Jonathan Harrington’s research while affiliated with Ludwig-Maximilians-Universität in Munich and other places

In ongoing sound changes, a coarticulatory effect is often enhanced as the coarticulatory source that gives rise to it wanes. But quite how phonologisation and these reciprocal coarticulatory changes are connected is still poorly understood. The present study addresses this issue through an acoustic analysis of metaphony, which like umlaut has its phonetic origins in VCV coarticulation, and which was analysed in three geographically proximal varieties spoken in the so-called Lausberg area in Southern Italy. The corpus was of 35 speakers producing mostly disyllabic words with phonetically mid stem vowels and suffix vowels that varied in phonetic height. The results of functional principal components analysis applied to the stem vowels’ first two formant frequencies showed a progressively greater enhancement to the vowel stem across the three regions that was characterised by raising, diphthongisation, and then further raising and monophthongisation. Suffix erosion was quantified by counting deletions and the degree of vowel centralisation. The analysis showed a reciprocal relationship between stem enhancement and suffix erosion across, but not within, the three dialects. Overall, the results suggest that a trade-off of cues between suffix and stem vowel has progressed to different degrees between the three varieties.

Anticipatory nasalization in the vowel-nasal (VN) sequence often leads to the reduction of the nasal coda consonant (Ohala & Busa, 1995), resulting in either a nasal vowel Ṽ (Rochet, 1976) or a shifted oral vowel V' (Cresci, 2019). Chéng-Dū (CD, 成都) and Chóng-Qìng (CQ, 重庆) are two metropolitan areas in southwest China, both as varieties of the Chéng-Yú (成渝) sub- group within the Chuān-Qián (川黔) group of Southwestern Mandarin (SWM, 西南官话) (Li, 2009). It was reported that both the nasal coda and any nasalization of /an/-rime have been lost in the CD variety, accompanied by a raising and fronting of the pre-nasal vowel (i.e., /an/→/ɛ/) (Liao et al., 2022, 2023). There is some evidence that this is a phonetically motivated sound change in progress (Liao et al., 2024a, 2024b). The goal of this study was to investigate the extent of nasalization of /an/-rime words across two generations and the CD and CQ regions. For this purpose, we analyzed data from 20 CD and 16 CQ speakers from sex-balanced old (mean age: 58) and young (mean age: 22) age groups. The speech materials consisted of 36 tokens per talker (1296 tokens in total) formed from (CG)V(N) with C = /t, th, p, ph/, G =/ø, j, w/ and with rimes = /a, an, aŋ/. The four lexical tones were equally distributed within each of the final categories. Each participant was recorded with a nasalance device, separating speech signals from the oral and the nasal cavity. The amplitude of the nasal and oral channels was extracted for the sonorant (G)V(N) interval. The nasalance score for each observation was calculated from An/(An+Ao), resampled to 100 data points and lowess smoothed (Cleveland, 1979). Each of these signals was DCT-transformed from which the DCT coefficients k0 and k2, which are proportional to the signals' mean and curvature (Harrington, 2010), were extracted. The extent of nasalization in the syllable final was assessed by calculating separately for each speaker the relative proximity of all tokens in the k0 × k2 space to the so-called oral and nasal anchors using the orthogonal projection ratio (op) (Stevens et al., 2019). The anchors were the mean values across the same speaker's (nasal) /aŋ/- and (oral) /a/-rime. Op values of −1 and +1 denote that tokens were coincident with the oral and nasal anchors respectively. The test was of whether the extent of nasalization in /an/-rime was affected by age group and region. For this purpose, a mixed model was applied with op values of /an/-rime as the dependent variable, DIALECT and AGE_GROUP as the fixed interacting factors, and the SPEAKER, GLIDE, and TONE as the random factors. The boxplots in the bottom panel of Fig. 1 shows (predictably) values around −1 and +1 for oral /a/- (blue) and nasal /aŋ/- (black) rimes. For /an/-rime (red), the boxplots show less nasalization in the CD than in the CQ variety and less nasalization for young than for old CD speakers (compare the red boxplots in the right panels). Compatibly, the statistical analysis showed that the op ratio for /an/-rime was significantly influenced by AGE_GROUP (p < 0.01), DIALECT (p < 0.001), and with AGE_GROUP significant in the Chengdu (p < 0.001) but not in the Chongqing variety. The analysis has shown that the Chengdu but not the Chongqing variety is undergoing nasal loss in /an/-rime words. In addition, the results show a sound change in progress, given that there is less nasalization for younger than for older Chengdu speakers. We suggest the possibility of a continuum of denasalization of /an/-rime nasalance in the Sichuan basin, with younger Chengdu speakers taking the lead.

