Languages form words by combining the sounds of the language. In many languages, however, certain combinations of sounds systematically do not occur, contrary to what is expected if the sounds of a language combined by chance. In Bolivian Quechua certain sounds may not co-occur in a root. While there are roots in Bolivian Quechua with a single ejective or aspirated sound, there are not roots with two ejectives or aspirates (for example, "k'ap'a", "thaqha" would be impossible roots). This project investigates native speakers' awareness of such combinatorial restrictions on laryngeal features in the lexicon of Bolivian Quechua. The goal of the research is to determine whether speakers of Bolivian Quechua are aware of these particular restrictions and, moreover, whether the systematic restrictions have a different status than "accidental gaps" in the lexicon. These questions are investigated by asking native speakers for acceptability ratings of nonce words (made up words). The experiments will contribute to an informed typology of co-occurrence restrictions and inform two broad questions in phonological research. First, the study will ascertain whether laryngeal restrictions have a synchronic status in the grammar. Some co-occurrence restrictions, and particularly laryngeal co-occurrence restrictions, are static generalizations about the shape of roots in a language and may not require a synchronic analysis. Previous studies of the psychological reality of root restrictions have not tested the kinds of laryngeal restrictions found in Quechua. Second, the comparison of laryngeal restrictions with accidental gaps will reveal whether speakers form generalizations about classes of segments in their language, or learn the attestation of individual combinations of segments. An additional component of the study is documentation and acoustic analysis of the laryngeal system of Bolivian Quechua, which contrasts voiceless unaspirated, aspirated and ejective stops. Ejectives are both relatively rare and understudied.
This study will contribute to the Quechua community by further documenting the phonetic properties of the sounds of Quechua. There are many aspects of the Quechua language that are rare among the world's languages, and differ from Spanish, the second language of most Quechua speakers. This project is linguistically and culturally important in contributing a detailed description of the acoustics of laryngeal contrasts in Quechua. No comprehensive phonetic description is available for Quechua. Accurate and thorough phonetic description is valuable both for future generations of Quechua speakers who may face language loss and endangerment and for any linguist who seeks to understand the phonology of this language.
In many varieties of Quechua, a family of languages spoken through the Andes mountains, words may not contain certain combinations of sounds. The current project investigated two restrictions on sound combinations within words, both of which restrict the occurrence of a set of sounds in the language called ejectives. Ejectives may occur in initial position in a word (e.g., p'acha 'clothes'), or in medial position of a word (e.g., mat'i 'forehead'), but a word may not contain two ejectives or an ejective that is preceded anywhere in the word by a non-ejective stop (e.g., p, t, k): there are no words like *k'ap'u or *kap'u. This project investigated whether Quechua speakers have learned these restrictions and feel that words with two ejectives or an ejective preceded by a non-ejective stop sound odd as words of Quechua. Participants in the study listened to words through headphones and were asked to repeat what they heard. The words were a combination of real words of Quechu, nonsense words that didn't violate a restriction of the language, and nonsense words that did violate a restriction in the language. All words in the study had medial ejectives. Accurate repetition of the medial ejective was compared between real words (e.g. [wirp’a] ‘lip’), well-formed nonsense words that are not ruled out by any restriction (e.g. [jump’a]), and ill-formed words with either two ejectives (e.g., [k'ap'u]) or a non-ejective stop followed by an ejective (e.g, [kap'u]). There were 15 words in each of the four categories, for a total of 60 tokens per speaker. 8 native speakers completed the experiment. Participants were more likely to make errors in producing a medial ejective in systematically absent words than in accidentally absent or real words. In addition, there is a small difference between the two systematically absent types of words: there are more errors on words that violate the restriction on roots with two ejectives (1a) than those that violate the restriction on a plain stop followed by an ejective (1b). The graph in Figure 1 shows the average rate of correct repetition on the four categories of words: "real" = real words, "gap" = accidentally absent, "ordering" = violates (1b), "cooccurrence" = violates (1a).
