Changing language input following market integration in a Yucatec Mayan community.

Like many indigenous populations worldwide, Yucatec Maya communities are rapidly undergoing change as they become more connected with urban centers and access to formal education, wage labour, and market goods became more accessible to their inhabitants. However, little is known about how these changes affect children's language input. Here, we provide the first systematic assessment of the quantity, type, source, and language of the input received by 29 Yucatec Maya infants born six years apart in communities where increased contact with urban centres has resulted in a greater exposure to the dominant surrounding language, Spanish. Results show that infants from the second cohort received less directed input than infants in the first and, when directly addressed, most of their input was in Spanish. To investigate the mechanisms driving the observed patterns, we interviewed 126 adults from the communities. Against common assumptions, we showed that reductions in Mayan input did not simply result from speakers devaluing the Maya language. Instead, changes in input could be attributed to changes in childcare practices, as well as caregiver ethnotheories regarding the relative acquisition difficulty of each of the languages. Our study highlights the need for understanding the drivers of individual behaviour in the face of socio-demographic and economic changes as it is key for determining the fate of linguistic diversity.

Introduction

Like many indigenous communities throughout the world, Yucatec Maya communities in Mexico are undergoing rapid change as they integrate into dominate market economies [1-7]. In this paper, we consider how these changes may affect children’s experiences, by asking whether the language addressed to children shifts during a period of ongoing market integration and why it might do so.

There are several reasons to predict language input change. First, during market integration (and associated opportunities for wage labor, access to education and market goods), parents may shift input away from minority local languages and towards majority surrounding languages because they believe that dominant languages are socially and economically advantageous for children to learn [3-10]. For example, Hill and Hill [11] and Rolstad [12] attribute the decline of Nahuatl speakers in Mexico over the 20th century to parents using Spanish (rather than Nahuatl) with their children in order to prepare them for the newly introduced bilingual schooling policies. But changes in the input received by children in small-scale communities could also be caused by factors beyond those that imply parental devaluation of native languages.

Market integration could lead to changes in childcare practices that shift caregiving away from older children to adults and consequently affect language input [13,14]. In many cultural settings, it is not parents who have traditionally served as infants’ primary interlocutors, but rather other children [15-17]. If changing market economies result in infants spending less time with older siblings, this could result in less time that infants are engaged in directed conversations overall (including conversations in a local language). Changes in childcare in emerging integrated contexts could result from several factors. For example, the reduction in fertility that often accompanies transitions from subsistence to skill-based economies could mean that there will be fewer older children to care for, and talk to, young children [14,18]. Moreover, if older children are expected to devote more time to their education, they may have less time to devote to household labour, including childcare [19,20]. Consequently, speech directed to young children might face overall decline, and, when combined with the other factors described below, affect local language input, even if there is no change in parental beliefs about the value of one language over another. Given the pivotal role of child-directed speech for the subsequent development of linguistic competences [21], these changes in input patterns could gradually result in fewer and fewer members of the younger generations becoming proficient in the local language.

Additionally, during the process of market integration, both parents and older siblings of young children in small-scale communities are more likely than previous generations to have had access to formal schooling provided by the majority community. This access both exposes caregivers to non-local languages (providing them with the ability to use the language in interaction with young children) and also introduces caregivers to the interaction patterns of the dominant culture. Schools generally model a pattern of interaction and socialization in which children are directly addressed in pedagogical interaction (e.g. [22]) Caregivers who have been schooled might integrate these socialization patterns into their childcaring practices with respect to both their local and non-local languages [14,23-25,28], by, for example, directing more speech to children. Alternatively, they might adopt pedagogical practices more narrowly, reserving the non-local language (also the language of school instruction) for directed interactions (those typical of school contexts). If so, directed speech to children would occur more often in the non-local language than the local language, again potentially affecting the relative ability of children to learn each of the languages.

Finally, broader cultural ethnotheories about language learning may affect caregiver’s input strategies during market integration. Rather than undervaluing the local language, parents may use directed speech specifically in a non-local language because they have beliefs about the necessity of doing so. Previous research suggests that, in many small-scale communities at greatest risk for language loss, development of the native language is thought to come from within the child, requiring little outside intervention [1,8,16]. In these contexts, directed linguistic input may be viewed as less important for supporting native language transmission than it is in cultural communities where directed teaching is considered critical [26-28]. Indeed, Shneidman and Goldin-Meadow [16] found that caregivers from the United States were 7 times more likely to direct speech to their infants than caregivers from rural Yucatec Mayan communities. Other studies have confirmed this general pattern––infants from market-economies are between 2 and 11 times more likely to receive child-directed speech (CDS) than infants from small-scale, subsistence economies [17,29,30].

Along the same lines, caregivers may believe that the non-native language is more difficult to learn than the native language, leading to a need for more focused input in the non-native language. In emerging bilingual contexts adults and older children are more likely to have learned the non-local language later in life and thus may have encountered difficulty in their own learning. This may lead them to the false conclusion that the non-local language is inherently more difficult to learn than the local language [3]. Such a belief could result in caregivers using more directed speech to children in the non-local language than the local language, even when they desire that children acquire equal competences in each.

In summary, four factors could explain the fact that children receive less directed speech in their native language than in the non-local language when small-scale societies are being integrated into majority cultures: 1) a general devaluing of the native language, 2) changes in child-care practices, 3) changing access to education and 4) specific parental beliefs regarding the importance of directed input for language learning. In order to assess the contributions of these factors, it is necessary to quantify changes in children’s input during periods of market integration, and determine how these changes intersect with changing demographics and belief systems. This is not only important to broaden our understanding of the consequences of market integration for indigenous communities but to predict the fate of the world’s linguistic diversity. As proposed by Lambert [10], the sustainability of minority languages might be unavoidably compromised where bilingualism is a side-effect of a general process by which individuals feel forced or incentivized to put aside their indigenous language for a more necessary or prestigious one. However, there is also the possibility of “additive bilingualisms” whereby individuals may want to invest in learning an additional language to be able to communicate with out-groups for particular purposes (such as education, commerce or travels) whilst retaining their ethnic (and linguistic) affiliation.

