Speech Sound Disorders Peter Flipsen Jr., Ph.D., CCC-SLP University of Tennessee, Knoxville http://web.utk.edu/~pflipsen
Outline • 1. General Issues • Assessment, Target Selection, Intervention, Measuring Progress • 2. Articulation vs. Phonological Disorders • 3. Speech Discrimination • 4. Childhood Apraxia of Speech • 5. Oral-Motor Exercises • 6. Dialects - Appalachian English • 7. Approaches to Intervention • Articulation, phonological, discrimination, CAS
General Issues - Assessment • The goal of every assessment is to decide if there is a problem or not. • When we say there is a problem, we may be right or we may be wrong. • Client may have had a really bad day and performed poorly. • Client may have been “in the zone” and did better than they usually do.
General Issues - Assessment • Risk of errors in our decisions greatest for borderline cases, but always a concern. • Given our reliance on standardized tests, we need to remember to consider “standard error of measurement” or SEM. • Accounts for the fact that all test scores are really just “samples” of ability and scores may vary from day to day. • Gives us a sense of how much they might vary.
Standard Error of Measurement (SEM) • Allows us to see where the “true ability level” is. • If a child achieves a standard score (SS) of 80 and the SEM is 5, then: • There is a 68% probability that his actual score is really somewhere between 75 and 85. • We are 68% confident that his true ability is within 1 SEM of his score that day. • There is a 95% probability that his actual score is really somewhere between 70 and 90. • We are 95% confident that his true ability is within 2 SEMs of his score that day.
SEM • Not always available but is generally found in most newer test manuals. • E.g., Photo Articulation Test – 3
SEM • GFTA-2 gives us the 90% and 95% “confidence intervals” directly for every score. • Hodson’s new HAPP-3 does NOT provide SEM values or confidence intervals. • Note: The HAPP-3 allows you to calculate “Ability Scores” which look like standard scores, BUT the test manual recommends you use percentile ranks instead.
General Issues – Assessment • What about unusual errors? • We know that unusual errors such as “lateralization” of fricatives are very resistant to intervention especially if left too long. • Need to avoid letting them become too established. • We need to find a way to justify working on these errors much earlier. • Any ideas?
General Issues - diagnosis • When we decide what the nature of the problem is, we assign a “category”. • Differential diagnosis. • Ultimately each category should mean a different approach to treatment. • We need to know how distinct each category is from other related categories.
General Issues - treatment • Change happens – we see it every day. • Many possible reasons: • Our intervention resulted in the change. • Child “figured out” what they need to do on their own (i.e., normal development). • Adult relearned a skill as physiological recovery progressed (i.e., spontaneous recovery). • Some outside influence led to the change (e.g., parent or spouse working with them).
General Issues - treatment • How do we know what caused the change? • If we do something and change follows, did we cause the change? • Even if the change is almost immediate, we still CANNOT be sure! • Something else may have been responsible for the change. • Still a long way to go here. • See June 13/06 issue of ASHA Leader.
General Issues - treatment • Recommended reading: • Reilly, S., Douglas, J., & Oates, J. (2004). Evidence Based Practice in Speech Pathology. Philadelphia, PA: Whurr Publishers.
General Issues - Measuring Progress • When we “monitor progress” we are really re-assessing skills to see if the client has learned what we’ve been teaching. • Re-administer a standardized test? • May be necessary to make decisions about whether a client is still “eligible for services”. • Need to consider SEM. • Doing this doesn’t really tell us if progress has happened.
Standardized Tests and “Measuring Progress” – Why Not? • 1. These tests are intended for a wide range of ages. • Designed for efficient administration and thus don’t sample very many behaviors at any one particular age. • Only sample each ability level superficially. • For speech sounds, they don’t test enough examples of those sounds. • Child may have over-learned those particular words. • Child may have a “fossilized form” for those particular words.
Standardized Tests and “Measuring Progress” – Why Not? • 2. Regression to the Mean. • Scores at the very low end or the very high end are not very common (relative to the entire population). • By sheer probability, when you retest, low scores are more likely to go up and high scores are more likely to go down. • Remember that statistically speaking, really tall parents tend to have shorter children than themselves and really short parents tend to have taller children than themselves.
More on Regression to the Mean • Every test score is a “sample” of ability and includes measurement error. • That’s why we consider SEM in assessments. • With a very low score, it means that many “sources” of measurement error were working against the child that day. • When we retest, it is much less likely that those “sources” will again be working against the child. • Scores are likely to improve just by chance.
