Lecturer's Précis - Ellis and Young (1988)

"Human Cognitive Neuropsychology"

Copyright Notice: This material was written and published in Wales by Derek J. Smith (Chartered Engineer). It forms part of a multifile e-learning resource, and subject only to acknowledging Derek J. Smith's rights under international copyright law to be identified as author may be freely downloaded and printed off in single complete copies solely for the purposes of private study and/or review. Commercial exploitation rights are reserved. The remote hyperlinks have been selected for the academic appropriacy of their contents; they were free of offensive and litigious content when selected, and will be periodically checked to have remained so. Copyright © 2002-2018, Derek J. Smith.


First published online 10:54 BST 3rd May 2002, Copyright Derek J. Smith (Chartered Engineer). This version [2.0 - copyright] 09:00 BST 3rd July 2018.

An earlier version of this material appeared in Smith (1997; Chapter 5). It is repeated here with minor amendments and supported with hyperlinks.


The Ellis and Young (1988) Transcoding Model

See firstly the supporting commentary for this material.

This is the first of the large scale psycholinguistic models to be widely published. It is derived from the earlier models via Ellis (1982), and is a good example of how to display highly modular cognitive processing in dataflow diagram (DFD) form. It shows the flow of information during both spoken and written language, makes some important theoretical judgements as to the cognitive modularity involved, and uses flowlines to interconnect the various resources.


The Ellis and Young (1988) Model: Numbers refer to the notes which accompanied the original. Input streams are shown at the top, and output streams at the bottom. Spoken language is shown to the left, and written language to the right. This gives four distinct processing routes, namely speech perception (top left), reading (top right), speech production (bottom left), and writing (bottom right). The central, or "semantic" system, is common to all four processing routes, but has been deliberately left "underspecified" to avoid swamping the surrounding detail. This is where the input routes terminate (with the creation of some sort of understanding of the material presented), and it is where the messages for production by the output routes actually originate (with the emergence of some sort of communicative intention). Following the lead set by Freud (1891), the semantic system is shown surrounded by four lexicons, one for each of the processing routes. These are where whole known word forms are stored. The two output routes are clear examples of motor hierarchies.

Bypass Routes: Some important routes allow information to flow around rather than through the central array of lexical-semantic modules. The main bypass routes are:

Route 11: This route allows the semantic system and the two auditory lexicons to be bypassed. This gives us the ability to copy unrecognised and essentially meaningless sounds, as for example when presented with material in an unknown language (compare Route 23 below). Route 15: This route allows the graphemes making up unfamiliar words to be "sounded out", and converted into their equivalent sound patterns. This then allows nonsense words, unfamiliar words, and foreign words to be spoken out loud either (a) mentally, by "talking to yourself" along Route 11 and then recognising what you are saying along Route 2, or (b) actually, by continuing downwards along the (unnumbered) speech production route. Route 18: This route allows sounds to be assigned letter equivalents. This then allows nonsense words to be written to dictation. If Routes 11 and 18 are combined, it allows the phonetic transcription of sounds other than those making up known words. Route 23: As for Route 11 above, but bypassing the two visual lexicons, and giving us the ability to copy unrecognised text, as when dealing with an unknown language.

Feedback Routes: Other important routes allow information to be fed back from a later process to an earlier one. These are of two sorts - single-ended arrows pointing up the page, and double-ended arrows. The main feedback routes are:

Route 10 (upper arrow): This route allows feedback from the speech production process to the word selection process. This allows speech output to be properly paced. New words or phrases are not released for articulation until the word or phrase currently being articulated is nearly and successfully finished with. Should articulation problems be encountered (a dry throat, say, or a mispronunciation), the whole process is immediately interrupted. Route 11 (upper arrow): This route allows internal feedback from the speech production process to the auditory analysis process. The authors suggest that this might provide "a mechanism for what we experience in everyday life as 'inner speech', where we appear to hear our own silent speech internally" (p226). Route 12: This route allows external feedback from the final act of speech production to the auditory analysis process. This is where we hear our own voice coming back to us through our ears. Route 22: As for Route 12, but for writing-reading rather than speaking-hearing. This is where we see our own writing coming back to us through our eyes.

Clinical Evidence: The model reflects a large body of clinical evidence, from Wernicke (1874) to Marshall and Newcombe (1973). Any one named process or information flow can be damaged more or less in isolation, and language behaviour will suffer in a very particular way as a result. Generally speaking, damage to the semantic "hub" of the model will result in what is known as a "deep" defect such as an agnosia, whilst damage to "outer ring" processes will result in a variety of dysphasias, anomias, dyspraxias and dysarthrias. Here are some specific defects:

  • Route 1: Damage here results in a pure word deafness: This is characterised by difficulty understanding or repeating heard language, but with intact comprehension, speech, reading, and writing.
  • Route 2: Damage here results in a word meaning deafness: This is characterised by difficulty understanding heard language, but with intact repetition and word production.
  • Route 4: Damage here results in a variety of defects, including dementia, deep dysphasia, and deep dyslexia, all characterised by mistakes and confusions on semantic grounds. Michel and Andreewsky (1983), for example, report a dysphasic patient who repeated "balloon" as "kite" or "red" as "yellow". Similarly, Coltheart (1980) reports dyslexic patients misreading "merry" as "happy" and "short" as "small".
  • Route 5: Damage here results in a peripheral dyslexia: This is characterised by difficulty attending to, grouping, or processing the superficial aspects of text.
  • Route 6: Damage here results in a "visual" dyslexia: This is characterised by difficulty identifying familiar written words, or telling apart visually similar words such as "calm" and "clam".
  • Route 7: Damage here results in a anomic aphasia: This is characterised by difficulty producing the name of an object whose function is perfectly well understood.

If this diagram fails to load automatically, it may be accessed separately at



Redrawn from Ellis and Young (1988:222). This version Copyright © 2002, Derek J. Smith.




Coltheart, M (1980). The semantic error: Types and theories. In Coltheart, M., Patterson, K., & Marshall, J.C. (Eds.), Deep Dyslexia. London: Routledge & Kegan Paul.

Ellis, A.W. (1982). Spelling and Writing (and Reading and Speaking). In Ellis, A.W. (Ed.), Normality and Pathology in Cognitive Functions. London: Academic Press.

Ellis, A.W. & Young, A.W. (1988). Human Cognitive Neuropsychology. Hove, UK: Erlbaum.

Michel, F. & Andreewsky, E. (1983). Deep dysphasia: An analogue of deep dyslexia in the auditory modality. Brain and Language, 18:212-223.

Smith, D.J. (1997). Human Information Processing. Cardiff: UWIC. [ISBN: 1900666081]