We address these questions by taking advantage of a “natural experiment” arising in the Yucatán peninsula as a result of economic development, new means of transportation, opportunities for wage labor, and access to education. Using video data collected from natural interactions, we compared the linguistic input received by 16-24-month-old infants growing up in the same Yucatec Mayan villages across two cohorts approximately 6 years apart (n = 21 in 2007/08, see ref. 20, and n = 15 in 2013/14). Previous work in this community showed that input heard by infants in 2007/2008 was primarily in the local language, Yucatec Maya, and provided by other children rather than adults [16]. Here, we ask whether these characteristics shifted in this community after a period of increased market integration. In addition, we conducted structured interviews with 126 adults from the villages (including all of the infant’s primary caregivers) in January 2019. These interviews provide demographic information related to the changing economy, educational practices, and care practices, and also shed light on caregiver beliefs and attitudes about language. In particular, they provide insight into how important caregivers thought Maya vs. Spanish was for their infant’s future success, and how difficult they thought the two languages were to learn. We examined whether language input changed over the period of market integration––in quantity, type (directed, overheard), and language choice (Maya, Spanish)––and whether those changes in language input were related to childcare, educational, and belief practices.

Results

We first used interview data and family histories to assess socio-demographic changes across the six-year period in the focal villages. As expected, there were general differences in variables related to increased contact with the majority culture across cohorts––greater access to education in the majority language, wage labor, and market goods (S1 Text, S1 and S2 Tables, S1 Fig). For example, in 2007, children could only complete primary school in their natal village; by 2013, they could also attend secondary school. Primary school attendance of caregivers had also doubled across cohorts (2007: Mean = 2.17, SD = 1.64; 2014: Mean = 4.09, SD = 2.51). Likewise, whilst 66% of fathers from infants in cohort 1 engaged in some form of wage labor (regardless of whether they also worked in agriculture), 82% of those from infants in cohort 2 did so. In cohort 1, the vast majority of fathers (89%) had a “milpa” (i.e. worked in agriculture) regardless of whether they also engaged in wage labour. However, this number had dropped to 63% in the second cohort, indicating that more households are transitioning to a complete reliance on market jobs. However, in line with findings by Schacht et al. [31] and Gaskins [1], many aspects of the villages had not changed during this period: Age of mother at first birth (2007: Mean = 20.08, SD = 3.03; 2014: Mean = 19.60, SD = 4.22), and average number of older siblings of focus infants (2007: Mean = 3.89, SD = 2.75; 2014: Mean = 3.19, SD = 2.52); no infant had younger siblings because the target infant was always the youngest of the family (weaning age ranged from 2–2.5 years).

In order to evaluate whether there had been changes in patterns of linguistic socialization, and the prevalence of Spanish in the speech heard by infants following market integration, we used transcriptions from the natural recordings. First, we assessed the number and type (child-directed, overheard) of utterances received by infants, and who produced the utterance (primary caregiver, another adult, older child). We fitted Bayesian multilevel zero-inflated Poisson models to the data, and found changes across cohorts in all three measures (see S1 Text). Mean number of utterances per hour directed to children decreased from 2007 to 2014 from primary caregivers (99.6 to 76.8 utterances) and, particularly, from other children (346.7 to 165.7), The number of directed utterances from other adults remained relatively stable (36.86 to 37.79) (Fig 1, top three graphs). In contrast, mean number of overheard utterances per hour increased slightly in primary caregivers (29.8 to 51.0) and in other adults (38.9 to 66.5), and decreased slightly in other children (104.6 to 93.9; Fig 1, bottom three graphs). For all input types, most utterances originated from other children (68.8% of utterances in cohort 1, 52.8% in cohort 2; S7 Table). The results were consistent across all villages (i.e. village-specific intercepts were roughly symmetrical around 0; S2 Fig). Thus, the overall amount of directed speech children heard, particularly from other children, decreased over the 6-year period; in contrast, the amount of overheard speech children heard changed very little.

Fig 1

Posterior predictive distribution of the mean difference in number of utterances of each type received by the average child from 2007 to 2013, as obtained from the Zero-Inflated Poisson model including “Cohort” as predictor variable and the number of utterances of each type as response variable.

These were obtained by averaging from 12000 samples from the posterior distribution (setting the standard deviations for the varying intercepts to 0). From top-left to bottom-right: Directed input from primary caregiver, directed input from adults, directed input from children, overheard input from primary caregiver, overheard input from adults, overheard input from children.

While there was no significant difference in age between cohort 1 and cohort 2, the age ranges of both cohorts were not completely overlapping (the age range for cohort 1 was 16.1–24 months and cohort 2 was 16.1–22.5 months). Because older children might elicit more speech from their caregivers, due to changing language competences, we assessed whether the decrease in overall input (and in particular input from primary caregivers) from cohort 1 to cohort 2 could be explained by child production or child age. We first fitted the same Bayesian multilevel zero-inflated Poisson model with “Number of directed utterances per hour received by target child from primary caregiver” as response variable and “Cohort” as well as “Number of utterances produced by the focal child” as predictor variables. Consistent with our previous findings there was a negative effect of “Cohort” on the number of utterances that infants received from their caregivers (Estimate = -0.43; 90%HPDI: [-0.53,-0.33]). However, the number of utterances produced by the focal child did not affect the number of directed utterances they received from their caregivers (Estimate = -0.01; 90%HPDI: [-0.01,0]). We next re-fitted all the models reported in Fig 1 and excluding from Cohort 1 all 23- and 24-month-old infants (N = 2). Results (see S3 Fig) showed that, without these children, the apparent reduction in CDS from the primary caregiver disappeared, perhaps indeed owing to the fact that mothers are more likely to direct speech to older infants. However, there was still a significant overall reduction in the total number of directed utterances received by infants from the second cohort, stemming from a reduction in utterances directed from other children. Thus, even when taking child production and age into account, infants from cohort 2 received less directed input than infants from cohort 1.

To examine whether the language of input shifted from Yucatec Maya to Spanish over the 6-year period, we fitted Bayesian logistic mixed models to the proportion of input in Spanish (vs. Maya) that infants received in child directed and overheard speech. Since code-switching within utterances was extremely rare (mean = 1.6 utterances per hour, SD = 2.08 and mean = 3.5 utterances per hour, SD = 7.09 in cohorts 1 and 2 respectively; out of an average of 464 utterances heard per hour) and did not vary across cohorts (t = -1.01, df = 14.59, P = 0.33), we excluded them from these analyses. We found significant increases in the proportion of child-directed input that infants received in Spanish (instead of Yucatec Maya) between 2007 (Mean = 0.21, 90% HPDI: [0.03, 0.47]) and 2014 (Mean = 0.67, 90% HPDI: [0.32, 0.93]) (Fig 2, left graph). In contrast, there were no significant changes in the proportion of overheard speech infants received in Spanish between 2007 (Mean = 0.27, 90% HPDI: [0.07,0.55]) and 2014 (Mean = 0.17, 90% HDI: [0.03, 0.37]) (Fig 2, right graph). Thus, caregivers altered the proportion of speech they directed to infants in favor of Spanish over the 6-year period, but did not change the proportion of speech overheard by infants (S8 Table). By 2014, caregivers (whether they were primary caregivers, other adults, or children; S8 Table) used Spanish in the speech they actively directed to infants, but primarily used Maya in the speech overheard by infants.

