Purpose
Analyze language structures, cross-linguistic patterns, and diachronic semantic change.
Key Datasets
- WALS (wals.info): 192 linguistic features across 2,679 languages — phonology, morphology, syntax, word order
- HistWords (nlp.stanford.edu/projects/histwords): Diachronic word embeddings for English, French, German, Chinese
- Universal Dependencies: 200+ treebanks, 100+ languages, dependency annotation
Analysis Types
- Typological analysis: Feature distributions, language universals, areal patterns
- Diachronic analysis: Semantic drift, grammaticalization, lexical change
- Syntactic analysis: Constituency/dependency parsing, word order patterns
- Phonological analysis: Sound inventories, phonotactics, prosody
- Corpus analysis: Frequency distributions, collocations, concordances
Protocol
- Language identification — Identify language family, branch, typological profile
- Feature analysis — Map relevant WALS features for target language(s)
- Comparative analysis — Cross-linguistic comparison using typological databases
- Statistical testing — Test for significant patterns (chi-square, Fisher's exact)
- Visualization — Geographic and phylogenetic visualizations of features
Rules
- Use ISO 639-3 language codes for unambiguous identification
- Cite primary grammars and fieldwork sources
- Distinguish descriptive from prescriptive claims
- Handle endangered language data with cultural sensitivity