Volume 4, Issue 2

It is likely that generalization of implicitly learned sound patterns to novel words and sounds is structured by a similarity metric, but how may this metric best be captured? We report on an experiment where participants were exposed to an artificial phonology, and frequency ratings were used to probe implicit abstraction of onset statistics. Non-words bearing an onset that was presented during initial exposure were subsequently rated most frequent, indicating that participants generalized onset statistics to new non-words. Participants also rated non-words with untrained onsets as somewhat frequent, indicating generalization to onsets that had not been used during the exposure phase. While generalization could be accounted for in terms of featural distance, it was insensitive to natural class structure. Generalization to untrained sounds was predicted better by models requiring prior linguistic knowledge (either traditional distinctive features or articulatory phonetic information) than by a model based on a linguistically naive measure of acoustic similarity.

While many phonological processes are local, consonant harmony is of interest phonologically because it can occur non-locally. Sibilant harmony in Navajo requires that sibilants within a word have matching anteriority specifications. The process is described as being sometimes mandatory and sometimes optional, but neither the statistical nature of the occurrence in optional settings nor the factors contributing to the optionality are fully understood. This paper provides preliminary investigation into these issues using the first person possessive morpheme, which is underlyingly /ʃi-/ but may harmonize to [si-]. Experiment 1, an online grammaticality judgment survey, reveals that the harmonized prefix is dispreferred in all environments. Experiment 2 presents acoustic data from three Navajo speakers: though none harmonize overtly, the spectral mean and lower bound of frication energy of the prefixal fricative are affected by the presence of [+anterior] sibilants in noun stems. The overall implication of these findings is that harmony is not only optional but is dispreferred or wholly absent for some speakers. While multiple factors are investigated, the only one that consistently affects harmony is adjacency of the trigger and target, indicating that, although consonant harmony may indeed be a non-local process, its occurrence is heavily mediated by distance.

Differences in rhythm between languages have been often attributed to differences in phonological properties such as syllable structure. This paper uses quantitative analyses to determine whether and how popular duration-based rhythm measures depend on the phonological structure of a language. Native speakers of five languages read a large corpus of comparable texts (approximately 371,000 syllables in total). Phonological properties of each language were specified as 11 variables, computed from the phonetic transcriptions. These variables were compared against published rhythm measures that captured variation in duration of consonantal and vocalic intervals. While the text-based measures discriminated well between languages, the values of rhythm measures overlapped substantially, showing that the languages are more alike in acoustic implementation than in their phonological description. Multilevel models demonstrated that the mapping between phonological properties and acoustics is much weaker than previously assumed: linear effects of the phonological variables explained less than a quarter of the total variance in rhythm measures. Instead, speaker was the main source of variation in those measures. Rhythm, in the sense of durational variability, depends far more on individual timing strategies than on the phonological structure of a language.

Russian normally does not have secondary stress, but it is variably realized in compounds. We examined the factors that contribute to secondary stress realization in a rating study, where listeners were asked to rate compounds pronounced without secondary stress and with secondary stress in various locations. We refine some generalizations from impressionistic descriptions: in compounds whose left-hand stems have mobile lexical stress, acceptability of secondary stress decreases with token frequency of the compound, and acceptability of pronunciations without stress increases with frequency. Ratings improve as distance between stresses increases, and this effect is gradient rather than categorical. We also identify new generalizations about secondary stress that relates to the properties of the left-hand stem. First, we identify a faithfulness effect: stress realization is optional on lexically stressed stems, but stress movement is strongly penalized. Second, we identify a sonority sequencing effect: secondary stress is not tolerated well on linker vowels in compounds, but acceptability improves significantly when the linker is the only vowel in a stem with a falling sonority cluster. Thus, the stress system distinguishes clusters with falling sonority from other types.

Adult second language (L2) learners often have difficulty perceiving and producing L2 speech segments with native-like accuracy. The Speech Learning Model (SLM; Flege 1995) attributes this to perceptual distance between native (L1) and L2 segments, the Perceptual Assimilation Model (PAM; Best 1995; Best and Tyler 2007) to L1 assimilation of L2 segmental contrasts. In a novel extension of these models, we examined whether experience with an L1 suprasegmental duration contrast can affect performance on L2 segmental vowel distinctions that are non-contrastive in the L1. Serbian employs an unusual type of relevant suprasegmental duration distinction, specifically morphological alternations in pitch-accented syllable duration. We assessed L1-Serbian participants' perception and production of L2-English /i-ɪ/ versus /æ-ɛ/, which involve both spectral and temporal distinctions but differ in magnitude of their temporal difference. The results indicate that experience with L1 syllable durations conditioned by pitch-accent systematically influences perception and production of L2 vowel contrasts.

Articulatory and acoustic studies have provided evidence that in word- initial and word-final positions, English /l/ exhibits substantial differences in ‘darkness‐: dark [ɫ] in word-final position is produced with a more retracted tongue dorsum and lowered tongue body than light [l] in word-initial position. The darkness of intervocalic /l/, however, is variable. While Sproat and Fujimura (1993) argue that /l/ darkness is on a continuum strongly affected by duration, Hayes (2000) maintains that the morphological status of intervocalic /l/s should affect whether they are produced as light or dark variants. In this study, ultrasound imaging is used to investigate whether the morphological affiliation of the /l/ affects the degree of tongue dorsum retraction and tongue body lowering and the acoustic characteristics of /l/ darkness. Six American English speakers produced three types of stimuli which were predicted to increase in darkness in the following order: (1) when /l/ corresponded with the onset of a suffix (e.g., flaw-less), (2) when /l/ corresponded with the final position of the stem word (e.g., tall-est), and (3) when /l/ was the final consonant of a stem word (e.g., tall). For both articulatory and acoustic measures, the predicted order was upheld. The strongest articulatory correlate of darkness was tongue body lowering, and acoustic differences were mainly manifested in F1 and normalized intensity. Phonological implications of these findings are discussed.

This paper offers an account for the cross-linguistic prevalence of phonological word-final vowel shortening, in the face of phonetic final lengthening, also commonly observed across languages. Two contributing factors are hypothesized: (1) an overlap in the durational distributions of short and long vowel phonemes across positions in the utterance can lead to the misidentification of phonemic vowel length and (2) the direction of bias in such misidentification is determined by the distributional properties of the short and long vowel phonemes in the region of the durational overlap. Because short vowel phonemes are typically more frequent in occurrence and less variable in duration than long vowel phonemes, long vowel phonemes are more likely to be misidentified than short vowel phonemes. Results of production and perception studies in Tokyo Japanese support these hypotheses.