The intelligibility of children who have severe & persisting speech difficulties : Why phonological process analysis is not enough Jane Speake Sara Howard & Maggie Vance Department of Human Communication Sciences University of Sheffield RCSLT Manchester 2012
Aims To consider how we assess children’s speech and how we analyse the data To explore similarities and differences in single word (SW) and connected speech (CS) production in the speech of 3 children with severe & persisting speech difficulties (PSD) To reflect on the implications of our findings, particularly in relation to intelligibility RCSLT Manchester 2012
Phonological assessment of children’s speech difficulties in the clinical context The most common approach is to collect data from a SW naming task, transcribing it more or less narrowly (or sometimes not at all), and to carry out a phonological process analysis (Skahan, Watson & Lof, 2007). RCSLT Manchester 2012
Phonological intervention for children’s speech difficulties in the clinical context The process analysis results in a set of descriptions which support clinicians in planning and delivering intervention – e.g. Child is stopping, fronting and reducing clusters; targets will be x, y & z; activities a, b and c will enable SLTs to effect positive change For most children this approach works well RCSLT Manchester 2012
However… This approach might imply that SW are the building blocks for utterance level productions and that ‘fixing the problem’ at that level, will simply ‘fix the problem’ (which of course for many children, over time, it does) But does it ‘fix the problem’ for all children and if not, why not? RCSLT Manchester 2012
“In learning to talk, children must gain knowledge of the phonological words and phrases of their native language and must learn the articulatory and phonatory movements required to produce these words and phrases in an adult-like manner” Stoel-Gammon & Sosa (2007) RCSLT Manchester 2012
Single word production versus connected speech production “phoneticians who study casual speech phenomena have consistently noted reductions that involve extreme changes from citation forms…” Johnson (2004: 7 ) “massive reduction in conversational English” Johnson (2004: 1) RCSLT Manchester 2012
Single word production versus connected speech production “Massive reduction” is not random: words typically retain key individual features, essential for listener recognition e.g. all variations of “until” retain /t/ Some segments (e.g. /t/, /ð/, and /ə/) are ‘incredibly vulnerable’, whereas /f/, /m/, /ʃ/, /ʧ/ and /ʤ/ are ‘practically invulnerable’ (Shockey, 2003:15) RCSLT Manchester 2012
Implications Speech development includes being able to handle the phonetics, phonology, and prosody of multi-word utterances For some children this may pose very different challenges from SW production This may be one underlying factor in the failure of children with PSD to respond to intervention based on single words RCSLT Manchester 2012
Persisting speech difficulties (PSDs) ‘Residual phonological errors’ Shriberg, Gruber & Kwiatkowski (1994) ‘Intractable speech disorders’ Wood & Scobbie 2003 ‘Persisting speech difficulties’ Pascoe, Stackhouse & Wells (2006) RCSLT Manchester 2012
Our participants 3 children described as having PSD 6 years+ Have received SLT intervention for their speech production for 3+ years Have significant intelligibility issues (typically- developing children are reported to be intelligible by 4;0: Coplan & Gleason, 1988) RCSLT Manchester 2012
Percent Consonants, Vowels, & Phonemes Correct PCCPVCPPC “Harry” (CA: 7;5) “Lily” (CA: 7;2) “Hamish” (CA:6;7) RCSLT Manchester 2012
Single words (SW) versus connected speech (CS) itemSWCS Harry‘spider’ Lily‘sheep’ Hamish‘aeroplane’ RCSLT Manchester 2012
Single word phonological process analysis Examples of common processes: StoppingSOCK / sɒk / [ t⁼ɒʔ ] Hamish DeaffricationCHIPS / ʧɪps / [ t⁼ɪpsː ] Harry Cluster Reduction FROG / fɹɒɡ / [ fɒkʼ ] Harry [ fɒʔ ] Hamish [ fːɒɡ̊ ] Lily RCSLT Manchester 2012
Connected speech Processes identified in SW also occur in CS but more frequently (particularly glottal realisations) But this does not fully explain the speech patterns observed or the lack of intelligibility experienced by listeners Further analysis of the connected speech data reveals longer domain patterns not identified in the SW data RCSLT Manchester 2012
Multiword utterances “Hyperelision” “Hyperarticulation” Unusual juncture at word boundaries Unusual prosodic (rhythmic) patterns Utterance harmony patterns Howard, Wells & Local, 2008 RCSLT Manchester 2012
“Hyperelision” vs “Hyperarticulation” “Hyperelision” Harry -and shall I tell you what “Hyperarticulation” Lily-near my Dad’s house RCSLT Manchester 2012
Hyperelision vs Hyperarticulation Hyperelision Harry -and shall I tell you what Hyperarticulation Lily-near my Dad’s house RCSLT Manchester 2012
“Hyperelision” vs “Hyperarticulation” “Hyperelision” Harry -and shall I tell you what Hyperarticulation” Lily-near my Dad’s house RCSLT Manchester 2012
Hyperelision vs Hyperarticulation Hyperelision and hyperarticulation in the same utterance (Howard, 2007) Hamish -under my bed got a mattress Harry -and landed on the boat’s top and decided to peck it RCSLT Manchester 2012
Utterance harmonies Hamish you can read my book [ˈʔɒ̃n ə ˈmĩ mãɪ ˈbʌʔ] Lily but I did get two and I kept them [ bəʔ aɪ ˈdɪd ˈdɛʔ ˈdu ən aɪ ˈdɛʔː dɛn ] RCSLT Manchester 2012
Intelligibility: Listener experience Single words % understood Conversational speech % understood Harry PCC 62.11; PVC Mean: Range: Mean Range: Lily PCC 44.9; PVC Mean: Range: Mean: Range: Hamish PCC 32.00; PVC Mean: Range: 0-40 Mean: Range: RCSLT Manchester 2012
Implications Children's speech difficulties are typically described in terms of phonological processes in SW The complex demands of multiword utterances increase the use and variability of phonological processes In addition, phonetic and phonological behaviours associated with multiword utterances affect intelligibility with interactions between typical CS processes and segmental difficulties RCSLT Manchester 2012
Implications The relationship between measures such as PCC and intelligibility, and the intelligibility of different types of utterance is neither predictable nor straightforward Listeners’ ability to understand disordered speech is highly variable Careful, detailed analysis of multiword utterances is essential in understanding and explaining the speech of individual children with PSD RCSLT Manchester 2012
Impact on intelligibility: a last word from Harry J: Do people ever say they can’t understand what you are saying? Harry: “Lots of times every day, a thousand times a day” RCSLT Manchester 2012
Acknowledgments Thanks to ‘Harry’, ‘Lily’ and ‘Hamish’ and their families This research constitutes part of the first author’s PhD study This research was supported by an ESRC Research Fellowship (Grant No RES ) “Connected speech and word juncture in typical and atypical speech development” awarded to the second author RCSLT Manchester 2012