The effect of non-word length on types of errors in repetition of Serbian preschool children

Abstract

Typical phonological error patterns observed on the word, syllable, and phoneme level in children’s language production are sensitive to and vary across (supra)segmental contexts. A valid procedure for investigating the effect of different contexts on errors is the non-word repetition task with varied phonological structures. Continuing on our previous findings (Popović et al., 2017), which show a significant effect of non-word length on total repetition accuracy, the aim of this study was to examine this effect on specific types of errors in repetitions. The study included 75 monolingual typically developing (TD) Serbian children, aged 3 to 7. The used non-word repetition task consisted of 48 non-words constructed by systematically varying the number of syllables (two, three, and four), and three other phonological parameters: the position of syllable stress, onset and coda. The children’s task was to repeat each prerecorded non-word after hearing it. The repetitions were audio recorded, transcribed, and coded for 16 types of errors on the word, syllable and phoneme level. Two-way factorial ANOVAs, with factors age and number of syllables (N Syll.), were performed for the six most frequent errors in the repetitions. The following significant effects and interactions are shown for the different errors: Stress shifting: age (F(4, 70) = 8.124, p < .001), N Syll. (F(2, 140) = 118.5, p < .001), interaction (F(8, 140) = 2.576, p = .012); Onset consonant cluster reduction: age (F(4, 70) = 13.35, p < .001), N Syll. (F(2, 140) = 26.92, p < .001); Coda consonant omission: age (F(4, 70) = 8.645, p < .001), N Syll. (F(2, 140) = 15.84, p < .001); Coda consonant addition: age (F(4, 70) = 7.474, p < .001), N Syll. (F(2, 140) = 16.89, p < .001), interaction (F(8, 140) = 3.34, p = .002); Consonant metathesis: age (F(4, 70) = 11.01, p < .001), N Syll. (F(2, 140) = 86.05, p < .001), interaction F(8, 140) = 5.650, p < .001); Phoneme substitution: age (F(4, 70) = 30.97, p < .001), N Syll. (F(2, 140) = 130.5, p < .001), interaction (F(8, 140) = 2.41, p = .018). It may be concluded that with increasing non-word length, the repetition accuracy at all three levels (word, syllable, and phoneme) decreases in Serbian TD children, especially at younger ages. Three-year-olds have mastered some parameters of the shortest non-words, but make a large number of all errors on the longer ones. With age, children master word length, and at the age of 7 they make minimal errors, except for phoneme substitutions and stress shifting on the longest four-syllable non-words.