Investigation of the articulatory patterns of young adults with Down syndrome using electropalatography

Tongue movements for speech in three adults with Down syndrome and one normal speaking control were investigatedusing electropalatography (EPG) and diadochokinetic rate (DDK)techniques. The subjects with Down syndrome all showedsubstantially larger areas of tongue-palate contact comparedwith the control subject for alveolar sounds /t, d, n, l/, anddifferences in fricative articulation. Undershoot (incompleteclosure) was a feature of velar articulations /k, g/ in the subjectswith Down syndrome. Longer closure duration, longer consonanttransition times within clusters and consistently slower DDKrates were also noted. Asymmetry and variabilitywere evident in two of the subjects with Down Syndrome.The results support the hypothesis that people with Downsyndrome have difficulties with coordinating the rapid tonguemovements necessary for clear speech with impaired tonguemotility and motor programming as well as phonological delay.Implications for treatment are discussed.

Download PDF

Hamilton, C. (1993) Investigation of the articulatory patterns of young adults with Down syndrome using electropalatography. Down Syndrome Research and Practice, 1(1), 15-28. doi:10.3104/reports.7


Children and adults with Down syndrome have a higher incidence of speech (pronunciation) problems than any other group of people with severe learning difficulties (Schlanger and Gottsleben, 1957; Blanchard, 1964). In addition, their spontaneous speech and general intelligibility (speech clarity) lags behind their other language skills (Lenneberg, 1967; Ryan, 1975; Rondal, 1978). Intelligibility is a problem even in older children with Down syndrome who have more advanced syntax (Bray and Woolnough, 1988).

Although many reasons for the speech difficulties in people with Down syndrome have been proposed, it is far from clear which of many possible factors are actually involved. Intervention approaches have varied over the years, reflecting trends in both linguistic theory and in therapy. Even after periods of intervention, intelligibility often remains the greatest barrier to communication outside the family. Buckley and Sacks (1987) found that over half the adolescent girls and about 80 % of the adolescent boys in their survey were rated by their parents as unintelligible to strangers. Their intelligibility rating may have included issues of language skill difficulty (i.e. in forming sentences) but in my experience the major hindrance to effective communication in people with Down syndrome is poor speech clarity (pronunciation).

It seems likely that effective intervention has not been offered to this population because of a lack of objective assessment which in turn has led to unclear diagnoses of these speech problems.

This study aims to investigate the use of one objective measurement tool of speech articulation in an attempt to shed light on the complex nature of these speech difficulties.

Three main types of speech difficulty are possible in any person developing speech: dysarthria, dyspraxia and phonological delay/disorder. Dysarthria is due to muscle weakness and results in the slowing down or difficulty in coordination of the articulatory movements of the tongue and lips during speech. These "articulatory difficulties" can cause distortion of the speech sounds either by changing a sound slightly (e.g. slurring) or by changing a sound so it sounds like a different sound (e.g. "s" becomes "t").

Dyspraxia is thought to be due to a problem in the planning and execution of the sequence of movements needed for speech. It can result in inconsistent attempts at sounds (i.e. a word is said differently every time) and difficulty in sequencing consonant and vowel sounds in a word.

Phonological difficulties involve the delay or disordered development of the use of speech sounds and generally result in the simplification of words (e.g. "poo" for "spoon"). There are recognized patterns of development in the acquisition and use of sounds in children who are learning to talk. Phonological delays are said to occur when a developmental pattern is observed in a child's speech but the child is at an immature stage. Disorders are said to be when patterns not found in development are found.

From a review of the current and past literature, it seems that the speech difficulties of people with Down syndrome have been attributed by different researchers to: macroglossia (large tongue), hypotonia (muscle weakness), phonological delay/disorder (delayed development in the use of speech sounds), or a motor programming deficit (dyspraxia) (difficulty in planning and executing the sequence of movements needed to produce speech).

a) Macroglossia (Large tongue)

Both early and relatively recent literature attributes the speech difficulties in people with Down syndrome to the typically large or protruding tongue and small oral cavity found in most of them (Benda, 1949; Engler, 1949; Strazzula, 1953; Olbrisch, 1982). Many plastic surgeons who perform tongue reduction operations (partial glossectomy) on people with Down syndrome have claimed improvements in speech clarity as a result (Lemperle, 1980; Rozner, 1983; Wexler, 1986).

