The main aims of the study were to examine the relative influence of phonotactic probability on nonword repetition accuracy using the Test of Early Nonword Repetition-Revised (TENR-R; Stokes & Klee, 2011) and to compare the sensitivity to these measures in Typically Developing (TD) and Late Talking (LT) preschoolers across a period of 18 months.

Aims: The main aims of the study were to examine the relative influence of phonotactic probability on nonword repetition accuracy using the Test of Early Nonword Repetition-Revised (TENR-R; Stokes & Klee, 2011) and to compare the sensitivity to these measures in Typically Developing (TD) and Late Talking (LT) preschoolers across a period of 18 months.

Method: The participants were enrolled in a long term project titled “Learning to talk: a research project on children’s early language development” (LTT; Klee, Stokes, & Moran, 2011-2015) and were recruited through early childhood specialists, by word of mouth from parents of the participants, and advertisement in nurseries. Data collection was carried out at three points in time: Time 1 (T1), Time 2 (T2) and Time 3 (T3). At T1 the participants were grouped into TD and LTs based on their language status assessed using the New Zealand adaption of the MacArthur-Bates Communicative Development Inventory: Words and Sentences (NZ CDI: WS; Reese & Read, 2000). A total of 134 participants between 24 and 30 months of age (M= 27.38, SD= 1.64) attempted the TENR-R at T1. At T3, a total of 138 participants between the age range of 42 and 49 months (M= 45.36, SD= 1.97) participated in the TENR-R. A computerised version of the TENR-R was administered at T1 and T3, and the participants were instructed to repeat the nonwords they heard.

Scoring and analysis: The audio and video recordings of the nonword repetitions on the TENR-R were transcribed using ‘Phon’ (Rose et al., 2006; Rose & MacWhinney, 2014) and scored for overall accuracy as well as percentage of consonant, vowel and phoneme accuracy. The phonotactic probabilities of the nonwords repeated accurately were calculated from the CELEX database and averaged for each child. Correlation and regression analyses were carried out to investigate the relationship between nonword repetition accuracy and phonotactic probability.

Results: The findings from correlation and regression analysis confirmed the presence of an association between phonotactic probability and nonword repetition accuracy. At T1, the TD participants were sensitive to Positional Segment Frequency (r(91)= .18, p< .05) while the LTs did not show sensitivity to any measure of phonotactic probability. At T3, both measures of phonotactic probability were found to correlate with nonword repetition accuracy in TD children (PSF average r(105)= .34, p< .001; BF average r(105)= .51, p< .001) and LTs (PSF average r(33)= .63, p< .001; BF average r(33)= .39, p< .001).

Conclusion: The results of the present study provide valuable insights into the nature of lexical representations in preschoolers. The presence of sensitivity to segmental information even at a very young age as well as in LTs at T3 along with the presence of the hypothesized sensitivity to Biphone Frequency at T3 is suggestive of the presence of an adult-like representational system that becomes fully adult-like with vocabulary growth and experience. There is a need for future studies to explore the nature of lexical representations from a continuity perspective and further expand the current findings.