Towards a Data-Based Definition of Cluttering
 |
About the presenter: Dr Robin Lickley is a Reader in Speech and Hearing Sciences at Queen Margaret University, Edinburgh, Scotland, UK. He completed his MA and his PhD at Edinburgh University. He spent a postdoctoral year at the University of Utrecht, Netherlands, before continuing his research at Edinburgh. His PhD and postdoctoral work focused on production and perception of typical disfluent speech in dialogue, before he began work on stuttering and cluttering. He is currently developing a multimedia-based corpus of stuttering and cluttering. |
Towards a data-based definition of cluttering
by Robin Lickley
from SCOTLAND
Let me begin by admitting that I don't know what I am talking about. What is cluttering? It's an evocative label and it appeals intuitively - a lay person can hazard a guess at what's involved. There are numerous books and papers on the topic, and there is even an International Cluttering Association. So somebody must know what I am talking about. The problem of a definition has been discussed plenty of times over the past decades, perhaps most recently by Scaler Scott and St Louis (2009), who bemoan the fact that there is an effective lack of progress in our understanding of cluttering since the special issue of the Journal of Fluency Disorders in 1996. In that volume, Bakker (1996) identified as a most fundamental need the establishment of "the validity of cluttering as an unique and independent diagnostic classification" (page 359). Fourteen years on, we are still there. There are so many diverse definitions or proposed characteristics of cluttering in the literature that more often than not it seems as if cluttering is what we do when we try to describe "cluttering". I begin this article with a selection of examples of how we have cluttered the description of cluttering at various different levels. I then propose a way forward - a new attempt to de-clutter the discussion of cluttering.
There is clutter in the use of generic terms to describe the disorder. If we assume that cluttering is in fact a definable disorder, then what is it a disorder of? Is it a speech disorder, a language disorder, or both (e.g., Weiss, 1964; Ward, 2006)? If so, what aspect(s) of speech or of language is(are) disordered? Is it a disorder of fluency (e.g., Ward, 2006; St Louis, Myers, Bakker & Raphael, 2007)? If so, what is "fluency", how can it be disordered and how is it disordered in the case of cluttering? Presumably, that has something to do with speech or language or both. Or is it a complex of learning disabilities (Op't Hof & Uys, 1974; Tiger, Irvine & Reis, 1980), which just happens to result in speech (or language) that can be considered to be cluttered?
There is clutter in the proliferation of symptoms and signs that are proposed as diagnostic criteria. A constellation of symptoms from different domains are proposed - speech motor control, language production, pragmatics, a range of other motor and cognitive domains...
Is cluttering a specific disorder or a syndrome representing a cluster of symptoms and signs? If it is a specific disorder is it, for example
- A disorder of the speaker's ability to regulate their rate of speech articulation, resulting in speech that is articulated at a rate which makes it hard to understand? Some researchers and clinicians would suggest that this is the principal characteristic.
- Or, as others imply, is it a disorder (perhaps of working memory) which leads to an excess of errors in speech and a resulting disfluent jumble as the speaker self corrects?
Both of the above have been suggested in the literature (at least as the major symptoms), and both appear to be possible disorders at some level of speech production, but there seems to be no logical reason why they should necessarily both share the same label: "Cluttering".
If cluttering is a syndrome and consists of a number of clinically observable signs and reportable symptoms, then this raises more questions: How many of these signs and symptoms can logically co-occur in a way that can be said to represent a single clearly-defined syndrome? How many of them have to be present in order for a diagnosis to be given with confidence - which ones are crucial to diagnosis and which are secondary?
Moving on to the various signs and symptoms that have been proposed, it seems that there are those who prefer to be relatively precise about their definition and those who take a more liberal approach. St Louis et al. (2007) represents the former, placing prominence on speech rate being abnormally rapid or irregular or both (before discussing other possible symptoms). Daly's Predictive Cluttering Inventory (2006) represents the latter, since it summarises many of the proposed indicators of cluttering rather than pinning the disorder to a particular symptom or sign. Here, 33 items are listed, the presence of any of which can add to the score that determines the presence of cluttering. Within the inventory, there is no difference in weighting between the items, as far as diagnostic value is concerned, so, for example, item 3 (Compulsive talker; verbose; tangential; word-finding problems) is equal to item 14 (Rapid rate (tachylalia)) and equal to item 21 (Language is disorganized; confused wording; word-finding problems). While not providing a clear and direct classification, Daly's approach does perhaps reflect the current state of knowledge (or perhaps the state of confusion) in the clinical world. There is a plethora of terms in use to describe symptoms of cluttering. Sometimes this reflects the fact that there is disagreement over what the key symptoms are (or, in fact, what cluttering is). Other times it happens when there is a lack of precision in the use of terms.