Others have questioned the effect of tongue size (Lenneberg, Nichols and Rosenberger, 1964; Rosenberg, 1982) including some plastic surgeons who have not found any significant effect on speech following tongue reduction operations (Margar-Bacal, Witzel and Munro, 1987; Klaiman, Witzel, Margar-Bacal and Munro, 1988 ; Parsons, Iacono and Rozner, 1987). In addition, some researchers, including Ardran and his colleagues (1972), the only researcher who used objective X-ray measures, have suggested that macroglossia in Down syndrome is an illusion due to the small size of the oral cavity; the need to mouth breathe because of a small nasopharynx (the upper part of the throat) and frequent upper respiratory tract infections; enlargement of the lingual tonsil in some cases; hypotonia of the tongue and lower lip; and forward position of the mandible with underdeveloped nasal and maxillary bones (the central part of the face) (Cohen and Winer, 1965; Krmpotic-Nemanic and Schulz, 1970; Ardran, Harker and Kemp, 1972; Purdy, Deitz and Harris, 1987; Yarom, Sherman, Sagher, Peled, Wexler and Gorodetsky, 1987 ).

A large tongue or a forward tongue position would not of itself guarantee a decrease in speech clarity - normal speech can be found with any anatomical deviation if the individual has learned to compensate for the defect (Bloomer, 1957).

b) Hypotonia (Muscle weakness)

There is strong evidence for hypotonia in Down syndrome muscles (McIntire and Dutch, 1964). Yarom et al. (1987) found evidence of degeneration of nerve endings and changes in the end plates of neuromuscular junctions in tongue tissue removed from people with Down syndrome in surgery. They suggest that this may be responsible for hypotonia in their tongues. Hypotonia in the tongue would lead to a "dysarthric" speech difficulty.

c) Phonological delay/disorder

Many studies over the past 15 years have focussed on the phonological development of people with Down syndrome (analysing their speech purely in terms of the acquisition and use of sounds) and have ignored or openly denied the possibility of articulatory difficulty (e.g. Dodd, 1984). Some have claimed speech development shows only delay (Stoel-Gammon, 1980; Rosenberg, 1982; Smith and Stoel-Gammon, 1983; Bleile and Schwarz, 1984; Rondal, 1988). Others feel there is disordered development (Dodd, 1975, 1976; Rosin, Swift, Bless and Vetter, 1988; Sommers, Patterson and Wildgen, 1988).

When errors are attributed to articulatory rather than phonological difficulties (e.g. Borghi, 1990), it is not clear how this decision was made. It seems that criteria are not based on objective measurement but on the theoretical background of the researcher and subjective auditory analysis (i.e. listening to the speech sample and then writing down phonetically what is thought to have been heard). That auditory assessments are far from accurate has already been shown (Oller and Eilers, 1975). This makes the need for objective measures of the speech of people with Down syndrome even more pressing.

It is important to note that even if the primary difficulty is one of articulatory control, compensatory strategies may give rise to whole sound changes which may then appear to be phonological simplifications (Crystal, 1981). For example, "skirt" could be produced "sirt" because of articulatory limitations or phonological simplification processes.

d) Motor programming deficits

Dodd (1976) proposed that people with Down syndrome have a central motor (muscle control) deficit resulting in problems with the planning and execution of fine motor (muscle) movements. Rosin, Swift, Bless and Vetter (1988) found that young adults with Down syndrome had more difficulty with speech production as demands for sequencing increased. They argue that difficulty with sequencing underlies both oral motor control and language, and suggest a pervasive processing problem in Down syndrome. Frith and Frith (1974) found evidence of impairment in the ability of children with Down syndrome to plan fine motor hand movements and this difficulty in establishing long-term motor programs for skilled movements has been related to cerebellar dysfunction in Down syndrome (Penrose 1963).

Difficulty in the planning and sequencing of sounds for speech would lead to a "dyspraxic" speech difficulty.

EPG data

The need for more objective data on the articulation patterns of people with Down syndrome is clear. The objective measurement tool chosen for this study was Electropalatography (EPG).

EPG is a technique developed over the past 30 years. It involves the use of an artificial palate made of acrylic, in which is embedded 62 silver electrodes arranged in 8 horizontal rows (see Figure 1.0).

