Alison Nicholas, Sharon Millard and Frances Cook
The Michael Palin Centre for Stammering Children, Finsbnry Health Centre, Pine St,
Parent-child interaction (PCI) therapy is an indirect therapy approach for young children who stammer. While there is growing evidence to indicate that it is successful in reducing stammering, there is little known about the impact that this approach may have on other aspects of child development. This study was undertaken to see whether there is evidence for the argument that modification of parents’ interaction styles may have a detrimental effect on children’s language development. Parent and child language and parental participation were measured pre and post therapy. No evidence was found to support the hypothesis that PCI has a detrimental effect on children’s language development.
Parent-child interaction therapy (PCI) is one indirect therapeutic approach used in the UK with preschool children who stammer. This approach was developed at the Michael Palin Centre for Stammering Children and is aimed at helping parents identify and develop interaction styles within the family setting which promote their child’s natural ﬂuency (Rustin et al, 1996). The approach is based on multifactorial theories of stamrnering which propose that neurophysiological, psychological, social and linguistic factors all contribute to the onset and development of stamrnering (Conture, 2001; Smith and Kelly, 1997; Guitar, 1998). Therapy is individually tailored according to each family’s needs and the approach is consistent with the principles of the Demands and Capacities model (Starkweather et al, 1990) where therapy is aimed at increasing or enhancing children’s capacities for ﬂuency and decreasing the potential demands on a vulnerable system. Parents make observations about their own interaction styles from videotape recordings, and then identify and practise behavioural changes to enhance their child’s ﬂuency. These changes may include following the child’s lead, increasing the balance of turns, asking fewer questions, making more comments and slowing down their own rate of speech.
A key principle underlying the PCI programme is that parents of CWS are not regarded as being unique or necessarily different from other parents in their interaction styles, but that CWS ﬁnd it harder to assimilate the input. Miles and Bernstein Ratner (2001) found that the input of parents of CWS is parallel to that of parents of ﬂuent children, but suggest that CWS may have more difﬁculties resulting from their underlying linguistic vulnerabilities. Felsenfeld (1997) has proposed that a child’s temperament may play a role.
Recently, speciﬁc concerns have been raised about the advice typically given to parents of CWS to reduce the linguistic demands they are placing on their children. These are based on the language acquisition literature which has found that children receiving reduced input make slower progress in their language learning compared to their peers (for reviews see Huttenlocher et al, 1991 and Snow 1986). Miles and Bernstein Ratner (2001) and Bernstein Ratner and Silverman (2001) have proposed that asking parents of CWS to simplify their language may have a detrimental effect on the child’s language development. They warn that there could be long-term consequences of parents simplifying their language, particularly for those CWS who also present with depressed language skills (Nippold, 1990; Bernstein Ratner, 1997 for reviews).
However the nature/nurture debate regarding language development is as yet unresolved. Although it is agreed that the nature of the language input is important for language learning, particularly in semantic development, the inﬂuence of the child and the skills they bring to the interaction are also viewed as been equally essential in creating facilitative contexts for language development (Smith 1998). There are many examples of children in impoverished linguistic environments who still acquire language (Bishop & Mogford, 1993). In addition, therapeutic approaches which -have similar components to those used in PCI for early stammering have been shown to be effective in encouraging language development in children whose language skills are delayed (see Law et al., 1998).
It is not yet known whether an indirect environmental (PCI) approach with CWS has an impact on other aspects of their development, including language development. In a study aimed at identifying the possible mechanisms responsible for the effects of the Lidcombe Programme (Onslow and Packman 1999), Bonelli et al (2000) examined whether this treatment might result in “extensive curtailment of language functioning” ( p430). Although all the children’s language measures fell within normal limits both before and after therapy, some of the children’s scores at the end of the study were described as not being as high as might have been expected given the time period of the study. They propose that although the language measures and developmental norms used may not have fully captured children’s language skills, it is possible that this treatment may inﬂuence the development of preschool children’s language development. Such ﬁndings clearly warrant further investigation.
