INT J LANG COMMUN DISORD, NOVEMBER–DECEMBER 2012, VOL. 47, NO. 6, 752 Letter to the Editor Regarding “Syrnyk, C. and Meints, K., 2012, Fishy fishes: the typicality of object stimuli used to assess children’s language in the Reynell Development Language Scales – III. Inter- national Journal of Language and Communication Disorders, 47(4), 437–450” We are puzzled and concerned by the aims and conclusions of Syrnyk and Meints’ study published in the July-August edition of the IJLCD (vol. 47, 4) and would appreciate an opportunity to respond to the observations and comments they make about a language test we co-authored. Syrnyk and Meints correctly observe that the Reynell Developmental Language Scales-3: RDLS-3 (Edwards et al. 1997) is one of most popular and highly regarded measures of child language, has good reliability and validity (pp 438–9) and age norms, percentiles and standard scores based a large representative sample of children (1074) living in the UK. Notwithstanding this observation, the authors present a straw man to knock down, viz the objects used in the test under estimate a young child’s ‘knowledge’ of language, as not all the objects used in the test are ‘typical’. This claim is based on four tasks. Syrnyk and Meints showed adults and four-year-old participants 15 pairs of photographs where one photograph was of a RDLS-3 object and the other picture within the pair was of a matched object judged to be more typical by participants in an earlier study (Meints et al. 1999). Adult participants ranked each item on a typicality scale (Ex- periment 1) and a forced choice task (Experiment 2). Having established that the photographs were more ‘typical’ exem- plars than the selected RDLS-3 objects, twenty-three four year olds participated in an intermodal preferential looking task (Experiment 3) and 21 four year olds of the same group participated in a further forced choice task (Experiment 4). They were asked to point to one of the two pictures in response to ‘which is the more normal-looking picture’. Given that the four-year olds would, we assume, be famil- iar with the names of the objects (although maybe not with the term normal), it is not clear how their picture preference is pertinent. Further, not all the test items were used: some very prototypical objects included in RDLS-3 such as pencil, sock and brush were excluded from this study. Thus, it is puzzling how the authors’ results bring them to the conclusion that the RDLS-3 can lead to over-diagnosis of speech and/or language (sic) disorders (pp 448,449). Not only do their tasks, meth- ods and materials differ from those used in the Scales but they seem to misunderstand the nature of this language test, the well-considered criteria used to select the vocabulary items and, most importantly, the purpose of using a developmental framework. This is not a test of vocabulary but a norm based test that investigates language status. Many young children commonly both under-extend and over-extend the word-object relation- ship on the pathway to ‘knowing’ a word. The early tasks in RDLS-3 involve selection of an object (not a picture) from a group of objects, and naming objects. While low typicality of an object may impede very young children, a self-citation of one of the authors (Meints et al. 1999) acknowledges that chil- dren start to associate nouns with less typical examples from 18 months. Walk into any nursery and watch the imagina- tive play of the children. The Scales map a steady progression across the ages tested. A child’s score is compared with his or her peers’, so indeed it is true that a child who has not yet as- sociated a word with a range of exemplars will score lower that a child who has. This is part of language development. This is what the test exposes. The development of ‘understanding’ of vocabulary is part of the complex process of language de- velopment. However, the finding is valid as the child’s score is relative to the cohort tested during standardisation of the test. Many clinical language tests involve complex skills, in- cluding the ability to recognise pictures and objects some of which may be non-typical exemplars; attend to tasks; co- operate with the tester and so on. Testing language in young children is tricky and in order to ensure that results are valid, standardisation using a large representative sample is vital. However, in common with all psychometric tests, the RDLS- 3 is not perfect and over the years that it has been used with thousands of children, we have received feedback, mainly about the use of the test in clinical situations. As a result, we have recently produced a new version of the test, the New Reynell Developmental Language Scales: NRDLS (Edwards et al. 2011). This new version, influenced by recent child lan- guage research and feedback from users, has parallel produc- tion and comprehension tasks (many of which are completely new), employs objects in the early stages and progresses to picture stimuli in the later stages. The test is easier to admin- ister and the number of objects has been reduced. Further, (Syrnyk and Meints please note) the fish has gone! Susan Edwards†, Carolyn Letts‡ and Indra Sinka§ †The University of Reading ‡Newcastle University §The Open University e-mail: s.i.edwards@reading.ac.uk References EDWARDS, S., FLETCHER, P., GARMAN, M., HUGHES, A., LETTS, C. and SINKA, I., 1997, The Reynell Developmental Language Scales III: 3 rd edition. (Windsor: NFER-Nelson) EDWARDS, S., LETTS, C. and SINKA, I., 2011, The New Reynell De- velopmental Language Scales (GL Assessment: Chiswick, UK) MEINTS, K., PLUNKETT, K. and HARRIS, P. 1999, When does an ostrich become a bird? The role of typicality in early word learning. Developmental Psychology, 35(4), 1072–1078. International Journal of Language & Communication Disorders ISSN 1368-2822 print/ISSN 1460-6984 online C  2012 Royal College of Speech and Language Therapists DOI: 10.1111/j.1460-6984.2012.00200.x