Using ultrasound tongue imaging to support the phonetic transcription of childhood speech sound disorders

Figures & data

Table 1. Error types, descriptions, example references and expected IPA transcriptions.

	Error Type	Ultrasound description^1	Example population/reference	Image panel in Figure 1	Possible IPA transcriptions
Imageable errors	1. Increased contact	Raising of tongue body and tip/blade towards or in contact with the hard palate	Functional Articulation Disorders (Gibbon, 1999)	Panel A	Simultaneous alveolar + postalveolar + palatal
	2. Retraction	Alveolar target retracted to velar or palatal	CAS (Cleland, Scobbie, & Wrench, 2015)	Panel B	Velar or palatal
	3. Fronted	Posterior target is fronted to palatal, postalveolar, or alveolar	Phonological disorder (Cleland et al., 2017)	Panel C	Alveolar, postalveolar, or palatal
	4. Complete closure	No visible groove (in the coronal ultrasound view)	CAS or residual speech sound error (Lundeborg & McAllister, 2007)	Panel D	Any lateral sibilant
	5. Open pattern	Uvular or pharyngeal articulation, or undershoot	CAS (Cleland, Scobbie, Roxburgh, Heyde, & Wrench, In Press)	Panel E	Uvular or pharyngeal consonant, or “lowered” diacritic
	6. Double articulation	Simultaneous production of two consonants: normally alveolar-velar	CAS (Hardcastle et al., 1987)	Panel F	Any double articulation
	7. Increased variability	Different tongue shape per repetition	CAS (Hardcastle et al., 1987)	Dynamic analysis required	Different transcriptions across repetitions
	8. Abnormal timing	Misdirected articulatory gestures of release of articulators with abnormal timing	CAS (Case & Grigos, 2016)	Dynamic analysis required	Any diacritic denoting timing, such as lengthening
	9. Retroflexion	Tongue tip retroflexion during any non-retroflex target	CAS (Cleland et al., 2017)	Panel G	Any retroflex consonant
	10. Coronal or dorsal production	Labial targets produced as coronal or dorsal consonants	A variety of populations (e.g., phonological impairment; Dodd & Lacano, 1989)	Panel H	Alveolar, postalveolar, palatal or velar
Non-imageable	11. Non-ultrasound error	-	A variety of populations	-	Errors unrelated to lingual position, such as voicing

^1.Based on descriptions from Cleland et al. (2020).

Figure 1. Example ultrasound images for various error types.

Table 2. Speakers’ demographic information.

Speaker	Sex	Age^a	Speech Disorder Reported by SLT^b	Languages Spoken at Home	Additional Information Reported by SLT^b
LN01	M	8;2	Phonological disorder	English	-
LN02	F	6;3	Articulation disorder	English	-
LN03	M	9;2	Articulation disorder	English	-
LN04	M	10;9	Articulation disorder	English	-
LN07	M	6;3	Motor speech disorder	English	Poor coordination
LN08	F	6;3	Phonological disorder	English	-
LN09	M	5;11	Inconsistent phonological disorder or CAS	English	-
LN10	M	6;4	Inconsistent phonological disorder	English	-
LN11	F	5;2	Diagnostic process ongoing	English	-
LN12	M	8;2	Motor speech disorder	English	Born at 31 weeks. Fine motor difficulties
GR01	M	9;1	Articulation disorder	English	-
GR02	M	9;2	Inconsistent speech sound disorder, query CAS	English	DLD, swallowing difficulties
GR03	M	8;11	Residual speech sound errors	English	Previous surgery for anterior tongue tie
GR04	M	6;2	Articulation disorder	English	-
GR05	M	7;5	Articulation disorder	English	-
GR06	M	7;5	Articulation and phonological disorder, with possible coordination difficulties	English	Difficulties with grammar
GR07	M	7;11	Articulation disorder	English	-
GR08	M	6;0	Possible CAS	English	-
GR09	M	12;11	Phonological disorder	English	Global developmental delay
GR11	M	8;4	Articulation disorder	English	Bilateral conductive hearing loss (wears hearing aids)
GR12	F	11;6	Motor speech disorder	English	ADHD and learning difficulties
GR13	F	11;2	Articulation disorder	English	-
GR14	M	10;10	Articulation disorder	English	-

^aAge reported in years; months. ^bReported by treating SLT (speech and language therapist), CAS was reported by the UK-based term “developmental verbal dyspraxia”, we use the international term “CAS” here.

Figure 2. Percent of children’s productions rated as always correct by the different types/pairs of transcribers (average of transcriber pair).

Table 3. Agreement between transcribers for rating (1) production as always correct or incorrect, (2) errors as imageable or non-imageable, and (3) classification of error types.