Photograph of an EPG palate

Figure 1.0 Photograph of an EPG palate showing the silver electrodes and two wire clips to fit around the molar teeth

Each palate is tailor made for each subject and covers the hard palate so that any tongue contact with the electrodes on the palate during speech articulation can be recorded (see Figure 1.1). The electrode positions are determined by anatomical dimensions of the palate so that different palate sizes and shapes can be compared with reference to specific rows and electrode positions (Hardcastle, 1989).

Figure 1.1. A computer generated display of tongue contacts (indicated by zeros) during the closure phase of /t/. This frame represents 10ms. Reference zones for the electrode positions and row numbers of electrodes are shown on the right.

From the EPG investigations already done, a range of normal patterns of tongue-palate contact for different sounds have been described (Hardcastle, 1984; Fletcher 1989; Hamlet, Bunnell and Struntz, 1986; Stone, Raphael and Shawker, in press). EPG has also been used to analyze the articulatory patterns of children and adults who have presented with a range of speech difficulties including dysarthria and dyspraxia (Hardcastle and Morgan, 1982; Hardcastle, Morgan Barry and Clark, 1985 and 198 7). The aim in the present study was to look for common patterns in the Down syndrome sample and compare them with existing EPG data of normal and disordered speech.

In order to attempt to differentiate between the different levels of breakdown proposed in the literature, a range of different monosyllabic and polysyllabic words/phrases were used and a test of Diadochokinetic rate (DDK) was completed. DDK testing involves counting the number of times certain sounds (e.g. p-p, p-t-k) can be repeated per minute. Difficulty in timing and sequencing of articulatory movements found in the polysyllabic words and DDK test would indicate difficulties at the motor programming level (dyspraxia).

Researchers have also shown interest in the processing skills of people with Down syndrome (their ability to take information from what they receive from their senses), claiming a deficit in auditory sequential processing compared with visual processing (O'Connor and Hermelin, 1965; Bilovsky and Share, 1965; Nakamura, 1965; Rohr and Burr, 1978; Rosin et al, 1988; Snart, O'Grady and Das, 1982). The use of visual systems, such as signing and reading, in teaching and facilitating speech and language learning in Down syndrome is widespread. Assuming that people with Down syndrome are able to convert letters and letter groupings to speech sounds (the "grapheme-phoneme" conversion route) (Ellis and Young, 1988), read words should be clearer than spontaneous output as information about the sequential order of each sound is given. Imitated words should not be as clear as read words if auditory processing deficits exist. My clinical experience supports this theory.

In this study, the same target words were elicited in three different conditions: picture naming (spontaneous recall), imitation (auditory input) and reading (visual input).



Three young adults with Down syndrome (Nikki, Michael and Paul) participated in this study with one non-disabled adult for comparison.

The young people with Down syndrome were selected on the basis of:

  1. having adequate comprehension to follow explanations of what was to be required of them and therefore be able to agree to take part.
  2. "typical" speech patterns for people with Down syndrome.
  3. differing levels of phonological development to allow analysis of fricatives, affricates, consonantal clusters as well as plosives.
  4. expressive language output of at least two word sentences to allow the use of phrases in the EPG target word lists.

The person recorded for comparison (Fiona) was a research speech therapist.

Speech, language and hearing tests were carried out on Nikki, Michael and Paul. These included: the TROG, which tests understanding of language and grammar; the LARSP, which is an analysis of expressive language (sentences) and PACS which is an analysis of phonology (speech sound acquisition and use). The results are shown below:

Speech and language scores

Name Age Hearing levels TROG LARSP PACS
Michael 17 High frequency
5yrs III
Nikki 17 NAD 6yrs III/IV
Paul 20 NAD 4yrs II
TROG (Bishop, 1982); LARSP (Crystal, Fletcher and Garman, 1976); PACS (Grunwell, 1985).


