Treating Children With Speech Sound Disorders

Lousada, M., Jesus, L., Hall, A., & Joffe, V. (2014). Research report: Intelligibility as a clinical outcome measure following intervention with children with phonologically based speech–sound disorders. International Journal of Language & Communication Disorders, 49, 584–601.

The majority of children on the caseloads of speech-language pathologists (SLPs) have speech–sound disorders (SSDs). When these SSDs affect intelligibility, what is the most effective intervention approach? This is the question asked by the researchers of this article.

Intelligibility is defined as how well a client’s speech is understood by other individuals (Pascoe, Stackhouse, & Wells, 2006). Although typically developing children of around 4 years of age are intelligible most of the time (Hodson & Paden, 1981), children with moderate to severe SSDs can show marked unintelligibility which can impact adversely on functional communication and social participation (Hustad, 2012). Improving intelligibility is typically a key long-term aim of SLPs working with all client groups with speech impairments. Two different methods of measuring intelligibility have been used with children with speech impairments: word identification tasks and rating scales (Ertmer, 2010; Miller, 2013; Pascoe et al., 2006). Word identification tasks require listeners to write down the words that they have understood to be said or involve the listener selecting words from a range of multiple-choice alternatives. The speech sample can be single words, sentences, or continuous speech, usually pre-recorded and randomized (Gordon-Brannan & Hodson, 2000). The listener’s responses are typically scored for the number of words they match correctly so that a percent intelligible score can be calculated (Gordon-Brannan & Hodson, 2000; Pascoe et al., 2006).

Rating scales, used typically with continuous speech samples, are of two types: interval scaling and direct magnitude estimation. Interval scaling is more typically used, requiring the listeners to rate speech samples (e.g., sentences) along a continuum of intelligibility (e.g., on a 4-point scale where 1 represents “completely unintelligible” and 4 represents “completely intelligible”). Direct magnitude estimation requires an estimate, usually a percentage, of parts of a sentence which are understood (e.g., an estimation of 50% would indicate that a listener has understood only about half of the message). The use of rating scales is potentially quick and easy to implement, but has some disadvantages: listeners could use different criteria on which they base their judgments, and ratings for the same children across raters can be widely variable.

The purpose of this investigation was to compare the effectiveness of two types of treatment approaches (phonological vs. articulation) using intelligibility as the primary outcome measure. Fourteen Portuguese children, aged 4.0 to 6.7 years, with phonologically based SSD participated in the intervention. Children were diagnosed as having phonologically based SSD after extensive assessment by an SLP, an audiologist, and a psychologist. Speech errors were not due to oro-motor difficulties or structural problems, as assessed through an oral examination, and all speech–sounds were stimulable when tested. Some children (n = 6) had nonverbal abilities within the average range (above 85), whereas the remaining eight had nonverbal IQ ranging between 85 and 62, and can be considered to have more general language learning difficulties.

The children were randomly assigned to phonological therapy (PT) or articulation therapy (AT; seven children in each group). The intervention for both groups consisted of 25 individual weekly sessions of 45 min in duration, divided into three blocks (9 + 8 + 8 weeks) without any breaks. Each block had a different target. Both groups were treated by the same SLP, who was blind to the aim of the study, to reduce bias and minimize the influence of some confounding variables. PT included a combination of phonological awareness activities from Gillon and McNeill’s (2007) program and auditory discrimination and listening activities from Lancaster (2008). The phonological awareness activities from Gillon and McNeill’s program included letter–sound knowledge, phoneme identity, and phoneme matching, blending, segmentation, and phoneme manipulation. Focus for the first two sessions of each block was on listening and discrimination activities, while the remainder of the sessions focused on phonological awareness. During the phonological awareness activities, productions of the target sounds were elicited.

OPEN IN VIEWER

AT consisted of a traditional AT approach following the “Van Riper Method” (Van Riper & Emerick, 1984). In this approach, the principal aim is to develop the child’s ability to discriminate and articulate the target sound correctly in isolation, syllables, words, phrases, and sentences. The first two sessions of each block focused on sensory-perceptual training, with the remaining sessions focusing on production. Therapy was undertaken on one target sound at a time. Different activities were used during the sensory-perceptual training (e.g., detect sound errors in the clinician’s speech). Direct instruction in the mechanism of sound production was used through different techniques (e.g., progressive approximation and phonetic placement).

Two types of speech samples were elicited at pre- and post-intervention points: single words and continuous speech. The children’s phonological abilities were assessed with 67 single words from a standardized phonetic–phonological test. To assess word intelligibility, four groups of words were selected from the 67 words. Groups of words were matched according to number of syllables and syllabic structure (equal or similar). A total of 56 words (14 per group) were included in this sample. This procedure ensured that each rater listened only once to each word and reduced the confounding variable of familiarity for one set of words over another (Ertmer, 2010).

A continuous speech sample was also obtained through a picture description task. Three pictures were specifically designed by the researchers and drawn by an illustrator to ensure that they depicted a wide range of everyday and familiar activities that would encourage speech and to maximize engagement by the children. They depict different scenarios and actions to elicit a broad-ranging speech sample.

Twenty-one people between the ages of 22 and 44, unfamiliar to the children in the study, were recruited to judge the children’s speech. Unfamiliar listeners were selected as the long-term aim of therapy for children with phonologically based SSD is to improve the ability to communicate with all people. The listeners were divided into seven clusters. Each cluster of three listeners rated one different set of words or a continuous speech sample. Each cluster of three listeners heard all the words or continuous speech sample produced by two children in both the pre- and post-treatment conditions. For the single words, judges listened to each word only once and orthographically transcribed each word. In the continuous condition, a rating scale was used as four children had severe speech difficulties and it was not always possible to know exactly what the children were saying. The participants were asked to listen to the samples and immediately afterward to classify them according to their degree of understanding using the rating scale supplied. The following instructions were provided: “Select the number that best corresponds to the child’s speech using this rating scale (“1—impossible to understand”; “2—very difficult to understand”; “3—difficult to understand”; “4—easy to understand”; “5—very easy to understand”). Judges listened to the continuous speech samples once only and did this without seeing the pictures used to elicit them.