Fig 2

Posterior means (bars) and 90% HPDI (error bars) for the models predicting the proportion of utterances in Spanish (Number of utterances in Spanish out of the total number of utterances of that type) an average infant (infant with an intercept at 0 for village) is exposed to.

To summarize, infants in cohort 2 generally received less directed input than infants in cohort 1 but, when that input was addressed to them, it typically came in Spanish. Indeed, changes in directed input were driven, fully, by decreases in directed input in Yucatec Maya (S12 Table). When we consider the total number of utterances per hour in directed speech heard by infants in each of the languages, we find that infants in cohort 1 heard an average of 397.3 Maya utterances per hour in directed speech, but infants in cohort 2 children heard only an average of 92.51 Mayan utterances.Why did caregivers decrease the amount of Maya they directed to infants? We turn to interview data gathered in 2019 to address this question. 126 adults from these villages (including all of the primary caregivers from the two cohort samples) were asked if it was more important to learn Maya or Spanish and, as a follow up question, why each language was important to learn (S1 Text). All of them reported that learning Maya and learning Spanish were equally important. Responses to questions about why it is important to learn Maya and Spanish are summarized in Table 1. In general, Spanish was seen as a functional tool important for necessities, such as communicating with doctors or going on shopping trips to nearby Spanish speaking cities. Maya, in contrast, was regarded as important for maintaining social relationships and cultural ties within the village.

Table 1

Answers to questions regarding why each language is important from interviewees.

Question	Answer	Count
Why is Maya important/necessary?	To communicate in the village, as everyone speaks Maya	34
	To preserve the language/prevent disappearance	18
	Some people don’t understand Spanish	13
	To know where one comes from, it is our cultural heritage/tradition	9
	One needs it for everything	4
	Learning Maya allows you to learn how to speak earlier in life	3
	It is pretty	3
	For work (requisite in hotels)	2
	Help old people translate	2
	To speak to one’s family	1
	It is more similar to English	1
	To teach one’s children	1
	If you don’t, people here make fun of you	1
	To go shopping	1
	To be able to work in corn fields (“milpa”)	1
	TOTAL	96
Why is Spanish important/necessary?	For going to the doctor	31
	For visiting cities	30
	To speak/understand visitors	16
	To fully express yourself	7
	To be able to work outside	13
	Now one needs to learn it for everything	6
	One needs it to talk to schoolteachers	6
	It is very popular now	6
	For understanding telenovelas	1
	For bureaucracy	1
	In case you get lost	1
	TOTAL	118

Note that counts do not add up to 126 because responses that did not answer the question were excluded as they offered no explanatory power.

In order to explore attitudes about the need for directed input in learning language, adults were asked: “Some people believe that language needs to be taught, while others believe that language will just come out on its own; what do you believe? And, as a follow-up, “Is that the same for Maya and Spanish learning?” In response to the first question, 81.9% (N = 98) of adults believed that infants needed to be actively instructed in how to speak. This pattern was found in the primary caregivers of infants in both the first (72.3%; N = 16) and second (60.9%; N = 9) cohorts, with no significant differences between cohorts (χ2 = 0.20, df = 1, p = 0.65). As an example, one mother stated: “Mothers need to speak to their babies for them to learn how to speak because alone they can’t. I have a nephew who only learned how to speak when he was seven because his mother didn’t teach him.” Interestingly, in response to the second question, 49.4% (N = 61) of the interviewed adults believed that infants learnt Spanish and Yucatec Maya in different ways (S11 Table). Specifically, 34.1% (N = 43) of interviewees thought Yucatec Maya was easier to learn than Spanish, sometimes reporting that Spanish required infants to be actively taught, whereas Maya could be learned from overhearing. This pattern raises the possibility that an additional factor responsible for the increase in directed speech in Spanish arose, not from devaluing Maya (and thus trying to substitute Spanish for it), but from the belief that child-directed speech is essential in order to learn Spanish, but not Maya.

To test this hypothesis, we coded caregivers’ responses to the above question based on whether the answer suggested that the respondent thought that Maya was easier to learn than Spanish, Spanish was easier to learn than Maya, the two languages were learnt in the same way, or the response was not codable along this dimension (e.g., “They learn Spanish better when they are older”) (See S11 Table for details on the coding of answers). We then used Bayesian logistic mixed models to assess whether the belief that Spanish was harder to learn than Maya predicted the proportion of input that caregivers directed to their infants in Spanish. We confirmed that children whose primary caregivers thought Spanish was harder to learn were more likely to speak to their children in Spanish both in 2007 (log-odds = 1.59, 90% HPDI: [1.06, 2.1]) and 2014 (log-odds = 1.1, 90% HPDI: [0.67, 1.56]) (Fig 3). The proportion of mothers who held the belief that Maya was easier to learn was not different across cohorts (58.3%, n = 7 versus 33.3%, n = 3; Fisher’s exact text, p = 0.21), nor was it related to how fluent they were in Spanish (χ2 = 0.47, df = 1, p = 0.50).

Fig 3

Posterior means (bars) and 90% HPDI (error bars) for the models predicting the proportion of utterances in Spanish (Number of directed utterances in Spanish out of the total number of directed utterances) primary caregivers directed to their infants in Cohort 1 (left) and Cohort 2 (right) according to whether they believed Yucatec Maya was learnt easier/faster than Spanish or not.

Discussion

Our results show, for the first time, that the quantity, type, source, and language input to Yucatec Maya infants changed as the communities became more connected with urban centers, and hence education, wage labour, and market goods became more accessible to their inhabitants. Infants in cohort 2 received less total directed input than infants in cohort 1 and, when they were directly addressed, most of their input was in Spanish. The net result of these changes was that infants in cohort 2 received far less directed input in the Yucatec Maya language than infants in cohort 1.

