Jingle and Jangle Fallacies in Levels of Representation

I have had several opportunities over the past few years to object when the investigators who conducted the Sound Start trial associated the framework outlined in our book with the theoretical foundation for their work. We have not had an opportunity to fully explore my objection and twitter is a bad medium for a discussion requiring this much complexity and nuance and therefore I am going to provide a rationale in some detail in my blog. McLeod et al. justify their approach by reference to certain psycholinguistic models with particular allegiance to Stackhouse and Wells’ (1993) important work. In our book, Francoise and I also pay homage to their model and note the historical lineage although our framework is drawn directly from work by Munson, Edwards, and Beckman (2005). Therefore there is no argument about the use of the terms input, output, and phonological processes or representations–as we note, this basic tripartite division of speech processes is more or less universal. The difficulty is that McLeod and colleagues divide up assessment and treatment tasks according to these category labels (input, output and phonological processes) differently from us and then cite our framework. It is jarring because the error appears to reflect both jingle and jangle fallacies. For those who have not encountered these amusing terms before, a jingle fallacy refers to the assumption that two concepts with the same name are the same when they are actually different; a jangle fallacy is the assumption that one concept that may be referred to with different names are therefore different when there is in fact only one concept. We can all agree that McLeod’s team and my team (and Munson’s team and so on) can all use a tripartite framework (input-output-phonological) and we can all agree to disagree about what which tasks go into which division (psycholinguistics have been doing that for a long time now and will continue to do so). However, when I am cited I would prefer to not have confusion about what I mean when I talk about input vs output processes.

I will begin with the point of agreement–the tripartite division of psycholinguistic processes, citing McLeod et al. (2017) directly: “Stackhouse and Wells (1997) … proposed three core elements: Input processes (i.e., detecting and perceiving speech…), cognitive-linguistic processes (i.e., creating, storing, and accessing lexical representations of words …), and output processes (i.e., producing speech…).

In our text, Francoise and I borrow heavily from Munson, Edwards, and Beckman to describe three types of phonological knowledge: Perceptual knowledge encoded in the form of acoustic-phonetic representations for speech sounds, abstracted from stored acoustic memories of words; articulatory knowledge encoded in the form of motor plans for syllables; and phonological knowledge, encoded as underspecified phonological units at all levels of the phonological hierarchy, and acquired as an emergent property of the lexicon itself. A variety of processes are proposed for acquiring and using these types of knowledge when perceiving, understanding, and producing speech.

The difficulty comes when we begin to assign different assessment or treatment tasks to these levels of processing or representation. In our book we describe input processes and input approaches to treatment as those that target specifically children’s acoustic-phonetic representations. Strong acoustic-phonetic representations provide support for speech perception and implicit phonological awareness. Assessment and intervention tasks will involve the provision of varied speech inputs, focusing on words but with systematic variation in acoustic cues and involving implicit learning strategies. Tasks that tap these processes may involve only listening to speech input or they may involve listening and talking–it depends upon the design of the task and the way that the children’s responses are analyzed. For example, on of my favourite studies that reveals the importance of “input processes” was conducted by Munson, Baylis, Krause, and Yim (2006). In their study children first listened passively to nonwords. After a distractor task they repeated nonwords, some of which they had previously heard during the passive listening task. Children with typical speech showed a benefit of the previous exposure in their repetition accuracy whereas children with a speech sound disorder did not show this benefit. You can see that this task that is largely dependent upon spoken responses is a measure of input processing! Speech perception tasks fall into this category most clearly when they reveal something about the nature of the acoustic cues that the child is using to make decisions about which acoustic-phonetic objects form a particular word or phonetic category. Some phonological awareness tasks are also input oriented when the child indicates that, for example, “hat” and “bat” sound similar by matching pictures even if they do not have high level metacognitive knowledge about what the similarity is.

Phonological knowledge is a more abstract form of knowledge that emerges from the organization of the lexicon and from explicit teaching, especially phonics and reading education in schools. It includes metacognitive knowledge of sublexical and subsyllabic units. Assessment and intervention tasks in this domain often involve high level expressions of this knowledge such as verbally identifying the common sound in the coda of the words “hat” and “boat” or indicating that [b] is at the beginning of the sound “boat” or differentiating 3-syllable from 2-syllable words.

