Consonant importance to word understanding varies across languages and sounds
Quantifying Consonant Contributions to Word Intelligibility via Acoustic Masking
Computation and Language
Summary
Not all consonant sounds contribute equally to how well we understand spoken words. The authors developed a way to measure how much each consonant matters by digitally silencing one sound at a time and seeing if speech recognition software still gets the word right. They found that common consonants tend to be less critical, while those that make bigger differences between words are more important. This varies across languages, meaning the impact of each consonant depends on the language spoken.
What this means in practice
- •For speech therapists: Prioritize specific consonants for therapy based on their measured impact on word understanding to optimize treatment time.
- •For speech recognition developers: Improve word recognition models by understanding which consonants contribute most to correct speech recognition across languages.
Authors
Eunjung Yeo, Kwanghee Choi, Krupaben Kothadia, Visar Berisha, Julie M. Liss, David R. Mortensen, David Harwath
Abstract
Consonants contribute unequally to whether a word is understood. Given the limited time available for therapy, ranking consonants by contribution to intelligibility helps prioritize intervention targets in motor speech disorders. However, measuring this contribution relies on perceptual studies that are difficult to scale. This paper presents a scalable method that measures consonant contribution using acoustic masking. We silence one consonant at a time in an isolated word and test whether an automatic speech recognition (ASR) model still recognizes the word. We define a consonant's contribution score as the proportion of its masked instances for which the word becomes misrecognized, which we refer to as the mask-induced misrecognition rate (MMR). We validate MMR against two linguistic factors previously reported to correlate with consonant contribution, namely phoneme frequency and functional load. We apply this analysis across four languages, English, Spanish, German, and Czech, using three ASR architectures, MMS (encoder-only), Whisper (encoder-decoder), and Qwen3-ASR (LLM-based). Using partial Spearman correlations, we find that phoneme frequency correlates negatively with MMR while functional load correlates positively. In other words, more frequent consonants are less disruptive when masked, whereas consonants carrying more lexical contrast are more disruptive. Further cross-language analysis shows that consonant rankings are not consistent, indicating that consonant contribution is language-dependent.