The present study was undertaken to investigate a) whether parents simplify their language or alter their conversational participation over a course of PCI therapy and b) the effects of PCI therapy on children’s language development. The aim was to identify the possible existence of trends and to establish whether further study with a larger group of subjects was warranted. The data used in this study was taken from a long-term efﬁcacy study of PCI therapy, where a multiple single Case design was used. Detailed analysis of the changes in dysﬂuency have been reported elsewhere (see Nicholas and Millard, in press; Millard and Nicholas, 2002) and indicate that the children’s fluency improved following therapy.
Participants were 6 children (4 boys, 2 girls) aged between 3:0 years and 4:11 years who had been diagnosed as having a stammering problem by two specialist speech and language therapists. All of the children had been stammering for a minimum of 12 months and none of them had attended for therapy in the 12 months prior to their recruitment into the study. They all came from a monolingual background, both parents were living at home and there were no identiﬁed medical problems.
Parents made 20-minute video-recordings of themselves and their child playing at home to provide naturalistic language samples for analysis. The ﬁrst ﬁve minutes of each recording were omitted and the following 10-minute samples were orthographically transcribed in random order.
The transcripts were coded using Codes for the Human Analysis of Transcripts (CHAT), which is part of the Child Language Data Exchange System (MacWhinney, 2000). Video recordings from the following points of time were analysed: Tl - 6 weeks prior to initial assessment, T2 - one week prior to initial assessment, T3 - 6 weeks post therapy. and T4 - 3 months post T3.
Computerized Language Analysis (CLAN) programmes were used to compute the mean length of utterance, number of different words, type-token ratio, vocd and Developmental Sentence Score.
Mean length of utterance (MLU): MLU (in morphemes) was used to measure both child and parent utterance length using the ﬁrst 50 utterances transcribed (as recommended by Miller and Chapman 1981). However ﬁve out of the total 48 transcripts did not contain 50 utterances (2 had 49 utterances, 1 had 48 utterances, 1 had 41 utterances and 1 had 28 utterances).
Developmental Sentence Score (DSS): DSS (Lee, 1974) was used to assess complexity of child syntax. Lee (1974) clearly deﬁnes what constitutes an utterance and recommends using 50 utterances to compute DSS. The mean number of utterances used to calculate DSS was 38 (range 11-50).
Type-Token Ratio (TTR), Number of Dzﬁerenz‘ Words (NDW) and Vocd: Lexical diversity was measured using TTR, NDW (Templin, 1957; Miller, 1981 ) and Vocd ( McKee et al 2000). Both TTR and NDW were calculated for each child and parent based on the same 50 utterances that had been used to calculate MLU. Given the number of limitations identiﬁed with both TTR and NDW (see McKee et al, 2000) Vocd was also used to assess lexical diversity. Vocd is a program that runs within CLAN and is regarded as being a more robust method of measuring lexical diversity as it is not sensitive to sample size and uses all of the data available.
Parents’ conversational participation was assessed using Fey's (1986) conventions for coding conversational acts. Utterances for which a verbal response or action was expected from the child were coded as requests for information (RQIN), requests for clariﬁcation (RQCL) or requests for action (RQAC). Statements and comments which required no action or response from the child were also identiﬁed. Fey distinguishes between comments and statements in this coding protocol. Since this therapy does not distinguish between these two acts and the distinction itself was not clear during analysis, statements and comments were amalgamated and referred to as Comments. Using CLAN, the frequency of occurrence of each type of conversational act was calculated, along with the total number of utterances made by each parent. The proportion of each typeiof conversational act was then computed.
Both parent and child mean length of turn (MLT) was assessed in terms of the mean number of utterances per turn. Mean length of turn has been described as providing an estimation of conversational participation with MLT increasing, as children become more skilful conversationalists (Sokolov and Snow, 1994). The proportion of child turns was also calculated to assess relative conversational participation of parents and their children.
To analyse changes in children and parents’ language (MLU, TTR, NDW and Vocd), a two factor mixed ANOVA (group X time) was used. Group (children VS. mothers and children vs. fathers) was the between-subject variable and time was the within-subject Variable. Main effects of group are not reported, as looking for overall differences between the children’s language and their parents’ language was not a purpose of this study. Main effects of time and interaction effects are reported to indicate changes over time and the direction of change for both the children and parents.
For DSS (only child data was analysed), a one-way repeated measures ANOVA was used to analyse change over time. See Table 1 for a summary of the results.
