Language and ethnobiological skills decline precipitously in Papua New Guinea, the world's most linguistically diverse nation

Abstract

Papua New Guinea is home to >10% of the world's languages and rich and varied biocultural knowledge, but the future of this diversity remains unclear. We measured language skills of 6,190 students speaking 392 languages (5.5% of the global total) and modeled their future trends using individual-level variables characterizing family language use, socioeconomic conditions, students' skills, and language traits. This approach showed that only 58% of the students, compared to 91% of their parents, were fluent in indigenous languages, while the trends in key drivers of language skills (language use at home, proportion of mixed-language families, urbanization, students' traditional skills) predicted accelerating decline of fluency to an estimated 26% in the next generation of students. Ethnobiological knowledge declined in close parallel with language skills. Varied medicinal plant uses known to the students speaking indigenous languages are replaced by a few, mostly nonnative species for the students speaking English or Tok Pisin, the national lingua franca. Most (88%) students want to teach indigenous language to their children. While crucial for keeping languages alive, this intention faces powerful external pressures as key factors (education, cash economy, road networks, and urbanization) associated with language attrition are valued in contemporary society.

Keywords: Papua New Guinea; biocultural diversity; ethnobiology; language attrition; language endangerment.

Fig. 1.

Languages studied in Papua New Guinea. (A) Language map (1) with the number of students surveyed. (B) Survey of 486 students speaking 37 indigenous languages at the Mt. Hagen Secondary School.

Fig. 2.

Indigenous language skills in present and future PNG populations. (A) Language skills (L2) of 6,190 students (female and male) and their parents. (B) Mean language skills (L2) for 110 well-sampled languages (n ≥ 10 students per language). (C) The proportion of fluent speakers among parents and students extrapolated to the entire 18- to 20-y-old cohort in PNG (Youths) and to the next generation of students (Next gen S) and all 18- to 20-y-olds (Next gen Y). (D) Language skills (L1) of the students, predictions from models characterizing the 18- to 20-y-olds in PNG at present (Youths) and in 30 y (PNG 30), and language skills assuming that PNG will come to match the mean socioeconomic parameters of lower-middle income countries (PNG LMI). Language skills were quantified as the number of body parts (from the total of 24) correctly named from photographs (L1) or by assessment by respondents for themselves and their parents on a four-point scale: no language skills (0), passive understanding (1), speaking but poorly (2), or fluent use (3) (L2).

Fig. 3.

Effects of language and socioeconomic factors on indigenous language skills and ethnobiological knowledge. GLMMs describe variability in student language skills (L1) and in knowledge of bird species (E1) and traditional plant uses (E2). The language-skills model incorporated 11 fixed variables divided into four classes: (A) language traits, endangerment (A3) and geographic regions (A4); (B) socioeconomic traits, birthplace urbanization (B1) and remoteness (B2) and parents’ education (B3) and employment (B4); (C) family language use, parents’ language skills (C1), home language use (C2), and whether parents speak the same first language (C3); and (D) student traits: traditional skills (hunting, fishing, growing food, house building, and medicinal plants) (D4) and contemporary technical skills (mobile phone and computer use) (D5). The variables were selected within each class (SI Appendix, Table S2) before being included in a global model (SI Appendix, Table S3). The bars show the AIC improvement due to the addition of each group (black) and each variable within each group into a model that includes all other variables, quantifying the marginal effect of each class/variable. The line plots show the shape of the effect of each variable across its range (except categorical A4) while keeping the other variables constant. Only significant (P < 0.05) variables are shown. The models describing variability in student knowledge of bird species (E1) and traditional plant uses (E2) used language skills (L1) and three classes of explanatory variables (family language use, socioeconomic traits, and student traits, including D1—gender) (SI Appendix, Tables S5 and S6). L1 is defined in Fig. 2 and other variables in Materials and Methods.

Fig. 4.

Language skills and ethnobiological knowledge. (A) Indigenous plant use by categories (Hill’s diversity ¹D = 4.15 for indigenous and 2.18 for Tok Pisin/English uses). (B) The 10 most common plant species listed in Tok Pisin/English medicinal uses. The respondents were asked to freely list up to 10 plant species with their indigenous names and traditional uses (E2). They provided 21,829 responses in indigenous languages (i.e., 35% of the maximum of 10 uses × 6,190 respondents) and an additional 5,458 responses in Tok Pisin/English when they could not name any plant in an indigenous language.