In some children there are discontinuities across these levels of knowledge even when the same unit is involved. For example, a child may be able to indicate that [bæθ] and [bæt] and [bæs] correspond to different pictures (i.e., have different meanings) but have an unclear sense of the acoustic cues that differentiate the phonetic categories that differentiate these words. Another child might have excellent acoustic-phonetic representations for these words and the phonetic categories that differentiate them but have immature metaphonological knowledge, being unaware that each word is composed of three phonemes and unable to tell you that they share the same head [bæ]. In our book, Francoise and I detail the kinds of tasks that can be used clinically to assess and remediate children’s knowledge at different levels of representation.

The disagreement we are having with McLeod et al is the classification of all the tasks in the Phoneme Factory computer intervention as being “input oriented” tasks. According to our framework, even though most tasks require the child to listen and then respond by selecting pictures or letters on a computer screen, these tasks all involve accessing phonological levels of representation and do not serve to strengthen the child’s acoustic-phonetic representations. Even the most basic level task involves associating sounds produced in isolation (e.g., [s], [d]) with a standard pictograph (e.g., [s] → “snake”). The authors mistakenly identify this task with the lowest level “input” process in Stackhouse and Well’s model, that is, speech discrimination, but it is not a discrimination task and the stimuli do not reveal anything about the children’s knowledge of the acoustic-phonetic cues that differentiate one category of speech sounds from another. All the tasks in the program are metaphonological tasks that that therefore tap phonological knowledge even though real words and word meanings are not always engaged.

At the recent NZSPA2019 Conference in Brisbane Jane McCormack divided up the phonological awareness assessment tasks that comprise the CTOPP into input and output task purely on the basis of whether a spoken response was required by the child. However, I would not agree that any of these phonological awareness tasks reveal the child’s acoustic-phonetic knowledge of speech sound categories and therefore there are no “input tasks” per se. All the tasks are tapping meta-phonological knowledge.

If this is still confusing, think of that child who says [s̪it], [s̪nek], [fes̪], and [buts̪], and who confidently identifies [mauθ] as the picture with teeth, but both [maus] and [maus̪] as the picture of the rodent. This same child is able to blend the sounds [m] – [au] – [θ] to recreate the word /mauθ/. If you ask her to say [maus] without [s] she answers [mau]. Here we have a child whose acoustic-phonetic and articulatory-phonetic knowledge of the /s/ phoneme is poor, explaining the consistent distortion in her speech; at the same time the child’s phonological knowledge of the /s/ – /θ/ contrast is good and her meta-phonological skills are good as well. Therefore, when treating this child, we would want to focus at the phonetic level. The Phoneme Factory intervention might be good for her future literacy skills but it would not be the best prescription for her speech articulation problem. We really want to have a clear understanding of the difference between these three levels of representation.

As a more general point it is really important when citing anyone to match up terms with concepts in a way that is consistent with the cited authors’ original intent. This is hard because the use of terms undergoes so much historical and theoretical change. The changes are good I think – Munson et al. help us to understand that many children with developmental phonological disorders have difficulties in the phonetic domains (acoustic-phonetic and articulatory-phonetic representations) whereas many children with language impairments have deficits in phonological knowledge in fact, a by product of smaller lexicons. Knowing how to assess and remediate children’s knowledge in these three domains will help us to target our interventions more effectively.

References

Baker, E., Croot, K., McLeod, S., & Paul, R. (2001). Psycholinguistic models of speech development and their application to clinical practice. Journal of Speech, Language, and Hearing Research, 44, 685-702.

McLeod, S., Baker, E., McCormack, J., Wren, Y., Roulstone, S., Crow, K., . . . Howland, C. (2017). Cluster-Randomized Controlled Trial Evaluating the Effectiveness of Computer-Assisted Intervention Delivered by Educators for Children With Speech Sound Disorders. Journal of Speech, Language & Hearing Research, 60(7), 1891-1910. doi:10.1044/2017_JSLHR-S-16-0385

Munson, B., Baylis, A., Krause, M., & Yim, D.-S. (2006). Representation and access in phonological impairment. Paper presented at the 10th Conference on Laboratory Phonology, Paris, France, June 30-July 2.

Munson, B., Edwards, J., & Beckman, M. E. (2005). Phonological knowledge in typical and atypical speech-sound development. Topics in Language Disorders, 25(3), 190-206.

Rvachew, S., & Brosseau-Lapre, F. (2018). Developmental Phonological Disorders: Foundations of Clinical Practice (Second ed.). San Diego, CA: Plural Publishing, Inc. https://www.pluralpublishing.com/publication_dpd2e.htm.

Stackhouse, J., & Wells, B. (1993). Psycholinguistic assessment of developmental speech disorders. European Journal of Disorders of Communication, 28, 331-348.

Previous Post
Leave a comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: