Accidental gap

Abstract: It has long been noted that phonological patterning is influenced by phonetic factors. But phonologists diverge on whether phonetic motivations take effect in synchronic or diachronic phonology. This article aims to tease apart the two theories by investigating native Mandarin speakers’ applications of two tone sandhi processes to novel words: the phonetically motivated contour reduction 21321/__T (T� 213) and the neutralizing 21335/__213 whose phonetic motivations are less clear. Twenty Mandarin subjects were asked to produce two monosyllables they heard as disyllabic words. Five groups of disyllabic words were tested: AO-AO (AO=actual occurring morpheme) where the disyllable is also a real word, AO-AO’ where the disyllable is nonoccurring, AO-AG (AG=accidental gap in Mandarin lexicon — legal syllable and tone but non-existent combination), AG-AO, and AG-AG. The first syllable is always 213, and the second syllable has one of the four tones in Mandarin. Results show that speakers apply the phonetically more natural 21321 sandhi more quickly and with greater accuracy than the 21335 sandhi. Theoretically, the study supports the direct relevance of phonetics to synchronic phonology by showing that there is a psychological advantage to phonetically natural patterns. Methodologically, it complements existing research paradigms that test the nature of the phonology-phonetics relationship, e.g., the study of phonological acquisition and the artificial language paradigm; when extended to other Chinese dialects, it can also provide insights into the long-standing mystery of how Chinese speakers internalise complicated tone sandhi patterns that sometimes involve opacity, near-neutralization, and syntactic dependency.

Abstract: Linguistic and Non-Linguistic Influences on Learning Biases for Vowel Harmony Sara Finley (sfinley@bcs.rochester.edu) Department of Brain and Cognitive Sciences, Meliora Hall Rochester, NY 14627 USA William Badecker (badecker@arizona.edu) Cognitive Science, Communications Building Tucson, AZ 85702 USA Abstract /– + +/. There are two possible harmonic outputs: [– – –], which changes the feature value of two of the input vowels, and [+++] which changes only one of the vowels in the input. A left-to-right spreading language chooses [– – –] even though two vowels change. Another possibility is to have no intrinsic direction of spreading, but to choose the harmonic output with the fewest changes from the input (in this case [+++]). This type of spreading is termed ‘majority rules’ because the direction of spreading is determined by the majority feature value of the input (Bakovic, 2000). One peculiarity is that while languages never use ‘majority rules’ to determine the direction of spreading, ‘majority rules’ grammars are extremely easy to produce in generative phonology 1 . Generative linguistics assumes that the non- existence of patterns in natural language implies that they should not be generated by the grammar. However, it is possible that the lack of ‘majority rules’ grammars is due to an accidental gap. Under this assumption, ‘majority rules’ patterns are grammatically plausible, but the lack of such languages is an accident of history and language sampling. One way of distinguishing between a principled restriction on the nature of vowel harmony languages and an accidental gap account is through testing for learning biases. If learners are biased against ‘majority rules’ languages and biased towards a directional harmony pattern, it suggests that the non-existence of ‘majority rules’ languages is a valid restriction on grammar. Because it is impossible to test learning biases for unattested languages in a naturalistic setting, as there are no naturalistic settings where a ‘majority rules’ grammar might be present, the artificial grammar learning paradigm is the best method for addressing this question. In an artificial grammar learning paradigm, it is possible to manipulate naturalness, complexity and statistical regularities in a way that is impossible with naturalistic studies of language learning. The present experiments test whether learners make use of the ‘majority rules’ strategy when making grammaticality judgments between harmonic items. We present an experimental paradigm in which learners are exposed to a harmony language that is ambiguous between directionality and ‘majority rules’. If learners are biased towards directional patterns and against ‘majority rules’ patterns, This paper addresses the question of the domain-specificity of learning biases for phonological processes. In two artificial grammar learning experiments we explore the role of learning biases in shaping the distribution of phonological patterns across the world’s languages. In Experiment 1, we demonstrate that learners are biased toward phonological patterns that occur in natural language, as opposed to patterns that are not found across the world’s languages. Specifically, learners are biased towards directional vowel harmony spreading processes. In Experiment 2, we exposed learners to a non-linguistic analogue to vowel harmony. Learners processed spreading such that learners favored the cross-linguistically valid pattern only when the first item of the series underwent spreading. This set of similarities and differences in learning may provide some insight into the origin of learning biases for spoken languages. Keywords: artificial grammar learning; phonology. Introduction The experiments presented in this paper address the hypothesis that typological restrictions on languages are due to learning biases (Slobin, 1973). Specifically, we address the distribution of vowel harmony across the world’s languages. Vowel harmony is a phonological process that induces statistical tendencies for words to share the same vowel quality along a particular phonetic dimension. In Turkish, which displays harmony for both backness and rounding, if the first vowel of the word is front and unround (with some exceptions), all following vowels must be both front and unround as well (Clements & Sezer, 1982). Thus, Turkish vowel harmony may be thought of as a directional spreading process in which the leftmost vowel spreads its feature (round, back) to the right. Vowel harmony languages exhibit both left-to-right and right-to-left spreading characteristics. The direction of spreading can be decided by the morphology of the language (stems are more likely to spread harmony than affixes (Bakovic, 2000)) as well as the characteristics of the input vowels (spreading [+Round] is more likely than spreading [–Round] (Korn, 1969)). The direction of spreading can also be set such that spreading always occurs from right to left or from left to right. One way in which the direction of spreading is never decided is by the number of changes from the input to the output of the phonological process. For example, consider the disharmonic input In ‘majority rules’ grammars, “ties” (e.g., two round and two unround vowels) are decided by a default strategy (lower-ranked constraint).

Abstract: Error patterns in children’s phonological development are often described as simplifying processes that can interact with one another with different consequences. Some interactions limit the applicability of an error pattern, and others extend it to more words. Theories predict that error patterns interact to their full potential. While specific interactions have been documented for certain pairs of processes, no developmental study has shown that the range of typologically predicted interactions occurs for those processes. To determine whether this anomaly is an accidental gap or a systematic peculiarity of particular error patterns, two commonly occurring processes were considered, namely Deaffrication and Consonant Harmony. Results are reported from a cross-sectional and longitudinal study of 12 children (age 3;0 – 5;0) with functional phonological delays. Three interaction types were attested to varying degrees. The longitudinal results further instantiated the typology and revealed a characteristic trajectory of change. Implications of these findings are explored.

Abstract: This paper is an examination of the tone sandhi phenomenon in Ningponese. Although there is a considerable disagreement on the exact tonal value of a citation form in the dialect, the consistency of sandhi form pronunciation is remarkable. Dissyllabic words of definite sandhi patterning are summarized in a chart according to Ancient Chinese tone categories. The non-occurrence of the combinatorily possible low level tone in the second syllable is taken to be significant, not likely to be a case of an accidental gap. O. Introduction In general, when we speak of Ningponese as a dialect different from other Wu dialects, we could mean two things, it denotes (1) the language spoken in the area known as Ningpo Prefecture (including Zhoushan) 宁波府(包括舟山) in the old days; (2) the language spoken within the City of Ningpo 宁波 and Yinxian 鄞县. We will use the term in the second sense. Although Wu dialects have been studied for more than one hundred years, there have been very few studies on Ningponese dialect either in Chinese or in English, Von Mollendorff's 1901 Ningpo Syllabary offered extensive information on the initials and the finals of the dialect, but because of limitations in the contemporary linguistic techniques, his records were deficient in detailed descriptions of tones. Chao 赵元任 (1928) included Ningponese in his Studies in the Modern Wu Dialects. Again, tones were not the main interest of the studies. In 1975, P. J. Rose at the Darwin College, Cambridge, included Ningponese in his research of tones. However, according to Michael Sherard (1979), there were no published accounts of Ningponese in the Western journals. Shi Wen-tao 施文涛 (1979), a native of Ningpo, in his effort to provide logographs for some of the syllables in Ningponese, approximated the phonetic values of the consonantal initials, finals and tones in Ningponese. My conclusions are often different from his, but rather than take issue point by point here, I shall only include his citation tones below for reference. The main body of this paper will concern itself with the tone sandhi phenomenon in Ningponese.