Table 1. Language variables ( p values reported; * denotes significant change)
No signiﬁcant main effects of time or interaction effects were obtained for child-mother data for any of the language Variables. Therefore there was no signiﬁcant change over time either in the children’s language when interacting with their mothers or in the mothers’ language.
No signiﬁcant main effect of time or interaction effects were obtained for the child-father data for TTR, NDW and Vocd. However an interaction effect approaching signiﬁcance was found for NDW (F[3,30] = 2.79, p =0.058), indicating that the trend was towards children increasing their N DW scores over time and fathers reducing their NDW scores.
No signiﬁcant main effect of time was obtained for the child-father data for MLU, but a signiﬁcant time X group interaction was obtained (F[3,30] = 3.78, p=0.02). Examination of the means indicated that the children increased their MLU and father reduced their MLU over time. Post hoc one-way ANOVA indicated that there was no signiﬁcant change between T 1 or T2 or between T1 and T3, but there was a signiﬁcant change between T1 and T4 (F(1,10) = 10.26, p =0.009). This change can therefore be attributed to the therapy.
No signiﬁcant change in the children’s DSS scores was found for the children when interacting with their fathers or when interacting with their mothers.
To analyse changes in children and parents’ MLT and total number of utterances spoken, a two factor mixed AN OVA (group X time) was used. As above, group (children VS. mothers and children VS. fathers) was the between-subject variable and time was the within-subject Variable and main effects of group are not reported. Main effects of time and interaction effects are reported to indicate changes over time and the direction of change for both the children and parents.
For proportion of child turns (only child data was analysed) a one-way repeated measures AN OVA was used to analyse change over time. See Table 2 for summary of results.
Table 2. Conversational participation: turn taking (p values reported; * denotes significant change)
No signiﬁcant main effect of time was found for MLT with child-mother data, but a signiﬁcant time x group interaction was obtained (F[3,30] = 8.39, p =0.00l). Examination of the means indicated that mothers reduced their MLT and children increased their MLT over time. Post hoc one- way AN OVA indicated that there was a signiﬁcant change between T1 and T2 (F[1,10] = 11.26, p =0.007), between T1 and T3 (F[3, 10] = 10.72, p = 0.008) and between T1 and T4 (F [l,10] = 28.44, p = 0.001). As a signiﬁcant change was obtained between T1 and T2, changes in MLT cannot be entirely attributed to the therapy.
A signiﬁcant main effect of time (F[3,30] = 4.33, p = 0.01) for MLT was obtained for child- father data indicating that there was a signiﬁcant change in MLT over time. A signiﬁcant time X group interaction (f[3,30] = 3.39, p :0.03) was also obtained for child-father data. Examination of the means indicated that fathers reduced their MLT and children increased their MLT over time. Post hoc one-way ANOVA indicated that there was no signiﬁcant change between T1 and T2, but that there was a signiﬁcant change between T1 and T3 (F[l,10] : 7.73, p=0.02) and between T1 and T4 (F[1,10] = 8.33, p=0.02). This change can therefore be attributed to the therapy.
No signiﬁcant main effect of time or interaction effects were obtained for child-father data for total number of utterances spoken. However with mother-child data a signiﬁcant main effect of time (F[3,30] = 3.752, p =0.02) was obtained indicating that there was a signiﬁcant change in the total number of utterances spoken over time. A signiﬁcant time X group interaction was also obtained (F[3,30] = 3.412, p=0.03). Examination of the means indicated that the total number of utterances spoken by mothers reduced and the total number of utterances spoken by the children increased over time. Post hoc one-way ANOVA indicated that there was no signiﬁcant difference between T 1 and T2 or between T1 and T3, but a signiﬁcant change between T1 and T4 (F[1,10] = 8.051, p = 0.02). This change can therefore be attributed to the therapy.
There was no signiﬁcant change in the proportion of children’s turns when interacting with their fathers (F[3,15] = 3.31, p :0. 128). However when interacting with their mothers, the proportion of the children’s turns signiﬁcantly reduced (F[3, 15] = 7.431, p = 0.04). Follow up analyses using t-tests (Bonferroni-corrected) indicated that there was no signiﬁcant change between T1 and T2, or between T1 and T3, but that there was a signiﬁcant change between T1 and T4 (t(5) =4.597, p =0.003). This change can therefore be attributed to the therapy.