There is thus further clutter in the way that proposed symptoms are presented. For example, in addition to rapid speech (tachylalia), authors often refer to the phonological process of elision (Daly's item 13 "telescopes or condenses words"), to coarticulation and lack of pausing at word and phrase boundaries, all as separate features of cluttered speech. In fact, these are all natural processes that occur normally in continuous speech and they will naturally increase as articulation rate rises (as discussed by Dalton and Hardcastle, 1989). Imprecise articulation, insofar as it is realised as incomplete articulatory gestures, and a low range of pitch, may also occur as a consequence of rapid rate: it takes time for the tongue to reach ideal articulatory targets and for the vocal folds to reach extremes of pitch, and rapid speech may lead to a failure to reach such targets. Counting features such as these as discrete symptoms, independent of rapid speech, adds redundancy to the plethora of descriptive features or proposed symptoms, if they are already predictable from the very presence of rapid speech. It might also seem reasonable to assume that faster speech may lead to more errors and more disfluencies. In fact, various sources of evidence suggest that the effect on disfluency of faster speech may be specific to repetition disfluencies. A recent (as yet unpublished) finding from our lab is that in the HCRC Map Task Corpus (Anderson et al., 1991), a corpus of typical speech (64 speakers, 150,000 words), the only type of disfluency that was significantly more frequent with faster speakers was repetition. More analysis is under way, but the suggestion that faster non-pathological speech results in higher repetition rates may be relevant to the presence of stutter-like repetition disfluencies in pathologically rapid speech. That other disfluency types (e.g., error repairs) do not increase in faster speech was also shown experimentally by Oomen and Postma (2001). Van Zaalen-op 't Hof (2009) finds that children who clutter (defined as having speech that was not stuttered but "either too fast or irregular, together with at least one other symptom of St Louis et al. (2007) working definition of cluttering" (page 79)) produced more repetition disfluencies than children with learning disabilities and typically developing controls. To summarise, one way to de-clutter the array of features of speech proposed as symptoms of cluttering would be to recognise that several of them are a natural result of rapid speech, rather than a discrete set of items to be checked off on a list.
And finally, there is clutter within measurement issues. For example, there is a lack of standard practice in how we measure speech/articulation rate and in how we assess disfluencies. How fast is fast speech? There is no standard measure that is universally adhered to. Words per minute? Syllables per minute? Syllables per second? When you measure rate, what, if anything, do you omit from the timing and the word/syllable count? If rapid speech is a primary marker of cluttering, then it is absolutely crucial that there be a standard measure of rate. It is common practice to measure rate "online", counting words or syllables while timing the speech on a stopwatch. This is simply not acceptable as a proper and reliable measure of rate, when there are variable amounts of pause and disfluency within speech. The distinction must be made between speech rate and articulation rate, speech rate being a count of word/syllables over time including pauses, disfluencies, etc., articulation rate representing the speed at which words or syllables are articulated, when all pausing and other "noise" is removed from the calculation of units and time. If cluttering involves speech that is articulated at such a fast rate that words become unintelligible, then articulation rate of those words is what should be measured. Successful treatment will presumably be evidenced by a slowing of the rate of articulation of previously incomprehensible words, not just by a slowing of overall speech rate. A reduction in overall speech rate can be achieved by increasing the frequency and duration of pausing, or even by increasing the level of disfluency, while not altering the articulation rate of words: It is therefore possible to reduce speech rate while maintaining high articulation rate and low comprehensibility and (possibly) making speech sound more unnatural and no clearer. It is entirely understandable that busy clinicians would prefer the simpler and quicker method of counting syllables or words over time, while including all the noise and pauses, but this method does not give a reliable measure of articulation rate. Similarly, regarding disfluency, it is very difficult to establish reliable methods for classifying types of disfluency and for differentiating between stuttered, cluttered and typical disfluency. But at the very least, standard methods for assessing disfluency rate need to be accepted by researchers, and they are not. For both speech rate and disfluency rate, if we can establish standard methods for measurement, then we need reliable measures of what is abnormally fast rate and what is abnormal variation in rate for a given language and of what is an abnormally high rate of typical disfluency. There do exist large corpora of natural speech in many languages and such figures are available for some of them, though classification of disfluency types and method of reporting rates differs between corpora, so the data need to be read with caution. Establishing best practice in measurement and reliable norms for common measures are both crucial requirements for differential diagnosis and outcome measurement.
