JCPSLP Vol 17 No 3 2015

The audio file was transcribed into a Word document and analysed using the following codes: a successful child initiation, successful caregiver initiation, connected utterance, failed utterance, behavioural directive, or caregiver expansion. The ratio of successful child to caregiver initiations was also calculated. These measures were selected as their presence, or absence, provides an indication of how supportive the HLE is of language acquisition. See the Appendix for the definitions adhered to for these codes. An inter-rater reliability analysis using the Kappa statistic was performed. A total of 297 utterances from six participants’ recordings were independently coded for comparison, and there was an almost perfect agreement, k = .949 (95% CI, .919 to .978, p < .005). Results The first aim of this study was to compare the quality of the HLE between the two groups. The non-parametric Mann Whitney U test was selected due to the small sample size in each group ( N = 5). This test is robust against violations to normality and is not affected by outliers in the data, and is therefore a suitable choice for this study. A p value of <.05 indicated a statistically significant difference between the groups. It was hypothesised that the HLE of the children who were D/HH would be less supportive of language development. This hypothesis was strongly supported (see Table 3). Five out of seven measures returned a significant result, with all but one generating a large effect. The second aim was to examine whether the quantity of language exposure was different between the groups. The hypothesised difference in adult word count, conversational turn count, and child vocalisation count was not supported. Adult word count in the D/HH group ( Mdn = 10335, Range = 4824–19401) was not significantly different to the NH group ( Mdn = 7261, Range = 2617–11857), U = 10.0, p = .690, r = .17. There was a non-significant difference in the conversational turn count of the D/HH group ( Mdn = 293, Range = 134–659) and the NH group ( Mdn = 409, Range = 247–666), U = 12.0, p = 1.000, r = .03. The child vocalisation count was also not significantly different in the D/HH ( Mdn = 1367, Range = 319–2567) and NH ( Mdn = 1334, Range = 887-3292) groups, U = 12.0, p = 1.000, r = .03. Discussion The aim of this study was to investigate differences in the quality and quantity of interaction in the HLE between children who are D/HH and their caregivers, and children

Table 2. Description of intervention received by D/HH children

Child

Intervention

HI Female 1 Weekly sessions at Telethon Speech and Hearing since 4/5 months of age. HI Female 2 Weekly sessions at Telethon Speech and Hearing since 3 months of age. Articulation therapy from 3 years. HI Male 3 Twice weekly sessions at Telethon Speech and Hearing since birth. HI Male 4 Weekly sessions at Telethon Speech and Hearing since 3 months of age. Recently changed to fortnightly sessions. HI Female 5 Nature of therapy received at Telethon Speech and Hearing not provided. Procedure The Language Environment Analysis (LENA) system is an audio-recording device used to collect and analyse verbal information about the HLE (Aragon & Yoshinaga-Itano, 2012). As part of the therapy program at Telethon Speech and Hearing, caregivers were provided with feedback surrounding the HLE using the LENA system. The audio recordings for this group were therefore pre-existing. The children wore the device continuously for up to 16 hours, and the caregivers were instructed on the use of the device, including communicating and interacting naturally. The caregivers rated how typical the day was on a 5-point Likert scale in terms of communication, events that occurred, and overall behaviour. No days were scored at 2 or below. The audio recordings for both groups were input into the LENA system, which automatically generated data for adult word count, conversational turn count, and child vocalisation count. Two participants did not record for the entire day, and consequently did not provide a complete recording. As such, measures were calculated from a 5-hour period (from 9am to 2pm) for all participants to ensure consistency. The LENA system calculated data for each of the five hours, and the hour with the median number of conversational turns was identified. This hour was coded to provide the conversational quality measures, thus ensuring a representative sample of the language environment.

Table 3. Descriptive measures, significance, and effect size for quality of interaction measures

Measure

D/HH ( N = 5) Median (range)

NH ( N = 5) Median (range)

Significance Effect size ( r )

Number of successful child initiations

17 (13–24)

32 (22–50)

.008**

.76 a

Number of successful caregiver initiations

16 (12–20)

22 (20–28)

.004**

.80 a

Number of connected utterances

44 (27–144)

196 (85–220)

.016*

.69 a

Number of failed utterances

30 (22–42)

22 (13–30)

.075

.50 a

Number of behavioural directives

17 (10–36)

9 (8–28)

.048*

.56 a

Number of caregiver expansions

5 (1–11)

18 (6–31)

.048*

.56 a

Ratio of successful child to caregiver initiations

1.26 (0.75–1.43)

1.28 (1.05–2.45)

.211

.30 b

Note. All significance levels were one-tailed. a large effect b medium effect * p < .05; ** p < .01.

147

JCPSLP Volume 17, Number 3 2015

www.speechpathologyaustralia.org.au