The diachronic change by which coarticulatory nasalization increases in VN (vowel-nasal) sequences has been modelled as an earlier alignment of the velum combined with oral gesture weakening of N. The model was tested by comparing American (USE) and Standard Southern British English (BRE) based on the assumption that this diachronic change is more advanced in USE. Real-time MRI data was collected from 16 USE and 27 BRE adult speakers producing monosyllables with coda /Vn, Vnd, Vnz/. For USE, nasalization was greater in V, less in N, and there was greater tongue tip lenition than for BRE. The dialects showed a similar stability of the velum gesture and a trade-off between vowel nasalization and tongue tip lenition. Velum alignment was not earlier in USE. Instead, a closer approximation of the time of the tongue tip peak velocity towards the tongue tip maximum for USE caused a shift in the acoustic boundary within VN towards N, giving the illusion that the velum gesture has an earlier alignment in USE. It is suggested that coda reduction which targets the tongue tip more than the velum is a principal physiological mechanism responsible for the onset of diachronic vowel nasalization.

Introduction. A sound change could be derived from phonological development (Ohala, 2012), social interactions (Labov, 1963), language contact (Boretzky, 1991), and other factors. It has been reported that the nasal coda in /an/-rime words is lost in the Chengdu variety (CD) of Southwestern Mandarin (Liao et al., 2022, 2023). However, it is yet not clear whether this sound change is the result of language contact with Standard Mandarin (SM), since the less dominant dialect variety (Chengdu) is often more prone to be subject to sound change. In this study, we collect real-time magnetic resonance imaging (MRI) recordings on native speakers from both varieties and compare the velum movement as a function of time in the target segments between these two Mandarin varieties. The objective here is to compare the velum opening between /an, aŋ/-rimed words in CD and SM, so as to investigate whether this sound change in CD results from language contact with the dominant variety (SM) or if it is a phonetically driven sound change. Methods. This study reports data from 4 native CD speakers (2 female) and 3 SM speakers (2 female). Each participant was recorded with the 3T MRI system in a supine position and was asked to read the carrier phrases in their respective dialect. With a Mandarin syllable structure being 'CGVN', the finals of the target words (GVN) were phonologically /(G)an/ and /(G)aŋ/, with G (glide) in /ø (null), j, w/. MR images were recorded, reconstructed with a frame rate of 50 fps, and synced with noise-suppressed audio. The velum opening signal was derived from the MR images from each vocalic interval, i.e. the vowel in the CGVN syllable structure, with vocal tract aperture algorithms (Carignan et al., 2020) in MATLAB, which were then resampled to 100 data points for each observation. A total of 718 tokens (after deleting some incorrectly produced items) were analyzed and were then put into the discrete cosine transformation (DCT). The resulting DCT coefficients, k0 and k1 that are proportional to the mean and linear slope respectively (Watson & Harrington, 1999), were then clustered by k-means with two centers. The accuracies of the clustering results (with regard to the actual rime type) were compared between speaker groups. Results. The velum opening signals as a function of time for each speaker group are shown in the left panel in Figure 1. For the native speakers of Standard Mandarin, the velum opening during the vowel segment between /an, aŋ/ rimes are quite similar: both approaching the maximum opening at the vowel offset, showing a largely opened velum pattern; meaning the nasal coda is firm. However, for the native speakers of the Chengdu variety, the velum movement approached the maximum for /aŋ/-rime, but not for /an/- rime. The right panel of Figure 1 exhibits the original rime type (in text) and the predicted rime type generated by the k-means clustering algorithm (in the respective color legend) for each speaker group on the k0 × k1 space. Red color denotes observations predicted as /an/ rimes, and black for /aŋ/ rimes. The main result was that the extent of the distinction between /an, aŋ/ differed between SM and CD: for Mandarin speakers, 106 / 157 tokens (male/female) were analyzed with accuracies of 50.94% / 42.04% which is close to chance; by contrast, among Chengdu speakers, 237 / 218 tokens (male/female) were correctly categorized with scores of 93.67% / 100.00%. Thus, the result shows a similar velum movement pattern for /an, aŋ/-rimes in SM but a clear distinction with nasal loss in /an/rime in CD. Discussion. These results illustrate the great difference in the velum movement in /an/-rime words between two Mandarin varieties. More specifically, and as Figure 1 shows, /a/ has about the same degree of nasalization preceding /n, ŋ/ in SM, whereas in CD there are marked differences: the vowel in /an/-rime is close to oral, meanwhile the vowel in /aŋ/-rime is almost as nasalized as in SM. Despite the fact that the Standard Mandarin being the dominant variety in China, the oralization of the vowel in CD /an/-rime (Liao et al., 2022) does not appear to be a result of language contact with the SM variety. Instead, the Chengdu variety but not Standard Mandarin is participating in a phonetically motivated sound change involving /an/ → /ãn/ → /ɛn/ → /ɛ/ → /ɛ/ that has also been observed for other languages and varieties (Hajek & Maeda, 2000; Ohala & Busa, 1995). Example words: '班' from /pan/ to /pɛ/, '攀' from /phan/ to /phɛ/.