Unless we can de-clutter the discussion of cluttering, there is little hope for major progress in research in the area, for improvements in clinical diagnosis, assessment and intervention. This situation is bad for research, bad for clinicians and unacceptable for clients.
So what should we do about it?
I agree with Bakker (1996), that there is (still) a basic need for clear diagnostic criteria, and with Ward (2006, page 152), who states: The search for a data-based definition must be an urgent priority. To formulate a data-based definition, a necessary first step in establishing diagnostic criteria, we need sufficient data.
Since the primary symptoms are related to speech (and, possibly, to language), the data should consist of high quality digital recordings, preferably audio and video, of the speech of people who are thought, by clinicians or researchers, to clutter. In addition, case notes should be made available, covering a range of areas considered to be important in diagnosis.
For at least two reasons, we need recordings to be provided by a range of clinics or labs. One problem with collecting data in this area is its sparsity: In addition to the fact that availability of cases is hampered by a lack of awareness, not only by clinicians, but more importantly by the clients themselves, cluttering appears to be considerably less prevalent than stuttering, so no one clinic is likely to be able to gather sufficient data within a reasonable time frame. A second reason why it is desirable to collect data from several disparate sites is that the resulting data set is more likely to represent the broad range of definitions that exists.
Since the primary symptoms are related to speech, we need, at the very least, some basic objective analysis of rate of articulation and rate and type of error and disfluency (including error repair, hesitation and typical and stutter-like disfluency).
The recordings and derived data should be made available to a number of researchers/clinicians with experience of cluttering, stuttering and typical disfluency, who can participate in structured discussion of the cases.
An easy way to achieve this would be to host the data on a secure computer server, allowing approved users access to the video content and to a discussion forum via a web site. I already have such a server available and can set up the web site relatively easily. The recordings could not be publicly available, for obvious reasons (for example, participants whom I have already recorded have agreed to have their recordings made available for training and research purposes, but not for fully open access on the internet). But we would of course aim to publish results of analyses and outcomes of the discussions at the earliest possible time.
As I said at the top, "cluttering" is an evocative and appealing label. It concerns me that many people have picked up the label and are continually trying to pin something (i.e., symptoms) to it, rather than clearly defining a clinical problem and then finding a label to pin to that. It may be that once we have compared cases and notes we will end up deciding that we're actually talking about two or three different disorders. That, at least, will be progress.
References
Anderson, A., Bader, M., Bard, E., Boyle, E., Doherty, G. M., Garrod, S., Isard, S., Kowtko, J., McAllister, J., Miller, J., Sotillo, C., Thompson, H. S. & Weinert, R. (1991) The HCRC Map Task Corpus. Language and Speech, 34(4) 351-360.
Bakker, K., (1996). Cluttering: Current Scientific Status and Emerging Research and Clinical Needs. Journal of Fluency Disorders 21 3-4, 359-365
Dalton, P. & Hardcastle, W. (1989). Disorders of Fluency and their Effects on Communication. London: Elsevier.
Daly, D. A. (2006). Predictive Cluttering Inventory. Ann Arbor, MI
Oomen, C. & Postma, A. (2001). Effects of time pressure on mechanisms of speech production and self monitoring. Journal of Psycholinguistic Research, 30 (2): 163-184.
Op't Hof, J. & Uys, I.C. (1974). A Clinical Delineation of Tachyphemia (Cluttering): A Case of Dominant Inheritance. South African Medical Journal, 48: 1624-1628.
Scaler Scott, K. & St Louis, K. O. (2006). A Perspective on Improving Evidence and Practice in Cluttering. In Perspectives on Fluency and Fluency Disorders, 19: 46-51. American Speech-Language Hearing Association.
St. Louis, K., Myers, F., Bakker, K., & Raphael, L. (2007). Understanding and treating cluttering. In E. G. Conture & R. F. Curlee (Eds.), Stuttering and related disorders of fluency (3rd ed., p. 297-325). New York: Thieme.
Tiger, R.J., Irvine, T. L. & Reis, R. P. (1980). Language, Speech, and Hearing Services in Schools Vol. 11 3-14 January 1980
Van Zaalen-op 't Hof, Y. (2009). Cluttering Identified: Differential diagnostics between cluttering, stuttering and speech impairment related to learning disability. PhD Dissertation, Universiteit Utrecht.
Ward, D. (2006). Stuttering and Cluttering: Frameworks for Understanding and Treatment. Hove and New York: Psychology Press.
Weiss, D. (1964). Cluttering. Englewood Cliffs: Prentice-Hall.
SUBMITTED: April 6, 2010