Theorists have proposed that shifts in input away from minority local languages and towards majority languages are due to a belief that dominant languages are more advantageous for children to learn than the minority language [8-10]. However, in the Yucatec Mayan community considered in this study, reductions in Mayan input did not seem to be due to speakers devaluing the Maya language––our interview data suggest that caregivers saw value in infants learning both Maya and Spanish, but for different reasons [32]. Spanish was seen as important for achieving pragmatic goals, such as going to town, finding work, or going to the doctor; Maya was seen as important for social goals, such as communicating in the village and maintaining the Mayan culture. One marker of the continued importance of Maya was the fact that, across cohorts, most overheard speech was in Maya throughout the period of change. This pattern indicates that Spanish was not replacing Maya overall, but only in speech directed to children.

We hypothesize that several factors, other than devaluing the local language, may contribute to the decreasing amounts of input in Maya that infants received. First, in this community, both in the past and currently, most input to infants comes from other children, not adults. We found an overall reduction between cohorts in child-directed input from other children, which might have resulted from infants spending less time with other children. These reductions are unlikely to be due to shifts in family composition following increased market integration (as has been suggested in other small-scale communities, e.g. [2,10,20] because we found no evidence that family size changed across cohorts (S1 and S2 Tables). It is possible, however, that schooling has gained importance over this period of time and that older children are focusing more on schoolwork and less on sibling care. Indeed, in a comparison of childcare patterns in 1992 and 2011 in a very similar Mayan community, Kramer and Veile [33] report that, even in the absence of changes in fertility or residence arrangements, the amount of time 7–10 year-old children dedicated to childcare decreased from 12% to a little over 6% (primarily traded off by care from fathers). In future research, we plan to evaluate whether similar changes in childcare practices have occurred in the target communities and, if so, why. Although we also observed a slight yet significant decrease in the quantity of child-directed input infants received from their primary caregivers in the second cohort when compared to the first one, this decrease can be explained by the presence of older children in our cohort 1 sample than in our cohort 2 sample.

The second factor that may contribute to the decreasing amounts of input in Maya that infants received is caregiver beliefs about the importance of directing speech to children in Maya vs. Spanish. We found that many caregivers endorsed a belief that Maya is more ‘naturally’ or ‘easily’ acquired than Spanish; parental effort is therefore better placed at teaching Spanish. This result mirrors findings from other indigenous communities in the Americas. For example, following the introduction of bilingual education programs in a K’iche Mayan community, Choi [34] found that many families did not support teaching K’iche at home or in school. The caregivers thought that the K’iche language was “naturally” learned without schooling, and therefore any attempts to teach it would be a waste of their and their children’s time. However, the same attitude did not apply to learning Spanish; the caregivers believed education to be necessary for Spanish to be learned. Several other researchers working in bilingual indigenous communities have reported similar beliefs (e.g. [8,35]).

Our results demonstrate that these beliefs matter for predicting caregiver behavior. Caregivers who thought that Maya was easier to learn than Spanish were more likely to use Spanish, and less likely to use Maya, when speaking directly to infants (Fig 3). Previous research indicates that speech directed to infants, and not overheard speech, predicts children´s later language competences [36]. This result holds even in Mayan communities where overheard speech is prevalent [16]. Given this finding, one possibility is that differential ethnotheries regarding language acquisition could cause majority languages to replace minority ones despite the fact that caregivers place equal value in both languages. Nonetheless, whether active valuation of the native language results in the majority language replacing the minority language, or in the coexistence of both tongues, is an important issue for future research.

Future research should also consider the role of the child’s production in influencing the amount of directed input that he or she receives. Although, in the current study, we did not find a relation between the amount of talk that children produced and the input they received from others, we did not conduct a more fine-grained assessment of children’s competencies in each of the languages. It could be the case that, if decreases in the input directed to children are language-specific (i.e. in Maya), children may become less competent in Maya, and that this decrement might then affect the amount of Maya the children elicit in directed input.

In conclusion, in the Mayan communities that were the focus of this study, increased market integration over the past decade has been associated with an overall decrease in speech addressed to infants, as well as a decrease in the proportion of total input that infants hear in Yucatec Maya, which is replaced by Spanish. Our results suggest that Maya caregivers are not actively trying to replace Yucatec Maya with Spanish because they believe that Spanish is more social or economically invaluable than Maya. Rather, we hypothesize that caregiver perceptions of the relative difficulty of learning Spanish vs. Maya, as well as changes in childcare practices, are factors associated with decreases in native-language input directed to children.

Materials and methods

Ethnographic context

All the data come from four Mayan villages located in the state of Yucatán about 80 miles to the southwest of Cancun (the largest Spanish-speaking urban centre in the region) (S1 Text). Traditionally, Maya families have made their living as subsistence maize farmers. Following the Mexican Revolution, the ejido land tenure system was established, whereby each Mayan village is conceded a plot of land on which to build a house, as well as surrounding country for crops, pasture, and woodland [37]. Ejido lands (as initially written) could not be owned, inherited, sold, or rented and their dominion resided within the village collective, which distributed them among married males [38].

Recently the Maya have experienced rapid socioeconomic changes due to the growth of lowland towns, creation of new roads, improved transport, greater availability of schools, and increased contact with Mexican and global cultures [38]. These changes allowed many individuals, particularly unmarried males, to work for wages in nearby Spanish-speaking urban centers, such as Cancun or Playa del Carmen [39]. However, wage jobs are seen as a supplement to agricultural work, needed to increase household productivity in times of need and not as a replacement for it (S1 Text). Nonetheless, in contrast to what many (e.g., [40,41]) have regarded as an inevitable and gradual process of acculturation, social and residential structures seem to have remained strikingly stable [1].

Another important change in the region has been the increased access of children from rural communities to education. In most schools (including those from the villages of the present study), textbooks are provided only in Spanish and, although some teachers do use both languages, Maya is principally employed for classroom management, whereas Spanish is the language of instruction [42].

Naturalistic video recordings

Families that contained a target infant between 16–24 months were video recorded for 60 minutes in their homes in natural interaction by the last author who has worked in the villages since 2006 and is familiar to each of the participating families. The first cohort of recordings was obtained in 2007–2008 [16] and the second cohort in 2013–2014. In total, 36 recordings of 29 infants were analyzed; 21 recordings in the first cohort (mean age = 20.88 months; SD = 2.75; 50% female) and 15 in the second (mean age = 18.17 months; SD = 1.35; 36% female). In both cohorts, an experimenter followed target infants wherever they went (see [43] for details). Family members were instructed to act as they would have had the experimenter not been present and were told that the infant was permitted to go anywhere that they normally would be permitted to go. Recordings varied in their location (inside and outside) depending on the actions of the infant and family members. Infants and family members typically engaged in activities like exploration outside, visits to neighboring households, eating, playing, grooming, etc. It was not uncommon for persons outside the nuclear family to appear in the video recordings (nearby extended family, shopkeepers etc.).

