“…At HV3 its loading for the semantic factor was negative, a rather startling result for a supposed indicator of semantic skill. Other researchers have reported a smaller effect size for D than for other measures (Owen & Leonard, 2002;DeThorne et al, 2008); similar results were obtained in two other recent studies of the WRRP data (DeThorne, Petrill, Schatschneider, & Cutting, 2010;DeThorne, Deater-Deckard, Mahurin-Smith, & Petrill, under review), each of which used independently derived values for D. The difficulty, then, is unlikely to be faulty calculation; perhaps it is that the serial type-token ratios from which D is calculated are no more effective than aggregate TTRs in distinguishing between typical and atypical language use (Watkins, Kelly, Harbers, & Hollis, 1995). If a single TTR is not useful in discriminating between typical and impaired language abilities, it is unsurprising that an agglomeration of TTRs is no more so.…”