  1. EPG palates were manufactured for each of the people in the study. In addition, practice palates (without electrodes) were made for Nikki, Michael and Paul. This was so that they could get used to wearing the palates before the EPG readings were taken.
  2. 54 target words were used to provide data for EPG analysis. The word list was taken from an EPG study by Hardcastle, Morgan Barry and Clark (1987) with the addition of 12 polysyllabic words/phrases (see Appendix).
  3. Black and white line drawings of all the target words were made on postcard size white card. Some use of color was allowed when a color adjective was required for the target response. No written words were visible on the pictures. For the reading condition, target words were hand written onto the card in black ink with lower case letters.
  4. The EPG hardware and software used in this study was developed by researchers at Reading University and IBM Scientific Center, Winchester ( Hardcastle, Jones, Knight, Trudgeon and Calder, 1989 ).


  1. Nikki, Michael and Paul wore their practice palates for an hour per day in the fortnight prior to their EPG analysis. In addition, practice palates were worn for the 2 hours immediately before the actual data collection. Fiona was an experienced palate user and so did not require these adjustment periods (Hamlet and Stone 1978).
  2. Nikki, Michael and Paul received four visits from the researcher to ensure recognition of the picture items and written words and to ensure familiarity with the procedures to be adopted in the data collection.
  3. The target words were presented in three different conditions in this order: (1) as spoken words for imitation, (2) as pictures, (3) as written words. The items were placed in random order within the three conditions. In the third condition, only those words that were able to be read (as assessed on the pre-EPG visits) were shown. One of the young people with Down syndrome was unable to read, another could only read 34 out of the 54 words, the third could read all the words. EPG patterns for all these test items were stored on disc for analysis.
  4. A DDK rate measurement was taken for Nikki, Michael and Paul using the time-by-count technique (Fletcher 1972).


Nikki, Michael and Paul's EPG patterns for different sounds were studied and compared with those of Fiona and the normal patterns described by Hardcastle (1984) and Hardcastle et al. (1987). The Reading University EPG software includes the facility to list the number of tongue-palate contacts across three zones of the palate: alveolar, palatal and velar (please see Figure 1.0).

These contact totals were taken at the point of maximum tongue-palate contact during each sound. The point of maximum contact is a measure that has been used widely by other researchers (Hiki and Itoh, 1986; Hardcastle et al, 1989; Fletcher, 1989) and although it is an arbitrary point, it provides a useful and easily identifiable landmark in the articulation of the sound, for comparison across subjects.

A summary of the tongue-palate contact patterns for each group of sounds analyzed is given:

Alveolar stops (/t,d,n/)

The most striking finding was the amount of palatal contact for these sounds in Nikki, Michael and Paul. All showed at least one row of contacts in the palatal zone in addition to full alveolar zone contact. Compare this with Fiona who only ever showed contact with the anterior two alveolar rows when articulating t, d or n sounds (see Figure 1.2 and Table 1). The actual pattern of tongue-palate contacts, traced onto a two dimensional representation of each subject's palate is illustrated in Figure 1.2. The pattern shown is that occurring during the point of maximum tongue-palate contact for /d/ in "dart". The actual EPG printout of the source frame is also shown. The use of more palatal and velar zone contacts is clearly visible in Nikki, Michael and Paul's EPG patterns.

Fig1.2.gif (24231 bytes)

Figure 1.2. Maximum contact patterns for /d/ in "dart"

Table 1. Summary of average maximum contacts for each subject in palatal zone for alveolar sounds /t,d,n/

Name Alveolar Palatal Velar
FIONA 13.62 9.33 7.04
NIKKI 14.00 20.37 10.78
MICHAEL 13.28 23.35 13.83
PAUL 13.29 17.65 8.43

Velar stops (/k,g/)

Nikki, Michael and Paul again showed differences from Fiona, with inadequate and incomplete tongue-palate contact (undershoot) being the most noticeable feature. However, Michael used almost total contact with the palate for /ng/. (See Figures 1.3 and 1.4 and Table 2).

Fig1.3.gif (22200 bytes)

Figure 1.3. Maximum contact patterns for /k/ in "keys"

Fig1.4.gif (21456 bytes)

Figure 1.4. Maximum contact patterns for /ng/ in "singing"

Table 2. Summary of average maximum contacts for each subject in each palatal zone for the velar sounds /k,g/

Name Alveolar Palatal Velar
FIONA 0 3.14 17.22
NIKKI 0 0 14.38
MICHAEL 0.21 0.47 12.04
PAUL 0.22 3.58 11.10

Laterals /l/

Again more palatal and velar contacts were evident in the young adults with Down syndrome compared with Fiona. Michael also showed less alveolar contact with the main constriction being at rows 3/4. His /l/ articulations showed a great deal of variability (see Figure 1.5).