Alternatives for measuring progress • 1. Conversational speech samples – for speech sound disorders the ultimate goal is performance in spontaneous speech. • A. For younger children (many errors) - have an unfamiliar listener transcribe (using regular spelling) and calculate % understood. • Track % understood over time. • Expectations for % understood = • Age in years / 4.
Alternatives for measuring progress • May also do phonetic transcription. • Calculate Percentage of Consonants Correct (PCC) and Percentage of Vowels Correct (PVC). • Compare to reference data from Austin & Shriberg (1997). • See handout
Alternatives to measuring progress • Use Means (and standard deviations) to calculate z-scores. • Z-score = how many standard deviations from the mean a raw score is. • z-score = (score –mean) / std. dev.
Alternatives for measuring progress • B. For older children (fewer errors) • Have an unfamiliar clinician transcribe phonetically and calculate % correct. • Probably only need to focus on the particular target sounds. • Clinicians can act as transcribers for each other.
Alternatives for measuring progress • 2. Systematic Probes – for each target sound, set aside some (e.g., 10) words containing the target sound that you don’t use for practice in therapy. • Bring these out every few weeks or so and ask the child to produce them. • Track % correct over time.
Articulation vs. Phonological Disorders • Now ASHA’s preferred term = Speech Sound Disorders. • Includes both “articulation disorders” and “phonological disorders”. • BUT is it reasonable to lump these two categories together? • Are they just two different names for the same thing? • Even if they are different, do we treat them differently?
Articulation Disorders • Group exercise. Answer the following: • What do we mean by an articulation disorder? • What specific behaviors do we observe?
Phonological Disorders • Group exercise. Answer the following: • What do we mean by a phonological disorder? • What specific behaviors do we observe?
Speech Sound Disorders • Articulation vs. Phonological Disorders • Are they the same thing? • If not, should we be doing something different for each of them?
Articulation Disorders • Problems with the physical aspects of producing speech sounds. • Not stimulable (or very poorly so). • Don’t ever produce the sound correctly. • Don’t produce the sound accidentally in place of some other sound. • Sometimes called phonetic disorders.
Phonological Disorders • Phonology = sub-domain of language. • The sound system. • How the phonemes and allophones are organized within a language. • Phonological disorder = a type of language disorder.
Phonological Disorders • Not a “production” problem. Child appears capable of producing the target but isn’t using it correctly. • Errors are stimulable, especially to the word level or beyond, • Target may also be produced accidentally in place of something else. • Sometimes called phonemic disorders.
Natural Phonological Processes • What does it really mean when we say that a child exhibited final consonant deletion? • Or velar fronting? • Or cluster reduction? • By themselves, do these labels really tell us what’s going on inside a child’s head?
Which Process? • If a child leaves off the /s/ in words like “hats” and “ducks”, what process is operating? • Final consonant deletion? • Stridency deletion? • Consonant sequence reduction (cluster reduction)? • Or is this just a failure to learn the plural morpheme?
Natural Processes vs. Linguistic Processes • The natural process labels that SLPs use are not the same as the phonological processes that linguists talk about. • Serious potential for confusion.
Processes vs. Processing • With the emergence of discussions of “phonological awareness”, we‘ve begun to look at psycholinguistic models of how the brain manages information (processing). • Are we talking about the same thing?
Processes vs. Processing • Just because we see errors that we can label as fronting, stopping, etc., this says ABSOLUTELY NOTHING about: • phonological awareness skills • short term memory skills, or • how the brain “processes” linguistic information. • Whether the problem is phonological or articulatory.
Processes vs. Patterns • There is no doubt that for many children their errors seem to follow patterns. • Capturing a child’s “pattern of errors” using labels such as stopping, fronting, etc. can be very useful clinically. • But even Barbara Hodson has suggested we call them “patterns” rather than “processes”.
What about speech discrimination? • If we assume normal hearing acuity (i.e., no hearing loss): • Is it possible to have difficulty producing speech sounds because of difficulty with speech discrimination? • Even if such a problem exists, is it possible to test it?
What about speech discrimination? • If the problem were one of “general inability to discriminate speech”, then no speech would be possible. • We do occasionally see children who have problems with discriminating speech specific to sounds they are not producing correctly. • Not at all clear how common this is. • Probably relatively uncommon, but we can’t ignore the possibility.
