PRiSM: Benchmarking Phone Realization in Speech Models

Phone recognition (PR) serves as the atomic interface for language-agnostic modeling for cross-lingual speech processing and phonetic analysis. Despite prolonged efforts in developing PR systems, current evaluations only measure surface-level transcription accuracy. We introduce PRiSM, the first open-source benchmark designed to expose blind spots in phonetic perception through intrinsic and extrinsic evaluation of PR systems. PRiSM standardizes transcription-based evaluation and assesses downstream utility in clinical, educational, and multilingual settings with transcription and representation probes. We find that diverse language exposure during training is key to PR performance, encoder-CTC models are the most stable, and specialized PR models still outperform Large Audio Language Models. PRiSM releases code, recipes, and datasets to move the field toward multilingual speech models with robust phonetic ability: https://github.com/changelinglab/prism.

Linear Script Representations in Speech Foundation Models Enable Zero-Shot Transliteration

Multilingual speech foundation models such as Whisper are trained on web-scale data, where data for each language consists of a myriad of regional varieties. However, different regional varieties often employ different scripts to write the same language, rendering speech recognition output also subject to non-determinism in the output script. To mitigate this problem, we show that script is linearly encoded in the activation space of multilingual speech models, and that modifying activations at inference time enables direct control over output script. We find the addition of such script vectors to activations at test time can induce script change even in unconventional language-script pairings (e.g. Italian in Cyrillic and Japanese in Latin script). We apply this approach to inducing post-hoc control over the script of speech recognition output, where we observe competitive performance across all model sizes of Whisper.

Languages in Multilingual Speech Foundation Models Align Both Phonetically and Semantically

Cross-lingual alignment in pretrained language models (LMs) has enabled efficient transfer in text-based LMs. Such an alignment has also been observed in speech foundation models. However, it remains an open question whether findings and methods from text-based cross-lingual alignment apply to speech. Building on prior work on spoken translation retrieval, we perform pronunciation-controlled experiments to observe if cross-lingual alignment can indeed occur in such models on a semantic basis, instead of relying on phonetic similarities. Our findings indicate that even in the absence of phonetic cues, spoken translation retrieval accuracy remains relatively stable. We follow up with a controlled experiment on a word-level dataset of cross-lingual synonyms and near-homophones, confirming the existence of both phonetic and semantic knowledge in the encoder. Finally, we qualitatively examine the transcriptions produced by early exiting the encoder, where we observe that speech translation produces semantic errors that are characterized by phonetic similarities to corresponding words in the source language. We apply this insight from early exiting to speech recognition in seven low-resource languages unsupported by the Whisper model, and achieve improved accuracy in all languages examined, particularly for languages with transparent orthographies.

Dialetto, ma Quanto Dialetto? Transcribing and Evaluating Dialects on a Continuum

There is increasing interest in looking at dialects in NLP. However, most work to date still treats dialects as discrete categories. For instance, evaluative work in variation-oriented NLP for English often works with Indian English or African-American Venacular English as homogeneous categories (Faisal et al., 2024; Ziems et al., 2023), yet even within one variety there is substantial variation. We examine within-dialect variation and show that performance critically varies within categories. We measure speech-to-text performance on Italian dialects, and empirically observe a geographical performance disparity. This disparity correlates substantially (-0.5) with linguistic similarity to the highest performing dialect variety. We cross-examine our results against dialectometry methods, and interpret the performance disparity to be due to a bias towards dialects that are more similar to the standard variety in the speech-to-text model examined. We additionally leverage geostatistical methods to predict zero-shot performance at unseen sites, and find the incorporation of geographical information to substantially improve prediction performance, indicating there to be geographical structure in the performance distribution.

Phonotactic Complexity across Dialects

Received wisdom in linguistic typology holds that if the structure of a language becomes more complex in one dimension, it will simplify in another, building on the assumption that all languages are equally complex (Joseph and Newmeyer, 2012). We study this claim on a micro-level, using a tightly-controlled sample of Dutch dialects (across 366 collection sites) and Min dialects (across 60 sites), which enables a more fair comparison across varieties. Even at the dialect level, we find empirical evidence for a tradeoff between word length and a computational measure of phonotactic complexity from a LSTM-based phone-level language model-a result previously documented only at the language level. A generalized additive model (GAM) shows that dialects with low phonotactic complexity concentrate around the capital regions, which we hypothesize to correspond to prior hypotheses that language varieties of greater or more diverse populations show reduced phonotactic complexity. We also experiment with incorporating the auxiliary task of predicting syllable constituency, but do not find an increase in the negative correlation observed.