How Linguistic Chickens Help Spot Spoken-Eggs: Phonological Constraints on Speech Identification
Bibliographic record
Abstract
It has long been known that the identification of aural stimuli as speech is context-dependent (Remez et al., 1981). Here, we demonstrate that the discrimination of speech stimuli from their non-speech transforms is further modulated by their linguistic structure. We gauge the effect of phonological structure on discrimination across different manifestations of well-formedness in two distinct languages. One case examines the restrictions on English syllables (e.g., the well-formed melif vs. ill-formed mlif); another investigates the constraints on Hebrew stems by comparing ill-formed AAB stems (e.g., TiTuG) with well-formed ABB and ABC controls (e.g., GiTuT, MiGuS). In both cases, non-speech stimuli that conform to well-formed structures are harder to discriminate from speech than stimuli that conform to ill-formed structures. Auxiliary experiments rule out alternative acoustic explanations for this phenomenon. In English, we show that acoustic manipulations that mimic the mlif-melif contrast do not impair the classification of non-speech stimuli whose structure is well-formed (i.e., disyllables with phonetically short vs. long tonic vowels). Similarly, non-speech stimuli that are ill-formed in Hebrew present no difficulties to English speakers. Thus, non-speech stimuli are harder to classify only when they are well-formed in the participants' native language. We conclude that the classification of non-speech stimuli is modulated by their linguistic structure: inputs that support well-formed outputs are more readily classified as speech.
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 | 0.000 |
| 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 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".