This study aims to examine whether the pre-nasal vowel change that commonly occurs with the phonologization of nasalization, such as the raising and possibly fronting (e.g. Latin 'manus' to French /mɛ/), extends beyond the language family and can be generalized to where the nasal trace is lost (e.g. /an/-rime with nasal coda loss in the Chengdu dialect). To do so, real-time MRI was applied to obtain vocal tract aperture signals in the palatal and hyperpharyngeal areas in different rimes (/a, an, i, o/) from four L1 Chengdu speakers. The distances of /a, an/ to anchors /i, o/ were compared in a transformed articulatory space, demonstrating that the vowel in /an/-rime is raised and fronted compared to that in /a/-rime. This concomitant vowel shift leads us to conclude that the universal pre-nasal vowel shift does extend to the nasal loss cases where the VN has become an oral vowel.

This study was concerned with understanding a case of diachronic vowel lengthening in the Gheg dialect of Albanian. Some of the long vowels of Gheg are thought to have emerged out of a historical process of compensatory lengthening following final schwa deletion. However, a confounding factor which may also explain why these vowels became longer is the effect of polysyllabic shortening being wiped out by schwa deletion. The main aim of this study was to disentangle the respective effects of compensatory lengthening and polysyllabic shortening on patterns of vowel duration in Gheg. An acoustic analysis of vowel duration in two word sets provided no evidence of an effect of polysyllabic shortening. Vowels were found to be longer, not shorter, in polysyllabic words, a fresh insight which suggests that vowel length may have instead been inherited from an open syllable and preserved when schwa was deleted.

The focus of the study is an agent-based model (ABM) concerned with simulating phonological stability and change using real speech data from a population of speakers. At the core of the model was a flexible and agent-specific association between acoustic exemplars and phonological categories. This was achieved by means of two general-purpose unsupervised machine learning algorithms: the first grouped exemplars into acoustic clusters, the second identified sets of clusters which largely contain exemplars of the same word classes. The model was tested first on data from Standard Southern British English, where a shift of /u/ to the front of the vowel space was expected not to cause any phonological re-categorisation. The simulation indeed showed phonological stability despite the phonetic change. Using the same settings, the ABM was then applied to data from New Zealand English in which a merger of /e@, I@/ diphthongs has taken place in the last 50 years. Compatibly with that, the simulation showed a shift from /e@/ towards /I@/ along with the neutralisation of the diphthong contrast. The results are discussed with respect to the conditions necessary to model a merger, taking into account both findings on marginal phonological relationships and principles from exemplar theory.