Probal Dasgupta, University of Hyderabad
Linguistics has been practising an antipurism that this paper argues should be developed explicitly. Scientific study of language presupposes the untenability of any purism, religious (sacred initial languages), historicist (past purer than decadent present), or codal (homogeneous codes purer than hybrids), forcing us into idealizations based on social assumptions about real or pure languages or items. As generativism outgrows the structural-formalist purism of an inner Form that would be defiled by Substance, generative representations begin to provide an extended interface between expression and content, not a unique interpoint. That interregion's dynamics reflects the duality of substances. Reference becomes impure not only because the chain spreads it out wide, but also because - it is argued here - paradigmatic ladders (which do for Delegation what chains do for Displacement) spread it deep down. Our case study works this out for the word /trikkhe/ which occurs only in the line /tin trikkhe nOy/ 'three threes are nine' in the Bangla multiplication table. Its referential labour co-involves but is not enslaved by /tin/ 'three'. Such a retooling of lexical reference takes the unexamined default purism out of a current debate between positions here called NC (lexical expression-content interfaces are uniquely provided for the computation so as to appear for LF as well as PF) and AM (given that much of the work is going into the PF-ward track, most or all lexicalization can occur there, feeding only skeletal scope information into LF), turning it into a more interesting debate which this paper tries not to resolve, but to highlight as a discussion-site we all need to examine more carefully, bringing the best of our methodologies to bear.