7 infants from the first cohort had more than one video. However, the recordings for those infants with multiple videos were never less than 6 months apart. Since we used hierarchical models (see Statistical Approach), we are confident that including more than one video per infant has not led to biases in inference. One female head of household had infants recorded in both cohorts.

Transcription and coding of naturalistic recordings

All audible speech from the video recordings was transcribed by local bilingual Yucatec Maya-Spanish speakers (who personally knew the children and their families) and divided into utterances, using the same criteria as Shneidman and Goldin-Meadow [16].

Each utterance was classified by the first and last authors on the basis of:

Who was speaking, primary caregiver (mother), other adult (over 11 years), or other children.

Whether it was directed to the infant or overheard by the infant. Speech was considered directed if it was addressed to the infant alone or if it was addressed to a group of individuals that included the infant. All other speech was categorized as overheard. Several cues were used together in order to categorize utterances as directed or overheard: Gaze direction, grammatical marking, utterance content, and proximity to infant.

The language in which the utterance was delivered. Utterances were classified as Spanish, Maya, or Code-Switch if speakers switched between languages mid-utterance. Because many object words (e.g., coche), nicknames (e.g., Gordo), kinship terms (e.g., tía), numbers greater than three (e.g., cinco), and calendar terms (e.g., Enero) are borrowed from Spanish even by monolingual Maya speakers, utterances containing these loan words (either embedded in a Maya utterance or alone) were classified as Maya (and not as a Code-switch or Spanish utterance). Only if speakers chose to use a Spanish word when there was a commonly used Maya alternative were those words and utterances classified as Spanish (for example if a speaker said “nariz” for “nose” instead of the Maya word “ni”).

The classification of all utterances according to the criteria a–c above was done from the transcriptions and videos together (so the identity of interlocutors and whether the interactions were directed or overheard could be checked) by the first and last authors.

Interview data

126 adults (female = 83; mean age = 35.81, SD = 14.87), including all but 5 of the mothers of the recorded infants in both cohorts, were interviewed in 2019. During the interviews, adults were asked questions regarding their social, economic, and linguistic profiles, as well as those of their family members, including all their infants. Interviewees could choose whether the questionnaire was administered to them in Spanish or Yucatec Maya. In the former case, the first author conducted the interviews; in the latter case, a local research assistant asked the questions from a previously verified translation. Both the first author and the research assistant were present during all interviews. See S1 Text for a copy of the questionnaire used and the coding of the responses.

All socioeconomic and demographic variables concerning the recorded infants and their families were obtained from these interviews. To assess the potential causal pathways driving any changes in linguistic socialization at the time of recording, for these analyses, we asked caregivers to answer in retrospective about their family situation (number of children, wage labour status, etc.) at the time of recording. Since some of the mothers’ answers concerned the wage labor status of their husbands, we verified them by independently asking their husbands the exact same set of questions. When possible, their answers were further validated with basic demographic records collected at the time of recording.

Informed consent

Informed, written consent was obtained from all adult participants included in the study and from all parents of the recorded children.

Statistical approach

Bayesian inference was used for all statistical analyses. In a Bayesian framework, each model conditions its data on prior probability distributions and uses Monte Carlo sampling methods to generate posterior distributions for its parameters. The priors are the initial probabilities for each possible value of each parameter. Regularizing priors were adopted, which are more conservative than the implied flat priors of non-Bayesian procedures, in order to prevent the model from overfitting the data given the limited sample size [44]. Having fit alternative parameterizations for all models, we believe that the results presented below are qualitatively robust to changes in priors.

Even if the four villages were very similar, they differed in the extent to which they were represented. Random intercepts for “village” were included in all models to account for the nested structure of the data and associated clustering [44]. Before conducting our analyses, we checked for multicollinearity among predictors using the generalized variance inflation factor (GVIF). All GVIF values fell below the lowest commonly recommended threshold of 4, indicating that our models should not suffer from multicollinearity [45]. Our use of regularizing priors should also reduce variance inflation [46].

Parameter estimation was achieved with RStan [47], running three Hamiltonian Monte Carlo chains in parallel until convergence was suggested by a high effective number of samples and R^ estimates of 1.00 [44]. This entailed in some cases 5000 samples per chain and in others 10000. In the former case we used 1000 as warm-up and in the latter 2000. We also visually inspected trace plots of the chains to ensure that they converged to the same target distributions and compared the posterior predictions to the raw data to ensure that the models corresponded to descriptive summaries of the samples.

For model comparisons, we used Widely Applicable Information Criteria (WAIC) which provides an approximation of the out-of-sample deviance that converges to the leave-one-out cross-validation approximation in a large sample [48]. Analyses were performed in R 3.5.2 using the brms package [49,50]. We present a complete description and justification of the priors, model specifications, model comparisons and model coefficients in S1 Text.

Ethical approval

All procedures involved in this study were reviewed and approved by the Ethics Commission of the University of Cambridge as well as the Ethics Committee of the University of Chicago. They are also in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Spanish level of the primary caregivers of the 14 infants recorded in Cohort 1 (left) and the 15 infants recorded in Cohort 2 (right).

(DOCX)

Click here for additional data file.

Marginal posterior distribution of changes in the proportion of utterances in Spanish directed to infants (left), overheard by infants (middle) and received by infants overall (right) taking into consideration variation across the different villages.

The solid lines are posterior means and the shaded regions are 80% HPDIs.

(DOCX)

Click here for additional data file.

Posterior predictive distribution of the mean difference in number of utterances of each type received by the average child from 2007 to 2013, as obtained from the Zero-Inflated Poisson model including “Cohort” as predictor variable and the number of utterances of each type as response variable but excluding those infants from cohort 1 with ages of 23 or 24 months.

These were obtained by averaging from 12000 samples from the posterior distribution (setting the standard deviations for the varying intercepts to 0). From top-left to bottom-right: Directed input from primary caregiver, directed input from adults, directed input from children.

(DOCX)

Click here for additional data file.

Descriptive statistics of the primary caregivers of the 14 infants recorded in Cohort 1 (left) and the 15 infants recorded in Cohort 2 (right).

(DOCX)

Click here for additional data file.

Descriptive statistics of infants recorded in Cohort 1 (left) and Cohort 2 (right).

