Generating Feature Vectors from Phonetic Transcriptions in Cross-Linguistic Data Formats

TL;DR

The paper addresses the challenge of comparing cross-linguistic speech sounds when existing feature inventories fail to cover all attested sounds. It introduces a dynamic pipeline that converts CLTS feature bundles into binary feature vectors, implemented in a Python package (soundvectors) and designed to work with Lexibank data. The approach uses a 39-feature ternary system with extensions for clicks, tones, complex tones, and diphthongs, applying a hierarchical processing workflow to produce robust, comparable vectors. Evaluation shows meaningful clustering consistent with established phonological classifications and substantial coverage across large multilingual datasets, enabling scalable downstream analyses in computational phonology and historical linguistics.

Abstract

When comparing speech sounds across languages, scholars often make use of feature representations of individual sounds in order to determine fine-grained sound similarities. Although binary feature systems for large numbers of speech sounds have been proposed, large-scale computational applications often face the challenges that the proposed feature systems -- even if they list features for several thousand sounds -- only cover a smaller part of the numerous speech sounds reflected in actual cross-linguistic data. In order to address the problem of missing data for attested speech sounds, we propose a new approach that can create binary feature vectors dynamically for all sounds that can be represented in the the standardized version of the International Phonetic Alphabet proposed by the Cross-Linguistic Transcription Systems (CLTS) reference catalog. Since CLTS is actively used in large data collections, covering more than 2,000 distinct language varieties, our procedure for the generation of binary feature vectors provides immediate access to a very large collection of multilingual wordlists. Testing our feature system in different ways on different datasets proves that the system is not only useful to provide a straightforward means to compare the similarity of speech sounds, but also illustrates its potential to be used in future cross-linguistic machine learning applications.

Figures (6)

• Figure 1: Workflow of vector creation.
• Figure 2: Cosine similarities between consonant vectors generated with our model.
• Figure 3: Cosine similarities between vowel vectors generated with our model.
• Figure 4: Two-dimensional reduction of feature vectors using PCA.
• Figure 5: Two-dimensional reduction of feature vectors using t-SNE.