Fig1.5.gif (22811 bytes)

Figure 1.5. Maximum contact patterns for /l/ in"leaf"

Asymmetry was a feature of /s/ and /sh/ in Paul and Michael, with increased palatal zone contacts for /s/ and a similar pattern for /sh/ (they were difficult to distinguish auditorily as well). In Michael the constriction was made at rows 3/4 and too much tongue-palate contact (overshoot) was a frequent feature. (See Figures 1.6 and 1.7).

Fig1.6.gif (24223 bytes)

Figure 1.6. Maximum contact patterns for /s/ in "seed" Fricatives (/s,sh/)

Fig1.7.gif (22912 bytes)

Figure 1.7. Maximum contact patterns for /sh/ in "shoe"

Overlapping articulations

In consonant clusters and syllable boundaries where consonant coarticulation (overlapping articulation) was expected to be found, only Fiona showed consistent patterns. Nikki, who had the clearest speech showed no overlapping articulations at all. For example "tickling" was produced as "tickeling" by Nikki and Michael (see Figure 1.8). Where faster transitions were seen, it was often at the "expense" of one of the sounds, i.e. assimilation occurred (see Michael's /sk/ in "skirt", Figure 1.9).

Fig1.8a.gif (101444 bytes)
Fig1.8b.gif (106115 bytes)

Figure 1.8. EPG Patterns for/kl/ in "tickling"

Fig1.9.gif (138195 bytes)

Figure 1.9. EPG patterns for /sk/ in "skirt"

Polysyllabic words

In Michael and Paul, these often showed such a degree of assimilation and simplification that the EPG patterns were not easily analysable. Nikki showed no coarticulatory patterns even in longer phrases and also tended to simplify by omitting sounds. She took longer to say the phrases than the control subject and showed longer closure durations and approaches to each sound.


Some apparent differences in the duration of initial plosives was also evident in the young adults with Down syndrome. Nikki in particular showed very long durations (see Table 3). Paul showed a great deal of variability in closure durations.

Table 3. Average closure duration of initial alveolar plosives (/t,d,n/)

Name Average duration
(number of 10ms frames)
across conditions
FIONA 9.17 frames
NIKKI 30.75 frames
MICHAEL 13.22 frames
PAUL 38.00 frames

DDK rates

These were significantly slower than the norms given. The experimental subjects found it extremely difficult to maintain a steady repetition of these sounds and had particular difficulty with the /p-t-k/ sequence (see Table 4).

Table 4 DDK rates for Down Syndrome Subjects

Name /p/ /t/ /k/ /f/ /l/ /pt/ /pk/ /tk/ /ptk/
NIKKI 9.0 7.3 10.5 7.0 6.0 13.5 16.0 16.0 27.0
MICHAEL 9.2 7.3 9.2 8.9 6.5 9.0 8.5 20.0 19.0
PAUL 8.0 9.0 7.0 8.3 9.0 9.1 14.0 13.0 *
"Norms" 4.8 4.9 5.5 5.5 5.2 7.3 7.9 7.8 10.3
KEY: 1. The scores represent the time in seconds required for 20 repetitions of single syllables, 15 repetitions of bisyllables (e.g. /pt/) 10 repetitions of /ptk/.
2. Each sound was produced as a CV (consonant-vowel) utterance - the schwa vowel (as in the) following the consonant.
3. Norms given are for 6 year olds (Fletcher 1972) - see text for explanation.
* Paul was unable to say this sequence of sounds even after repeated prompting.
Michael's /tk/ was actually /t-t/ for more than half of his attempt.

Differences between conditions

The EPG patterns for the different sounds did not differ across the three conditions (imitation, reading and spontaneous speech) for any of the subjects in the study.

Individual differences

Nikki had the clearest speech. She also spoke very slowly - this could have been a compensatory strategy, but it is interesting to note that she also had very slow approach phases to her sounds: compare her approach to closure for /n/ in "knot" with that of the others (see Figure 2.0). It is unlikely that this is a learned strategy - it may be indicative of difficulty in executing rapid tongue movements or indicate some underlying dysfluency.