(DOCX)

Click here for additional data file.

Descriptive statistics of all female adults interviewed in 2019 (n = 83).

(DOCX)

Click here for additional data file.

Descriptive statistics of all male adults interviewed in 2019 (n = 46).

(DOCX)

Click here for additional data file.

Nature of wage work of the adult males (n = 28) that participated in wage labour.

(DOCX)

Click here for additional data file.

Comparison of Zero-Inflated Poisson models predicting whether the number of utterances heard by infants in one hour had changed across Cohorts with equivalent Poisson models.

(DOCX)

Click here for additional data file.

Posterior predictive mean number of utterances of each type received by the average child from the Zero-Inflated Poisson model including “Cohort” as predictor variable and the number of utterances of each type as response variable.

These were obtained by averaging from 12000 samples from the posterior distribution (setting the standard deviations for the varying intercepts to 0).

(DOCX)

Click here for additional data file.

Posterior predictive distributions for the changes in proportion of input in Spanish across cohorts for a child from an average village (in this context, average refers to setting the estimates of the standard deviations for the varying intercepts to zero).

“ll” refers to “lower limit” and “ul” to upper limit.

(DOCX)

Click here for additional data file.

Comparison of Poisson models predicting the number of directed utterances received by infants in one hour.

(DOCX)

Click here for additional data file.

Self-reported beliefs about Spanish acquisition by the 94 adults that were fluent Spanish speakers.

The fact that the total is greater than 66 is because some women gave more than one answer to the question.

(DOCX)

Click here for additional data file.

Beliefs by primary caregivers concerning the ways in which infants learnt the different languages.

Counts are the number of females that gave a specific answer to the question of whether they thought infants could acquire competences in Yucatec Maya in the same way as those in Spanish.

(DOCX)

Click here for additional data file.

Raw averages across cohorts of the number of utterances per input type, both overall (rows 1–6) and in Spanish (rows 7–12).

(DOCX)

Click here for additional data file.

Posterior predictive distributions from the Zero Inflated Poisson models assessing changes in the number of utterances received by infants from their primary caregiver across cohorts and as a function of the number of utterances produced for a child from an average village (in this context, average refers to setting the estimates of the standard deviations for the varying intercepts to zero).

“ll” refers to “lower limit” and “ul” to upper limit.

(DOCX)

Click here for additional data file.

Anonymised data.

(CSV)

Click here for additional data file.

(DOCX)

Click here for additional data file.

(DOCX)

Click here for additional data file.

21 Jan 2021

PONE-D-20-39416

Why local languages disappear: a study of changing language input following market integration in a Yucatec Mayan community

PLOS ONE

Dear Dr. Padilla-Iglesias,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Mar 07 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Marcela de Lourdes Peña Garay, Ph.D

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Your ethics statement should only appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please move it to the Methods section and delete it from any other section. Please ensure that your ethics statement is included in your manuscript, as the ethics statement entered into the online submission form will not be published alongside your manuscript.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This article reports changes in infant-directed speech that occurred between 2007 and 2013 in four Yucatec Mayan villages with populations between 400 and 600 inhabitants. The villages are close enough to the major urban centers of Cancun and Playa del Carmen to make it practical for men to travel to these towns for employment. The Yucatan Peninsula has seen major development in the international and national tourist economy for many decades, which attracts more villagers to these urban centers. This study offers a unique portrait of the impact that these changes have made to the language environment of Yucatec Mayan infants and thus to the forces responsible for language shift in the Yucatan.

The paper reports important research that contributes unique data to the study of language shift. Nevertheless, there are several issues the authors should address before their paper is published.

The title of the study suggests a causal connection between market integration and language change, and the first sentence of the Discussion (p. 8) states “Our results show, for the first time, that the quantity, type, source, and language given to Yucatec Mayan infants changed as the community became more connected with urban centers.” There are two difficulties with this claim. First, the paper does not supply details about the market integration taking place in the region between 2007 and 2013. Were there specific developments that took place or do the changes in input reflect prior changes or other factors? Second, the paper does not report any effects that changes in the input might have had on the children’s language production. Did the children continue acquiring Maya?

The main finding of the study is that “the amount of directed speech children heard in either language, particularly from other children, decreased over the 6-year period” (p. 5). This result is at odds with explanations of language shift in that the Yucatec communities aren’t shifting languages so much as refraining from talking to their children. The authors do not offer an explanation for the reduction in speech to children. Does the reduced speech to children reflect the forces of market integration or the effect of other changes taking place in the Yucatan?

Restricting attention to the speech directed to the children, the authors report an increase in the proportion of Spanish utterances directed to the children (p. 5). The overheard speech did not exhibit a similar shift, and thus the result leads to the question of what factor would explain this shift? The authors phrase the issue (p. 6) as one of decreased use of Maya rather than an overall decrease in their speech to infants. This perspective is at odds with the statement (p. 5) that the parents also decreased their use of Spanish.

The ethnographic description of the households and villages is too thin to assess the impact of other factors and thus evaluate the significance of the four factors identified in the study. One omitted factor is the introduction of satellite tv to Mexican villages. Many remote Mexican villages now have access to cable tv, and households with cable tv watch it much of the day, but especially in the evenings. Television dissolves the social fabric by subtracting from the time that families previously spent gossiping in the local language with friends and family. The paper should state the number of homes that had tv in the two cohorts.

The introduction of cable tv offers a direct tie to the forces of market integration in that the men who work in the tourist sector would see television as a marker of social status and want to take television back to their own village. Men with cable tvs would have more status than men who just worked in the milpa and so television would take over the village. This hypothesis recognizes a pathway for importing urban behavior to the home village leading to a shift in culture that underpins the shift in language. A factor like “the general devaluing of the native language” does not identify the source of such a devaluing and how it becomes general. The paper does not identify a pathway that ties language devaluation to market integration.

The paper mentions in passing a more sinister factor in the absence of Yucatec teachers and language materials in the village schools (p. 10; Supplementary text, p. 2). The possibility has to be admitted that the absence of such support is a deliberate policy on the part of the national government to suppress the use of local languages. Many indigenous teachers are sent to teach in urban areas, while teachers from urban areas are sent to teach in rural schools. Teachers from outside the village are more likely to model intolerant attitudes towards the local “dialects”. It would be helpful if the paper supplied some information about where the village teachers come from and their attitudes towards Yucatec.

