* Add basic tests for Tamil (ta) * Add comment Remove superfluous condition * Remove superfluous call to `pipe` Instantiate new tokenizer for special case