Fig2.0a.gif (94885 bytes)
Fig2b.gif (275983 bytes)
Fig2c.gif (239141 bytes)

Figure 2.0. EPG patterns for "knot"

Michael often released his closures atypically, in anterior to posterior direction (tongue tip/blade first). Michael showed the largest areas of contact of all the subjects. He also appeared to have the largest tongue (from observation) and the shallowest palate. In addition, he had the most difficulty in adjusting to the EPG palate. These factors may all be related to the EPG patterns found in his sample.

Paul's speech was the least intelligible. He had a tendency to place the middle of his tongue up to the hard/soft palate areas during speech (particularly strange as he had a very steep, high palate). (See Figure 2.1)

Fig2.1.gif (280313 bytes)

Figure 2.1. Paul's EPG printout for "leg" showing the persistent line of velar contacts (see text for explanation)

Phonological processes

Developmental phonological patterns were also found in the data: final consonant deletion (e.g. "see" for "seed"), consonant harmony (e.g. "gog" for "dog"), fronting ("fis" for "fish"), stopping of fricatives ("dot" for "dolls"), cluster reduction (e.g. "bak" for "black"). However these can not account for the articulatory differences described above.

Summary of results

In brief, the young adults with Down syndrome shared the following differences from Fiona (whose EPG patterns were consistent with the normal data (Hardcastle op.cit.)):

In Michael and Paul, asymmetry and variability were features of fricative articulation.


Various theories have been proposed regarding the level of impairment in the speech production processes of people with Down syndrome.

The results in this study lend some support to the theory of an articulatory level of impairment which could result from hypotonia.

The increased tongue-palate contact patterns found in this study were also found in dysarthric children by Hardcastle (1987).

Movement of the tongue body up and back to the velar area for /k,g/ sounds also appeared to be slower in Nikki, Michael and Paul compared to Fiona. /k,g/ sounds were produced as fricatives (as in the Scottish "loch") or omitted. This "undershoot" is also a feature of dysarthric speech (Hardcastle, 1985).

Overshoot for fricatives (too much tongue-palate contact) resulting in complete closure, evident in Michael's speech, was also found in Hardcastle's 1985 sample.

The absence of any adequate auditory or articulatory difference between /s/ and /sh/ which is common in dysarthria was also evident in Michael and Paul.

The absence of coarticulatory patterns in consonantal clusters is another feature of dysarthric speech (Hardcastle, 1989) and was also found in Nikki, Michael and Paul. Presumably the difficulty of moving both the tip/blade (front) and the body (back) of the tongue to different places on the palate at the same time requires more accurate control than is possible in hypotonic tongues (Butcher, 1989).

It is also possible that some of the immature phonological patterns that were found in Nikki, Michael and Paul's speech may have been a by-product of limitations in tongue movement rather than evidence of delayed phonological maturation.

In all of the young adults with Down syndrome, there were some alveolar sounds where the tongue tip actually protruded forward from the incisor teeth during their articulation. For these sounds, the tongue tip was not used at all in articulation. As the somewhat larger area immediately posterior to the tip (the tongue "blade") was presumably used instead, this alone might account for the increased areas of palatal contact. This may lend credence to the macroglossia argument - that the larger size and more forward tongue position are the reasons for poor speech clarity in people with Down syndrome (Fucci and Petrosino, 1981).

However, one could also question why the tongue tip was not used during speech - in my clinical experience of other children with speech difficulties, a tongue tip protrusion pattern at rest does not usually affect articulation of alveolar plosives - /t,d,n/ (i.e. the tongue tip is still used in articulation). It must also be noted that for the alveolar sounds where the tongue tip was inside the mouth, the pattern of more electrode contact was still evident. This suggests some difficulty in controlling the tongue tip itself. This finding would also support the hypotonia argument.

Difficulties in motor programming for speech articulation would lead to dyspraxic type abnormalities in the timing and sequencing of movements. These are also evident in the data collected.

Dyspraxic speakers show long closure durations (Hardcastle, 1985) - also found in the young adults with Down syndrome in this study. In addition, Hardcastle noted slow approach and release phases for closures in his dyspraxic subjects - a feature of Nikki's EPG data also.

Abnormally long transitions between different elements in consonant clusters, another feature of dyspraxic speech, is also found in Nikki's and Michael's data.