The authors should clarify the following points:

1. Participants: The paper should state the number of participants, if any, who were included in both cohorts. Including the same families in both studies might lead to an observer effect and possibly depress the number of utterances they produced. How many of the original families participated in the interviews? The utterances were classified by speaker: “primary caregiver (mother), other adult (over 11 years), or other children” (p. 11), but the paper only provides information about the primary caregivers (S1 Table) and the target infants (p. 11; S2 Table). It should provide similar tables for the other speakers. All of the participant information should be presented in one place. In the discussion on p. 9 the paper states it is possible “that older children are focusing more on schoolwork and less on sibling care.” This should be obvious from the number of older children present in the two cohorts.

2. Language samples: The families were “recorded for 60 minutes in their homes in natural interaction” (p. 11). The authors should add more information about these recording sessions such as whether they were made inside or outside the house, and the proportion of time spent looking at books. Changes in the use of books between the two cohorts might account for the changes in the input.

3. Maya versus Spanish: “Utterances were classified as Spanish only if the entire utterance was in Spanish. Utterances where code-switching between languages occurred were not included in this category and neither were utterances in which Spanish words were used as nicknames” (p. 11). It is unclear whether utterances with code-switching were excluded entirely or coded as Yucatec. The discussion of Figure 3 (p. 8) defines the proportion of Spanish utterances as the “Number of directed utterances in Spanish out of the total number of directed utterances” so it appears that the authors coded the utterances as either entirely Spanish or other. This interpretation is at odds with the statement “while infants in cohort 1 heard an average of 397.3 Mayan utterances per hour in directed speech, infants in cohort 2 children heard only an average of 92.51 Mayan utterances” (p. 6). On p. 5 the paper states “To examine whether the language of input shifted from Yucatec Mayan to Spanish over the 6-year period, we fitted Bayesian logistic mixed models to the proportion of input in Spanish (vs. Mayan) that infants received in child directed and overheard speech.” The paper should clarify what was counted as a Mayan utterance, and address the question of whether the code-switched utterances were excluded from both the Mayan and Spanish counts.

Without knowing how the proportion of code-switched utterances changed between the two cohorts, it’s impossible to tell if the number of Spanish utterances changed significantly. The paper should report the number of code-switched utterances. The code-switched utterances should have been analyzed for the number of utterances with verbs in Spanish or Yucatec. In fact, the proportion of utterances containing Yucatec or Spanish verbs would be a more sensitive measure of language shift than the proportion of Spanish utterances because it would obviate the need to decide whether or not individual words were Spanish. An increase in the number of utterances consisting entirely of Spanish nouns and noun phrases (un carro, chicle, galleta) would not suggest a significant shift in language.

4. Villages: The paper states that the recordings were made in four villages (p. 10), and that the villages were very similar (p. 12). The Supplementary text adds information about the populations (p. 2). All information about the villages should be presented in one place in the paper. The paper should provide more details about the relative levels of roads and schools in each village, the proportion of the men who participated in wage labor from each village, and the number of participants from each village in each cohort. These details would help readers understand the degree to which the changes took place regionally rather than principally in one or two villages.

This is a solid investigation and I encourage the authors to add further details to their paper.

Reviewer #2: The paper examines factors involved in language shift by analyzing interview data, family histories, and transcripts of natural data (to assess input) over a 6-year period in 4 Yucatec Mayan villages where language shift to Spanish is in process. The analysis supports the central argument, that Maya language input has decreased due to social changes that are not explicitly related to language but rather to other social changes (e.g. increased numbers of Mayans working for wages in Spanish-speaking cities, decrease in the availability of older siblings to care for young children because they now attend formal schooling at higher grades than previously) which directly affect language input. The paper is clearly researched and argued; the data are clearly presented and all analyses explained thoroughly and well-documented. The supplementary materials provide robust data supporting changes in education, in the caretakers, and in the language used by them with the children, and in the fathers' earning sources. (I did not see data supporting an increase in market goods or directly linking an increase in wage labor to Spanish, maybe I missed it.)

The finding that Mayan communities find Spanish more practical than Yucatec Maya is not surprising or novel, given similar findings in most shifting ecologies; the novelty is perhaps a bit overstated in the paper. The significant contribution of this paper is the very rigorous way it has determined the attitudes and beliefs of the speaker communities and the actual input the children, in both directed and overheard speech. It is also provides a model for future work in other communities to understand the causes of shift.

Minor revisions:

The relevant social and demographic changes discussed here are encapsulated by the authors with the term market integration, and I would recommend that the authors define what they mean by market integration in this paper early on, as it is used in a non-standard way, OR change the terminology to specify increased contact with urban centers (which is specified on p. 8) OR increased education/wage labor/market goods (p. 4). The discussion of the relevant urban contact is provided under Materials & Methods on p. 16 of the manuscript. A sentence definining what is meant in the beginning of the paper would be welcome.

I would further note that it is odd to think about periods of market integration:

"during periods of market integration, both parents and older siblings of young children in small-scale communities are more likely than previous generations to have had access to formal schooling provided by the majority community" (p. 2).

Market integration is not usually conceptualised as a transitory state. Rather, markets usually start out not integrated and then improvements in road conditions or whatever barriers there were to integration would make them better integrated. So here my sense is that the intent of the authors is a before/after comparison, and the wording could be tweaked.

Reviewer #3: see attachment--uploaded attachment would exceed character count, I think;

The paper presents one very significant finding, based on a comparison of two different cohorts of Maya Yucatec speakers in two different periods of language-acquisition studies separated by 6 years. I found this main empirical finding striking and important (although not without some questions). I think the work should be published if only to disseminate this well founded and potentially useful result. It is, in brief, that over the very short 6 year span, there was significant change in the sort of direct language input infant Yucatec Maya language learners received. It went from being largely in Maya to almost entirely in Spanish (with some empirical caveats I will shortly mention).

Reviewer #4: [see attachment] [see attachment][see attachment][see attachment][see attachment][see attachment][see attachment][see attachment][see attachment][see attachment][see attachment][see attachment] - Sorry it wouldn't let me submit without 200 characters here!

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Clifton Pye

Reviewer #2: No

Reviewer #3: Yes: John B. Haviland

Reviewer #4: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PadillaetalReview.pdf

Click here for additional data file.

Submitted filename: Padilla-Iglesias_PLOS_SM.pdf

Click here for additional data file.

Submitted filename: PLoS One Review.pdf

Click here for additional data file.

Submitted filename: PONE-D-20-39416_reviewer.pdf

Click here for additional data file.