Speech Discrimination • Two possible problems: • 1. Problems discriminating sounds as produced by others (external discrimination). • 2. Problems discriminating sounds when produced by self (internal discrimination). • One example of this is the “Fis” phenomenon.
Can we test speech discrimination? • We can test external discrimination easily but ultimately we cannot ever really test internal discrimination. • We all hear our own speech differently than others do because of bone conduction. • Can’t get inside someone else’s head.
Testing Speech Discrimination • Common approach = minimal pairs test. • Present two words side by side (one contains the target, one contains the error). • Only requires comparison within working memory. • Doesn’t require the child to compare what they hear against their own internal representation. • Often present only one example – could guess.
Testing Speech Discrimination • Need a way to allow for comparison against internal representation and to prevent guessing. • Locke’s (1980) “Speech Production-Perception Task” (SP-PT) does both of these things. • Still based on production by someone else but probably as close as we’ll get.
SP-PT • Create a unique test for each of the child’s errors. • Compare child’s usual error to the target. • Include a similar sound that child can discriminate (ensures task is understood). • Present multiple examples to account for possible guessing.
SP-PT • Key = examiner presents one example at a time of a possible version of the target. • Child must compare what they hear with their internal representation and then decide: • Was it correct or not correct? • Record child’s responses.
Target / / Error / / Control / / Stimulus - Class Response 1. / / - Control yes - NO 2. / / - Error yes - NO 3. / / - Target YES - no 4. / / - Target YES - no 5. / / - Error yes - NO 6. / / - Control yes - NO 7. / / - Control yes - NO 8. / / - Target YES - no 9. / / - Error yes - NO 10. / / - Target YES - no 11. / / - Error yes - NO 12. / / - Control yes - NO 13. / / - Error yes - NO 14. / / - Target YES - no 15. / / - Control yes - NO 16. / / - Error yes - NO 17. / / - Target YES - no 18. / / - Control yes - NO Correct response shown in uppercase letters. Misperception = 3+ mistakes on Error.Mistakes: Error ____ Control ____ Target____
Target / ‘ / Error / f / Control / s / Stimulus - Class Response 1. / s / - Control yes - NO 2. / f / - Error yes - NO 3. / ‘ / - Target YES - no 4. / ‘ / - Target YES - no 5. / f / - Error yes - NO 6. / s / - Control yes - NO 7. / s / - Control yes - NO 8. / ‘ / - Target YES - no 9. / f / - Error yes - NO 10. / ‘ / - Target YES - no 11. / f / - Error yes - NO 12. / s / - Control yes - NO 13. / f / - Error yes - NO 14. / ‘ / - Target YES - no 15. / s / - Control yes - NO 16. / f / - Error yes - NO 17. / ‘ / - Target YES - no 18. / s / - Control yes - NO Correct response shown in uppercase letters. Misperception = 3+ mistakes on Error.Mistakes: Error ____ Control ____ Target____
Testing Speech Discrimination • Another option is the SAILS software program. • http://www.avaaz.com • Computer program that is intended to teach discrimination. • Includes an assessment tool.
All or None? • For any given child, will all of their errors fall neatly into “articulation”, “phonological” or “perceptual” categories? • Maybe but not necessarily. • Need to evaluate each error sound. • Treat each sound based on the type of error that it is.
Childhood Apraxia of Speech (CAS) • Now ASHA’s preferred term for “Developmental Apraxia of Speech”. • Group exercise. Answer the following: • 1. What are the core characteristics of CAS? [i.e., what behaviors set it apart from other speech sound problems?]
Childhood Apraxia of Speech (CAS) • See handout “ASHA’s draft position statement”. • Based a thorough review of the available evidence. • Still being discussed and fine tuned (i.e., not yet the final word but close).
CAS • “… (CAS) exists as a distinct diagnostic subtype of childhood (pediatric) speech sound disorder that warrants research and clinical services.” • Note: even with this, there may still remain some who claim it doesn’t exist.
CAS • “… (CAS) is a subtype of severe childhood speech sound disorder due to unidentified neurological differences likely of genetic origin. The core deficits arise at linguistic or early speech motor processing levels. Symptomatology, which changes with age, may include age-inappropriate vowel/diphthong errors, unusual and variable errors in repeated attempts at words, increased number and severity of errors with increasing word and utterance length, and prosodic disturbances. CAS places a child at risk for persisting problems in speech, language, and literacy.” • ASHA Ad Hoc Committee on Childhood Apraxia of Speech, 2006