Proceedings Fonetik 2009 - Institutionen för lingvistik

Proceedings, FOETIK 2009, Dept. of Linguistics, Stockholm Universitymore often in the language input of the child, itis easier to recognize.Estimations of children’s use of open vs.close class words may be based on appreciationsof types and occurrences. For example, ina longitudinal study on four Swedish childrenand their parents it was shown that the 20 mostfrequent types of words stand for approximately35-45% of all the word occurrences in childlanguage, as well as in adult’s speech directedtowards children (Strömqvist, 1997). And evenmore notably, there were almost none openclass words among these 20 most frequentwords in child language or in child-directedspeech (CDS) in the Swedish material. On thecontrary, close class words such as de, du, va, e,ja, den, å, så constituted the most commonword forms. These word forms were most oftenunstressed and phonologically/phonetically reduced(e.g. the words were monosyllabic, andthe vowels were centralized). Nevertheless, itshould be mentioned that the transcriptionsused were not disambiguated in the sense thatone sound might stand for much more than thechild is able to articulate. For example, the frequente might be generalized to signify är (eng.is), det (eng. it/that), ner (eng. down) etc.In the current study, the questionnairesbased on parental reports prompted for wordstypes produced by the child. The use of wordswas differentiated by whether the word in questionwas used “occasionally” or “often” by thechild, but no estimations of number of wordoccurrences were made. Therefore the materialsused in the current study allow only forcomparison of types of words used.Based on the earlier study by Strömqvist weshould thus expect our data to show large andmaybe growing proportion of close class words.For example, the proportion of open vs. closeclass words measured at three different timepoints, corresponding to growing vocabularysizes, could progress as follows: 90-10%, 80-20%, 70-30% etc. But on the other hand, thetypically limited amount of close class words inlanguages should be reflected in the sample andtherefore our data should – irrespective of thechild’s vocabulary size – reveal large and stableproportion of open class words as compared toclose class words, measured at different timepoints corresponding to growing vocabularysizes (e.g. 90-10%, 90-10%, 90-10% etc.).Eriksson and Berglund (1995) indicate thatSECDI can to certain to extent to be used forscreening purposes to detect and follow upchildren who show tendencies of delayed oratypical language development. The currentstudy is a step in the direction for finding referencedata for typical development of open vs.close class words. Atypical development ofclose class words might thus give informationon potentially deviant grammatical development.MethodThe Swedish Early Communicative DevelopmentInventory (SECDI) based on parental reportsexists in two versions, one version onwords & gestures for 8-to 16-months-old childrenand the other version on words & sentencesfor 16-to-28-months-old children. In this studythe latter version, divided in checklists of 711words belonging to 21 semantic categories, wasused. The inventory may be used to estimatereceptive and productive vocabulary, use ofgestures and grammar, maximal length of utterance,as well as pragmatic abilities (Eriksson &Berglund, 1995).SubjectsThe subjects were 24 Swedish children (13girls, and 11 boys, age range 6.1- to 20.6-months by the start point of the project) randomlyselected from the National Swedish addressregister (SPAR). Swedish was the primarylanguage spoken in all the families with the exceptionof two mothers who primarily spokeFrench and Russian respectively. The parents ofthe subjects were not paid to participate in thestudy. Children who only participated duringthe first part of the collection of longitudinaldata (they had only filled in the version ofSECDI for 8-to 16-months-old children) wereexcluded from the current study resulting in 28completed forms filled by 17 children (10 girls,7 boys, age range 14- to 43-months at the timepoint of data collection). The data collected wasa mixture of longitudinal and cross-sectionaldata as follows: 1 child completed 4 forms, 1child completed 3 forms, 6 children completed2 forms, and 9 children completed 1 form.MaterialsTo estimate the number of open class words thesections through A2 to A12, as well as A14 andA15 were included. The semantic categories ofthese sections are listed in Table 1. Section A1-Sound effects/animal sounds (e.g. mjau) andA13-Games/routines (e.g. god natt, eng. goodnight) were not considered as representative127