RQIN, RQAC, RQCL and Comments
For RQIN, RQAC, RQCL and Comments, where only parents’ data was analysed a series of one-way repeated measures ANOVA was used to analyse change over time. See Table 3 for summary of results.
Table 3. Conversational participation: F ey’s (1986) conversational acts ( p values reported; * denotes signiﬁcant change)
Fathers signiﬁcantly reduced their %RQIN (F[3, 15] = 17.6, p=0.009). Follow-up analyses, using t-tests (Bonferroni-corrected) indicate that this change can be attributed to the therapy. There was no signiﬁcant change in the %RQIN between T1 and T2, but there was a signiﬁcant change between T1 and T3 (t(5) = 5.75, p : 0.003) and between T1 and T4 (t(5) = 11.503, p <0.00l).
The trend for mothers was also to reduce %RQIN, but results were just below signiﬁcance (F[3,15] = 6.375, p=0.053).
There was no signiﬁcant change in father’s %RQAC, but mothers signiﬁcantly reduced their %RQAC (F[3, 15] = 14.196, p = 0.01). Follow up analyses, using t-tests (Bonferroni-corrected) indicated that there was no signiﬁcant change between T1 and T2, but that there was a signiﬁcant change between T1 and T3 (t(5) = 3.916, p=0.005) and T1 and T4 (t(5) =3.636, p =0.007). This change can therefore be attributed to the therapy.
There was no signiﬁcant change in %RQCL for both fathers and mothers.
Although the pattern of change was towards both mothers and fathers increasing %Comments, this did not reach signiﬁcance for mothers and approached signiﬁcance for fathers (F[2,15] = 5.67,p = 0.06).
Normative Comparisons for NDW, MLU and DSS
All the children presented with differing language proﬁles pre- and post-therapy. Children’s MLU, DSS and NDW performance at T1 and T4 was compared with available normative data by Lee (1974), Miller (1981) and Templin (1957) respectively, with the following results.
Table 4. Children ’s language performance at T1 and T4 in relation to available normative data.
As can be seen from these results, at T1 two of the children (S1 and S5) presented with language measures within normal limits for their age or in advance of their age on some of the measures. One child (S3) presented with delayed MLU and NDW scores, but DSS scores within normal limits, and three children (S2, S4 and S6) presented with MLU and DSS scores within normal limits but a delayed N DW score.
For the individual children, between 7 and 9 months elapsed between T1 and T4. Post therapy a similar pattern of scores emerged. Apart from S2, S5 and S6, those children presenting with language skills within normal limits, advanced or delayed for their age at T1, showed the same proﬁle at T4, that is they continued to present with language skills within normal limits, advanced or delayed for their age at T4. However, for S2, his NDW score was delayed for his age at T1, but within normal limits at T4. For S5 and S6 their DSS scores were advanced for their age at T1, but within normal limits at T4. In addition, for S6, no normative data for MLU was available for his age at T4 (526 yrs), however his scores fell at least within the normal range for a child aged 5 years. Overall the mean change for the group was positive over the time period of the study. For MLU the mean observed change for the group was 0.88, for NDW was 16.3 and for DSS was 0.19.
This study was undertaken to see whether there is evidence for the argument that modiﬁcation of parents’ interaction styles during PCI therapy may have a detrimental effect on children’s language development.
Firstly, we shall look at the changes made by children and parents in their language over the time period of study. From the variables investigated, there was no signiﬁcant change in the mothers’ language over the course of the study as measured by TTR, NDW, Vocd and MLU. In addition, there was no signiﬁcant change in the father’s language on the measures of lexical diversity. Fathers did however signiﬁcantly reduce the length of their utterances over the time period of the study. Similarly, there was no signiﬁcant change in the children’s language when interacting with their mothers and no change in the children’s lexical diversity when interacting with their fathers. There was also no change in children’s complexity of syntax as measured by DSS. However the children did signiﬁcantly increase the length of their utterances when interacting with their fathers. The evidence in this study suggests that PCI does not result in the curtailment of children’s language functioning and indeed the increased MLU with fathers may indicate that there are some beneﬁts for language development. These ﬁndings also suggest that although parents may focus on reducing the level of their language to match that of the child, they do not reduce their lexical diversity, which is important for vocabulary development (Snow, 1995).