	Ultrasound (T1^a)	Ultrasound (T2)	Audio (T1)	Audio (T2)	Audio (T3)	Audio (T4)
(1) Always correct or incorrect productions
Ultrasound (T1)	-
Ultrasound (T2)	0.86^b Almost perfect^c	-
Audio (T1)	-^d	0.59 Moderate	-
Audio (T2)	0.62 Substantial	-^d	0.62 Substantial	-
Audio (T3)	0.53 Moderate	0.56 Moderate	0.61 Substantial	0.62 Substantial	-
Audio (T4)	0.42 Moderate	0.48 Moderate	0.49 Moderate	0.59 Moderate	0.56 Moderate	-
(2) Imageable or non-imageable error
Ultrasound (T1)	-
Ultrasound (T2)	0.86 Almost perfect	-
Audio (T1)	-^d	0.58 Moderate	-
Audio (T2)	0.57 Moderate	-^d	0.59 Moderate	-
Audio (T3)	0.51 Moderate	0.56 Moderate	0.59 Moderate	0.58 Moderate	-
Audio (T4)	0.43 Moderate	0.48 Moderate	0.47 Moderate	0.58 Moderate	0.53 Moderate	-
(3) Classification of error type
Ultrasound (T1)	-
Ultrasound (T2)	0.72 Substantial	-
Audio (T1)	-^d	0.51 Moderate	-
Audio (T2)	0.44 Moderate	-^d	0.48 Moderate	-
Audio (T3)	0.42 Moderate	0.40 Fair	0.52 Moderate	0.44 Moderate	-
Audio (T4)	0.32 Fair	0.35 Fair	0.38 Fair	0.44 Moderate	0.46 Moderate	-

^aT1 = Transcriber 1, T2 = Transcriber 2, T3 = Transcriber 3, T4 = Transcriber 4. ^bCohen’s kappa calculated with a 95% confidence interval. ^cDescriptors based on those listed in Viera and Garrett (2005). ^dInterrater reliability not calculated as these transcriptions were completed by the same transcriber. Darker colours represent higher agreement between raters.

Figure 3. Error types identified by transcriber group/pair (average values of raw count across dataset).

Gibbon, F. E. (1999). Undifferentiated lingual gestures in children with articulation/phonological disorders. Journal of Speech, Language, and Hearing Research, 42(2), 382–397. https://doi.org/10.1044/jslhr.4202.382

 PubMed Web of Science ®Google Scholar

Cleland, J., Scobbie, J. M., & Wrench, A. A. (2015). Using ultrasound visual biofeedback to treat persistent primary speech sound disorders. Clinical Linguistics & Phonetics, 29(8–10), 575–597. https://doi.org/10.3109/02699206.2015

 Google Scholar

Cleland, J., Scobbie, J. M., Heyde, C., Roxburgh, Z., & Wrench, A. A. (2017). Covert contrast and covert errors in persistent velar fronting. Clinical Linguistics & Phonetics, 31(1), 35–55. https://doi.org/10.1080/02699206.2016.1209788

 PubMed Web of Science ®Google Scholar

Lundeborg, I., & McAllister, A. (2007). Treatment with a combination of intra-oral sensory stimulation and electropalatography in a child with severe developmental dyspraxia. Logopedics, Phoniatrics, Vocology, 32(2), 71–79. https://doi.org/10.1080/14015430600852035

 Google Scholar

Hardcastle, W. J., Barry, R. M., & Clark, C. (1987). An instrumental phonetic study of lingual activity in articulation-disordered children. Journal of Speech, Language, and Hearing Research, 30(2), 171–184. https://doi.org/10.1044/jshr.3002.171

 Web of Science ®Google Scholar

Case, J., & Grigos, M. I. (2016). Articulatory control in childhood apraxia of speech in a novel word-learning task. Journal of Speech, Language, and Hearing Research, 59(6), 1253–1268. https://doi.org/10.1044/2016_JSLHR-S–14–0261

 Google Scholar

Dodd, B., & Lacano, T. (1989). Phonological disorders in children: Changes in phonological process use during treatment. International Journal of Language & Communication Disorders, 24(3), 333–352. https://doi.org/10.3109/13682828909019894

 Google Scholar

Cleland, J., Lloyd, S., Campbell, L., Crampin, L., Palo, J. P., Sugden, E., … Zharkova, N. (2020). The impact of real-time articulatory information on phonetic transcription: Ultrasound-aided transcription in cleft lip and palate speech. Folia Phoniatrica Et Logopaedica, 72(2), 120–130. https://doi.org/10.1159/000499753

 PubMed Web of Science ®Google Scholar

Viera, A. J., & Garrett, J. M. (2005). Understanding interobserver agreement: The Kappa statistic. Family Medicine, 37(5), 360–363.

 PubMed Web of Science ®Google Scholar