The great difficulty experienced by Nikki, Michael and Paul in rapidly repeating polysyllabic Consonant-Vowel sequences in the DDK test is also indicative of motor sequencing (dyspraxic) problems (Henry 1990). The subjects' erratic and arrhythmic production of these sequences is also typical of dyspraxic difficulty (Henry 1990).

Asymmetry and variability, particularly noted in the /s/ and /sh/ articulations of Michael and Paul are also features of dyspraxic speech (Johns and Darley, 1970; Hardcastle, 1989 and personal communication).

The three different conditions in this study did not appreciably affect the performance of any of the subjects. This may have been due to practice effects (the same words were used in all conditions and had been practiced beforehand). In addition, specific training in using the grapheme-phoneme route for word recognition may have been necessary before beneficial effects would be seen.

From the results discussed above, it seems that there is evidence in these young adults with Down syndrome, for breakdown both in the ability to move the tongue and lips in order to articulate speech sounds accurately and in the ability to select, plan and sequence the sounds needed. These are in addition to phonological difficulties. All of these factors would contribute to making the speech of people with Down syndrome unintelligible.

There are a number of clinical implications of these findings.

In the past there has been a tendency to focus on intervention to improve phonological rather than articulatory skills. It seems clear that tongue motor coordination must be an area for work as well as encouraging phonological development. There is a need for further studies evaluating the effects of any possible treatment methods (e.g. Proprioceptive Neuromuscular Facilitation (Langley and Darvill, 1979). Perhaps EPG could be used to measure any articulatory changes.

The practice of removing part of the tongue, widely heralded as a technique to improve cosmetic appearance and speech is certainly contra-indicated by these findings. As it is clear from the results of this study that the speech difficulties of people with Down syndrome can not be attributed purely to macroglossia, removing weak tongue muscle tissue would seem at the very least an ineffective treatment and certainly inadvisable.

The use of EPG in treatment has been documented (Morgan Barry, 1989) and it may be an effective way of providing visual feedback for this client group. Certainly the experimental subjects showed considerable interest and understanding of their tongue-palate patterns shown on the visual display.

Further research is needed into what happens to articulation and phonology in spontaneous speech and in longer utterances. There is evidence for increased error rate in spontaneous speech in people with Down syndrome (Sommer, Patterson and Wildgen, 1988).


In this study, the young adults with Down syndrome showed evidence of impaired muscular control of the tongue resulting in slow movements, difficulty in tongue tip control and coarticulation. Difficulty in planning sequences of articulatory actions was also found, resulting in long closure durations, long transitions between consonants within clusters, erratic DDK output, asymmetry and variability in some articulations.

It is clear that the speech difficulties experienced by people with Down syndrome are complex. This study suggests that they are due to impairments at virtually every level of the speech production process - selecting the sounds (phonological), planning their sequence (motor programming) and carrying out the movements of the tongue and lips (articulatory). Assessment and treatment must therefore reflect this complexity and clinicians must be prepared not only to work at these different levels but also to evaluate which treatment approaches are effective for the different impairments. By this means, it is hoped that people with Down syndrome will be able to communicate more intelligibly and consequently enjoy more opportunities for an independent life.


Target words used for EPG analysis

  • dart
  • lamb
  • cot
  • deer
  • leg
  • chain
  • shark
  • key
  • dolls
  • leaf
  • book
  • well
  • car
  • girl
  • beak
  • knot
  • sun
  • mouse
  • fish
  • zoo
  • sheep
  • brush
  • seed
  • shop
  • bush
  • shoe
  • racer
  • kitkat
  • headlight
  • clock
  • tractor
  • catkin
  • weekday
  • tickling
  • deckchair
  • bookshop
  • star
  • box
  • bikeshop
  • fishcake
  • hats
  • squashkit
  • skirt
  • ghostbusters
  • scooby-do
  • soldiers
  • climbing
  • recordplayer
  • white kitten
  • black jumper
  • bird singing
  • cars crashing
  • dogs eating
  • red kettle
  • Kylie Minogue
  • Jason Donovan



    This research was supported by a grant from the College of Speech and Language Therapists. The author wishes to thank Professor W J Hardcastle and Fiona Gibbon at the Department of Linguistic Science, Reading University, for their help and advice in carrying out this study.