Ie-cor lexical database: Moscow school's methodology and error analysis

The new lexical database, IE-CoR, compiled along the lines of the Moscow School's methodology, is a step forward. The IE-CoR dataset consists of 170-item wordlists for 161 lects. The lexicographic quality is high, thanks to an international team of over 80 linguists working for four years. IE-CoR reflects the methodology advocated by the Moscow school of historical linguistics for almost 15 years. This includes key principles from Kassian et al. (2010) and adopted in the Global Lexicostatistical Database. These principles significantly boost the quality of the resulting dataset. However, Heggarty et al.'s online supplement claims that "IE-CoR protocols and meaning definitions differ radically" from Kassian et al. (2010). Heggarty et al. propose morphological simplicity as a criterion for selecting basic words. This criterion suggests simpler words have selective advantage, but lacks supporting examples. Five IE-CoR wordlists (Bulgarian, Hittite, Gothic, Tsakonian Greek, Kashmiri) were checked. The results for Bulgarian were excellent, with no errors found in ca. 100 checked concepts. Hittite's results were acceptable, though some errors were noted. Gothic's list was slightly biased towards etymological equivalents over corpus-supported synonyms. Tsakonian Greek and Kashmiri suffered from undetected loanwords. These loanwords skewed their placement in phylogenetic trees. Tsakonian Greek, influenced by vernacular Greek for millennia, contains many undetected loans. The IE-CoR list doesn't mark these loans, rendering it "basically useless for phylogenetic purposes." The Kashmiri list contains undetected borrowings from Indo-Aryan languages. These undetected loans placed Kashmiri incorrectly in Heggarty et al.'s tree. Despite the improvements, the tree classification is statistically weak. It also contradicts established knowledge, making it an unreliable basis for further research.