Tone mergers in spontaneous speech and gaps in the tone inventory
Bibliographic record
Abstract
We investigate the status of three ongoing tone mergers, comparing Heritage Cantonese in Toronto and Homeland Cantonese in Hong Kong, using conversational recordings from the Heritage Language Variation and Change (HLVC) Corpus (Nagy 2009). The mergers, which have been reported from experimental tasks in several Cantonese dialects (cf. Bauer et al. 2003, Mok et al. 2013, Zhang 2018) are: T2/T5 忍 jɐn35 / 引 jɐn23; T3/T6 印 jɐn33 / 孕 jɐn22; and T4/T6 仁 jɐn11 / 孕 jɐn22. In connected speech, many contextual variables influence the acoustic value of a tone in a given syllable (cf. Stanford 2016), so each token extracted from Labovian sociolinguistic interviews is coded for the segmental value of its onset, nucleus and coda, its position in the utterance, whether it is in a compound word, and the tones of the adjacent syllables. We have 7,495 tokens from 32 speakers (12 Generation 1, 12 Generation 2, 8 Homeland), but our most robust analysis moves forward with 2,400 tokens, excluding tokens that appear only in contexts where the other tone of the pair is not found. After normalization of syllable duration and speaker mean pitch, and conversion to semitones to account for differences in speaker pitch ranges (Zhu 1999, Edmondson et al. 2004), we find that two measures best represent the extent of each merger: (a) pitch at the 90% duration mark of each token and (b) the slope of the pitch track from 10% to 90% duration. Mixed Effects Models are fit to the data with, e.g., T2 vs. T5 as a binary dependent variable, the pitch measurements and the above-mentioned contextual factors as fixed effects, and word and speaker as random effects. If pitch emerges as significantly distinct for the two tones when contextual factors are thus controlled for, there is no merger. Comparing models fit to the data from each generation group, we determine whether the same social and/or linguistic factors condition the tone merger and measure how merged eachtone-pair is. Preliminary analysis shows the merger to be more advanced in the two heritage generations (which do not differ from each other) than the homeland group for T2/T5 and T4/T6. We are eager to discuss possible connections between gaps in the tone inventory (e.g., no T4 with /d/ onset, no T6 with /t/ onset) and mergers in progress. Are the previously reported mergers, based only on minimal pairs where both tones occur with the same onsets over- or under-stating the status of the merger? Do the gaps indicate mergers completed long ago?
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".