LED-A: calculating linguistic distances
LED-A is a web app for calculating linguistic distances with Levenshtein distance using phonetic IPA transcriptions.
The Levenshtein Edit Distance App (LED-A) was originally developed to measure pronunciation distances between dialects using phonetic transcriptions. It offers five variants of the Levenshtein distance and a binary measure for assessing lexical differences.
Pairwise distances can be explored through hierarchical clustering and multidimensional scaling (MDS), with several methods available for both. Users can also upload externally computed distance matrices for statistical analysis.
What’s new in LED-A
Recent extensions have substantially broadened LED-A’s functionality. Acoustic distances can now be calculated directly from speech recordings using Mel-frequency cepstral coefficients (MFCCs) and Dynamic Time Warping (DTW), eliminating the need for manual transcription. Acoustic measurements are optimized for perceptual similarity, while Praat’s gender changer reduces differences attributable to speaker gender.
LED-A also supports syntactic distance measurement using part-of-speech (PoS) n-gram frequency vectors derived from parallel texts in Excel or CoNLL-U format. Distances are calculated as one minus the correlation between these vectors.
Who is involved
Wilbert Heeringa
Researcher
Wilbert Heeringa is a researcher at the Fryske Akademy, working on dialectometry, phonetics and statistics. He uses computational methods to measure dialect variation and change in vocabulary, morphology and pronunciation, and develops tools for phonetic research such as Visible Sounds.
Hans Van de Velde
Researcher
Hans Van de Velde is a researcher in sociolinguistics at the Fryske Akademy. He studies language variation and change, multilingualism and phonetics, including the pronunciation of Frisian and Dutch in Fryslân, the evaluation of variations, and the development of experimental sociolinguistics.