Comparison of children’s language performance pre and post therapy with available normative data was also made. Overall the mean change in the language scores of the children increased over the time period of the study. The individual children presented with differing proﬁles of language pre and post therapy. At the beginning of the study, three of the six children presented with language measures within normal limits or in advance of their age on some of the measures. One child presented with delayed MLU and NDW scores, but DSS scores within normal limits, and two children presented with MLU and DSS scores within normal limits but delayed NDW scores. Children categorised as having language skills within normal limits, advanced or delayed for their age at the beginning of the study were described in the same way at T4. However two of the children changed from being described as advanced for their age on the DSS to having language skills within normal limits. Overall however ﬁndings do not suggest that PCI impairs children’s language development.
There were some interesting changes observed in the parents’ conversational participation. Although there was no change in the proportion of parents’ utterances that were RQCL, mothers signiﬁcantly reduced the proportion of their utterances that were a RQAC and fathers signiﬁcantly reduced the proportion of their utterances that were RQIN. Similar ﬁndings for RQCL and RQIN were reported by Bonelli et al (2000). They proposed that this change may be due to the children becoming more active participants in the conversation following the therapy or that parents were better at eliciting language from their children and were less reliant on asking questions to achieve this. It could also be argued that parents reduced their RQAC and RQIN as part of the therapeutic process, in an attempt to encourage the children to take more lead in the conversation and play. Parents also often aim to increase the proportion of their utterances that are Comments as part of therapy, as comments do not necessitate a response and they result in increased shared focus of attention. While there was some evidence that this was occurring, the data did not reach signiﬁcant levels. Larger subject numbers may yield stronger results. Findings of this study suggest that parents are able to make the changes aimed for during PCI therapy and that these changes are maintained over time.
Although the children signiﬁcantly reduced the proportion of their turns when interacting with their mothers, the children’s MLT increased with their mother and their total number of utterances spoken increased, whilst the mothers’ reduced. The children were therefore taking less turns over time when interacting with their mothers, but they had more to say in each turn. There was no signiﬁcant change in the proportion of children’s turns when interacting with their fathers, no change in the total number of utterances spoken by either the children or fathers, but the children’s MLT increased and father’s reduced over time. The children were therefore taking as many turns with their fathers, but were again saying more in each turn.
In summary, the naturalistic data sample analysed in this study suggests that as a group, CWS continue to develop language over the period they are receiving PCI therapy and for a period of time post therapy. While there was no control group, baseline and comparative data was available and indicated that the children continued to score within the range that would be expected according to their ages and language level at the start of the study. There was therefore no strong evidence for the suggestion that PCI impedes language development in CWS. There was evidence that the parents were participating differently at the end of the study, using a more facilitative style of interaction, while maintaining their range of vocabulary. Further studies evaluating the changes parents make as part of the therapeutic process may help to identify the essential components of PCI therapy.
This study was undertaken with the support of Islington Primary Care Trust, who received a proportion of the funding from the NHS Executive. The views expressed in this paper are those of the authors and not necessarily those of the NHS Executive. We thank Tina Bory, James Au Yeung, Steve Davies, Marcin Szczerbinski, Tim Pring and all the staff and the families at the Michael Palin Centre for their support with this study.
Bernstein-Ratner, N. E. (1997a). Stuttering: A psycholinguistic perspective. In R. Curlee and G. Siegel (Eds.), Nature and treatment of stuttering: New directions. (2nd ed., pp. 99-127). Bos :on: Allyn & Bacon.
Bernstein Ratner, N., & Silverman, S. (2001) Parental perceptions of children’s communicative development at stuttering onset. Journal of Speech, Language, and Hearing Research, 43, 1252-1263.
Bishop, D., & Mogford, K. (1993). Language development in exceptional circumstances. Hove: Taylor and Francis
Bonelli, P., Dixon, M, Bernstein Ratner, N., & Onslow, M. (2000). Child and parent speech and language following the Lidcombe Programme of early stuttering intervention. Clinical Linguistics and Phonetics, 14, 427-446.
Conture, E.G. (2001). Stuttering: Its nature, diagnosis, and treatment. Boston: Allyn & Bacon.
