.; PAULA, A. S. Palatalização das oclusivas alveolares [t] e [d] com a semivogal [j] em contexto anterior na cidade de Maceió. Revista Leitura, Maceió, v. 1, n. 60, p. 102-122, jan./jun. 2018.
• SANTOS, L. F. Realização das oclusivas /t/ e /d/ na fala de Maceió . 1996. Dissertação (Mestrado em Letras e Linguística) - Programa de Pós-Graduação em Letras e Linguística, Universidade Federal de Alagoas, Maceió, 1996.
• SOUZA NETO, A. F. Realizações dos fonemas /t/ e /d/ em Aracaju – Sergipe . Aracaju: Ed. da UFS, 2014.

• 1
  This codification provides information about the informant’s city, age, sex, and education.
• 2
  Project funded by CNPq (406218/2012-9) and approved by the Ethics Committee of the Federal University of Alagoas, Report No. 621,763.
• 3
  Original: “ um relato retrospectivo da experiência pessoal de um indivíduo, oral ou escrito, relativo a fatos e acontecimentos que foram significativos e constitutivos de sua experiência vivida. ” ( CHIZZOTTI, 2011 , p. 101).
• 4
  The variables age and schooling are continuous. Therefore, there are no totals, percentages, and relative weights. The results of the interaction between these variables will be presented graphically.
• 5
  In this Figure, and in the others that will follow the same model, the points for each factor represent the relative weight; and the dashes at the end of the lines indicate the confidence interval, within which is 95% of the probable estimates (which leads us to 5% error, and at a significance level of 0.05). When the line crosses the 0.50 RW, it means that we cannot say that the effect of the factor differs from the average effect (or neutral point) with a significance of 0.05.
• 6
  Original: “ A regra incipiente mostra preferência por realizações sensíveis que passam despercebidas, isto é, para posições menos complexas na estrutura da sílaba ou para menor força prosódica, ou para segmentos fonológicos menos complexos e representações subjacentes menos abstratas. Diminuindo o efeito externo, isso impede a crítica de comportamento linguístico incomum, mais prejudicial para a expansão de uma regra e sua consequente generalização .” ( BISOL, 1991 , p. 117).

Publication Dates

• Publication in this collection
  01 Nov 2021
• Date of issue
  2021

History

• Received
  19 Feb 2019
• Accepted
  08 Oct 2020