Vowel harmony preferences in infants growing up in multilingual Ghana: Effects of experience/exposure to languages with and without tongue root harmony

Phonotactics acquisition is a fundamental task for every language learner. Hence, this topic has received considerable attention in the literature on the acquisition of phonology (Jusczyk et al., 1993; Mattys et al., 1999; Sebastián-Gallés & Bosch, 2002). The term phonotactics refers to the rules or constraints that govern possible phoneme (i.e., speech sound) combinations, phoneme positions, and syllable structures in a language. These rules or constraints are language-specific: for example, in English, the sound [ŋ] can only occur at the end of a word (e.g., ring /rɪŋ/) but not the beginning of a word, which would, however, be possible in other languages such as Akan (e.g., /ŋkwan/'soup'). Infants must, therefore, learn these permissible sound combinations and related rules of their native language(s). The present study focuses on vowel harmony (VH). VH is a language-specific phonotactic constraint on the co-occurrence of vowels in a word or syllable. This phonotactic constraint is attested in many locally and genetically unrelated languages. VH is a key aspect of the phonotactics of most African languages. The co-occurrence constraint in VH languages is that consecutive vowels within a word must share specific phonological features. For example, in Advanced Tongue Root (ATR) harmony languages like Akan, a Kwa language, vowels are grouped into two based on their tongue root feature [+ATR] and [-ATR]. It is, therefore, required that vowels within the same word agree in terms of their ATR feature. Hence, words in Akan must (with few exceptions) contain either only +ATR vowels (e.g., /kukuo/ 'pot', /efie/ 'house') or only -ATR vowels (e.g., /adaka/ 'box', /ɛwʊɔ/ 'honey'). Some literature has been published on VH in infants (see Altan et al., 2016; Hohenberger et al., 2017; LaCross, 2015; Mintz et al., 2018; Van Kampen et al., 2008). These previous studies report that infants demonstrate language-specific phonotactic knowledge biases in processing VH information from about 6 months of age (Altan et al., 2016; Hohenberger et al., 2017; Van Kampen et al., 2008). Recent evidence suggests that infants learning a VH language show a listening preference for vowel-harmonic syllable sequences consistent with the VH rules of their native languages (Turkish: Altan et al., 2016; Hohenberger et al., 2017; Van Kampen et al., 2008; Hungarian: Gonzalez-Gomez et al., 2019), but not when their language has no VH (French: Gonzalez-Gomez et al., 2019; Germany: Van Kampen et al., 2008). This preference has been interpreted to reflect knowledge of the presence or absence of VH in their native language, thus, suggesting an effect of language-specific phonological acquisition. However, it has been proposed that the acquisition of VH, despite its language-specificity, may be facilitated by universal innate or perceptual biases (see Moreton, 2008 for an informative discussion). Following these proposals, all infants may be perhaps born with a universal (innate or perceptual) preference for harmonic sequences. However, this universal preference disappears over time if not reinforced by input from the surrounding language(s). This may also explain why German- and French-learning infants, for example, show no preference for neither harmonic nor non-harmonic syllable sequences. It could be that these infants have lost the universal bias for a VH preference at some point in their language development because of their non-exposure to VH language. An infant study that supports this possibility of a universal bias for VH comes from Mintz et al. (2018). However, in their study, English-learning infants at 7 months showed a harmony bias only when tested in a segmentation task but not in an auditory preference task. To this point, very little is known about the development of phonotactic knowledge in infants growing up with more than one language. So far, only one study by Sebastián-Gallés and Bosch (2002) has investigated 10-month-old bilingual infants' perceptual preferences of consonant clusters that were attested only in one of their native languages. As a result, infants showed a preference for attested over unattested clusters, and this effect was not modulated by language exposure. As it is a linguistic reality for many infants in the world to grow up with two or more languages that differ in their phonotactics, research on this issue is relevant to broaden our understanding of how exposure to diverse languages can affect their perceptual processing. For instance, the type and quantity of input (e.g., exposure to both VH and non-VH languages and maybe the unequal amount of input in each language) might lead to significant differences in how phonological constraints are acquired and, in effect, might influence perception. Particularly in Africa, multilingual language acquisition from birth is the norm. In light of this, the present study focuses on the acquisition of native language phonotactics for the first time. Specifically, the study focuses on VH in infants being multilingually raised in Accra, the capital of Ghana (West Africa), with minimally one language following the Advanced Tongue Root VH rules of Akan, a language that is widely spoken in Accra. This study addresses the following questions: (1) Do multilingually raised infants acquiring an ATR harmonic language in addition to non-harmonic languages show sensitivity to the harmonic and non-harmonic patterns in their languages? (2) Are infants' preferences for harmonic or non-harmonic patterns modulated by the amount of exposure in each of their native languages? Gaining insight into how exposure to diverse languages (VH and non-VH languages) shapes infants' phonotactic preferences will help us better understand how language experience and its potential learning biases influence infant's phonotactics acquisition and the extent to which amount of language exposure to VH and non-VH languages influences perceptual preferences.