Conture, E.G., & Melnick, K.S. (1999 . Parent-child group approach to stuttering in preschool children. In M. Onslow and A. Packman (Eds.), The handbook of early stuttering intervention. (pp.17-52) San Diego: Singular Publishing Group.
Felsenfeld, S. (1997). Epidemiology and genetics of stuttering. In R. Curlee and G. Siegel (Eds), Nature and Treatment of Stuttering: New Directions. (2nd ed., pp. 3-23). Boston: Allyn & Bacon.
Fey, M. (1986). Language intervention with young children. Needham, MA: Allyn & Bacon.
Guitar, B. (1998). Stuttering: An Integrated Approach to its Nature and Treatment (2nd Ed.). Baltimore: Williams and Wilkins
Huttenlocher, J ., Haight, W., Bryk, A., Seltzer, M., & Lyons, T. (1991). Early vocabulary growth: Relation to language input and gender. Developmental Psychology, 27, 236-248.
Law, J., Boyle, J ., Harris, F., Harkness, A., & Nye, C. (1998). Screening for speech and language delay: A systematic review of the literature. York: NHS Centre for Reviews and Dissemination.
Lee, L. (1974). Developmental sentence analysis: A grammatical assessment procedure for speech and language clinicians. Evanston, IL: Northwestern University Press.
MaCWhinney, B. (2000). The CHILDES project: Tools for analysing talk (3rd ed.). Mahwah, NJ: Erlbaum.
McKee, G., Malvern, D., & Richards, B. (2000). Measuring vocabulary diversity using dedicated software. Literary and Linguistic Computing, 15, 323-338.
Miller, J & Chapman, R.S. (1981). The relation between age and mean length of utterances in morphemes. Journal of Speech and Hearing Research, 24, 154-161.
Millard, S., & Nicholas, A. (2002). Evaluazing the effectiveness of therapy with young children who stammer. Paper presented at the 2002 meeting of the Special Emphasis Panel in Treatment Efﬁcacy afﬁliated to the American Speech and Hearing Association.
Miller, J. ( 1981). Assessing language production in children: Experimental procedures. Baltimore, MD: University Park Press.
Miles, S. & Bernstein Ratner, N.- (2001). Parental language input to children at stuttering onset. Journal of Speech, Language, and Hearing Research, 44, 1116-1130. 116 Theory, research and therapy in ﬂuency disorders
Nicholas, A & Millard, S. (in press). A study to investigate the effectiveness of parent-child interaction therapy. Proceedings of the Sixth Oxford Dysfluency Conference.
Nippold, M. (1990). Concomitant speech and language disorders in stuttering children: A critique of the literature. Journal of Speech and Hearing Disorders, 55, 51-60.
Onslow, M., & Packman, A. (1999). The Lidcombe Programme of early stuttering intervention. In Bernstein Ratner and E. C. Healey (Eds.), Stuttering research and practice: Bridging the gap. Mahwah, NJ: Lawrence Erlbaum
Rustin, L., Botterill, W., & Kelman, E. (1996). Assessment and Therapy for Young Dysfluent Children: Family interaction. London: Whurr Publishers.
Smith, L. (1998). Predicting communicative competence at 2 and 3 years from pragmatic skills at 10 months. International Journal of Language and Communication Disorders, 33, 127-148.
Snow, C. (1986). Conversations with children. In P.Fletcher and M. Garman (Eds.), Language Acquisition (2nd ed., pp 69-89). Cambridge, UK: Cambridge University Press. '
Snow, C.E. (1995). Issues in the study of input: ﬁnetuning, universality, individual and developmental differences, and necessary causes. In P.Fletcher and B. MacWhinney (Eds.), The handbook of child language. Oxford, UK: Blackwell Publishers.
Sokolov, J .l., & Snow, C.E. (Eds.) (1994). Handbook of research in language development using CHILDES. Hillsdale, NJ: Lawrence Erlbaum Associates.
Starkweather, C.W, Gottwald, S.R., & Halfond, M.M. (1990). Stuttering prevention: A clinical method. Englewood Cliffs, NJ: Prentice-Hall.
Templin, M. (1957). Certain language skills in children: Their development and interrelationships. Westport, CT: Greenwood Press.
Wall, M.J., & Myers, FL (1995). Clinical management of childhood stuttering, (2nd ed.). Austin, TX: Pro-Ed.