24 Feb 2021

Please see the attached "Response to reviewers.docx" document.

Submitted filename: Response to reviewers.docx

Click here for additional data file.

6 Apr 2021

PONE-D-20-39416R1

Changing language input following market integration in a Yucatec Mayan community

PLOS ONE

Dear Dr. Padilla-Iglesias,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by May 21 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Marcela de Lourdes Peña Garay, Ph.D

Academic Editor

PLOS ONE

Additional Editor Comments (if provided):

All reviewers and I find the subject presented in this manuscript very interesting and original. However, there are still important issues that must be clarified to advance with the publication consideration in PLOS ONE.

I agree with reviewer 4 that providing more data supporting the statement that socioeconomic change in the analyzed periods, disentangling the role of market integration and secondary schooling in the linguistics changes, and clarifying the concept of language extinction based on current models, would confer the manuscript greater impact.

In addition to te comments of reviewer 4, the suggestion of reviewer 3 about to tone down or rephrase the claims made on the basis of the survey data, may benefit the clarity of the ms.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: All comments have been addressed

Reviewer #3: All comments have been addressed

Reviewer #4: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Partly

Reviewer #4: Partly

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: Yes

Reviewer #3: I Don't Know

Reviewer #4: I Don't Know

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Manuscript #: PONE-D-20-39416.R1

Title: Changing language input following market Integration in a Yucatec Maya Community

I thank the authors for their attentive responses to my comments. They have responded well to my queries and I would now accept their paper for publication.

The study reports two key findings, but the interpretation of the results remains problematic due to the inherent complexity of sociolinguistic data. I am especially concerned about the explanation for the decrease in the number of utterances between cohorts (S7), which the authors link to changes in the villages’ market integration. The alternative is that this difference is due to some other difference between the cohorts. The authors’ responded to my initial concern about this decline by noting that:

"In fact, changes in directed input were driven (fully) by decreases in input in Yucatec Maya. Input in Spanish was very rare in the first cohort (mean of 7 utterances per hour) and increased in cohort 2 (80 utterances per hour). Thus, two changes were taking place simultaneously: a reduction in overall speech directed to children and a shift in the language that was used (from Maya to Spanish) in instances where speech was directed to children."

The description in the text should be modified to clarify the fact that the decline in input reflected decreases in Maya utterances. The authors might consider replacing table S7 with a table that just reports the number of Maya utterances in directed input across the two cohorts. I combined the numbers in S7 and S8 into a single table in order to simplify the comparison:

Number of utterances S7 Proportion of Spanish S8

mothers Other adults children mothers Other adults children

Cohort 1 99.55 36.86 346.69 10% 1% 27%

Cohort 2 76.81 37.79 165.72 29% 66% 61%

I don’t see how to reconcile the numbers in the authors’ response with the numbers in S7 and 8. For Cohort 1, 10% of 99.55 ~ = 9.9 utterances (close to 7 utts/hr.) For Cohort 2, 29% of 76.81 ~ = 25 utterances (not close to 80 utt/hr.) This confusion suggests it would help if the authors supply the number of utterances in Maya and Spanish as well as the proportions. It’s good practice to provide both numbers and proportions. My calculations suggest that the mothers in Cohort 1 produced a mean of ~90 Maya utterances, and mothers in Cohort 2 produced a mean of ~51 Maya utterances, while their Spanish utterances increased from 9.9 to 25.

These numbers still don’t account for the decrease in Maya utterances between cohorts assuming that the children’s basic interactions wouldn’t change between cohorts. This is one of the findings that as the authors state “Our results show, for the first time, that the quantity, type, source, and language input heard by Yucatec Maya infants changed as the communities became more connected with urban centers, and hence education, wage labour and market goods became more accessible to their inhabitants.” The authors link the decrease in speech from other children to a decrease in time the older children spent with the target children due to schooling (p. 9 and response to reviewer). This explanation doesn’t explain the decrease in the mother’s directed speech between cohorts.

If the target children’s behavior changed between cohorts then the target children’s behavior, not the behavior of other speakers, would account for changes in the number of utterances. One change in the target children’s behavior might reflect the difference in the children’s ages between the cohorts (Median 22 mos in Cohort 1 and 18 mos in Cohort 2, S2). These differences could correspond to differences in the mean lengths of the target children’s utterances from a one-word stage to a two-word stage. Thus, differences in the mothers’ CDS between the cohorts might reflect differences in the target children’s language production rather than changes in market integration. Or both factors might be at work.

In their response to my previous question about the children’s language production, the authors state that they “did not quantitatively assess children’s proficiency on each of the languages…” This observation makes sense from the perspective that CDS is the primary driver of the children’s language development, but ignores the point I raise here that the children’s language production might account for the decrease in the mothers’ directed speech in Maya. It would be a simple matter to calculate the target children’s mean length of Maya and Spanish utterances in words across the two cohorts to test this hypothesis. An alternative would be to acknowledge the target children’s language as a possible explanation for the decrease in speech between cohorts. It would help to add the authors’ new observations about the children’s current use of Maya and Spanish in order to add a valuable perspective on the limited effects of market integration. It might yet prove to be the case that despite changes in the economy, the villages will persist in their use of Maya.

Reviewer #2: not applicable. I have no comments to make to the author and am annoyed that I need to fill this box in, having already indicated above that the changes addressed my concerns.

Reviewer #3: One of my main suggestions, namely to tone down or background the claims made on the basis of the survey data, the authors chose to ignore, and it is thus difficult for me to reassess the paper given my reservations about its original design, which remains largely unchanged.

Reviewer #4: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Clifton Pye

Reviewer #2: No

Reviewer #3: No

Reviewer #4: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PadillaetalReview-1stRevision.pdf

Click here for additional data file.

Submitted filename: PloS One take 2.pdf

Click here for additional data file.

14 May 2021

Please see attached letter.

Submitted filename: Response_to_reviewers.docx

Click here for additional data file.

26 May 2021

Changing language input following market integration in a Yucatec Mayan community

PONE-D-20-39416R2

Dear Dr. Padilla-Iglesias,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Marcela de Lourdes Peña Garay, Ph.D

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

The authors rewrite the manuscript to answer all the concens made by the reviewers.

In the current state the manuscript provides new data and may inspire further discussion and research.

Reviewers' comments:

2 Jun 2021

PONE-D-20-39416R2

Changing language input following market integration in a Yucatec Mayan community

Dear Dr. Padilla-Iglesias:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Marcela de Lourdes Peña Garay

Academic Editor

PLOS ONE