Processing pronunciation variation with independently mappable allophones
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
• A long-standing merger of /n/ to /l/ in Cantonese has produced [n]- and [l]-initial pronunciation variants that are effectively allophones of syllable-initial /l/ and /n/. • [n] and [l]-initial allophones are distinguishable in perception. • In recognition and encoding paradigms, [n] and [l] allophones are processed neither equivalently nor distinctly when the targets bear the more common [l]-initial allophone. • Error rates are high when the targets bear the [n]-initial allophone. • [n] and [l] are allophonic variants independently mapped to a phoneme, with connection strengths varying. Sound change can present synchronic variation with categorical pronunciation variants. This is the case in Cantonese, where syllable-initial /n/ is merging with /l/, occasionally creating homophones (e.g., lou5 腦 “brain”/ 老“old”) and giving rise to [n]- and [l]-initial pronunciation variants that are allophones. This pronunciation variation offers insight into how variation is processed in spoken word recognition because [n] and [l] in Cantonese are not associated with an orthographic standard. Across four experiments, we examine the perception, recognition, and encoding of Cantonese [n] and [l], and use Bayesian analyses where gradient interpretations are more straightforward. We observe perceptual evidence that these allophones are distinguishable (Exp 2). In recognition (Exp 1) and encoding (Exp 3) paradigms, we find that the [n] and [l] allophones are processed neither equivalently nor distinctly when the targets bear the more common [l]-initial allophone. When the targets bear the [n]-initial allophone (Exp 4), we observe high error rates, and somewhat contradictory results. Altogether, the results suggest that [n] and [l] are allophonic variants independently mapped to a phoneme, with connection strengths varying as a function of the frequency, such that the more common [l]-initial pronunciation demonstrates an overall recognition advantage.
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.
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 | 0.000 |
| 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) | 0.000 | 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 it