Lecturer's Précis - Crosson (1985)
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 16:05 BST 28th August 2002, Copyright Derek J. Smith (Chartered Engineer). This version [2.0 - copyright] 09:00 BST 3rd July 2018.
An earlier version of this material was contained in Smith (1997). It is repeated here with additional detail and supported with hyperlinks.
Crosson's (1985) Subcortical Language Processing Model
A detailed analysis of the role of subcortical structures in language is provided by Bruce Crosson of the University of Florida (Crosson, 1985). To begin with, he implicates the thalamus in a phenomenon known as thalamic aphasia:
"The issue of thalamic aphasia has been a controversial one with some authors questioning the existence of such a syndrome. [Nevertheless] an increasing number of scientists are beginning to accept and speculate about a role for the dominant thalamus in language. [The features of thalamic aphasia] are (a) relatively fluent language with substitutions of one word for another and with spoken language sometimes deteriorating into incomprehensible jargon, (b) comprehension that is less impaired than the type of language output would normally indicate, and (c) relatively preserved repetition [Refs]. Perseveration and a lack of spontaneous speech are also common." (Crosson, 1985:264-265.)
Crosson then puts forward an explanatory model, which suggests certain functions of the cerebral cortex, the thalamus, and the basal ganglia, and which hypothesises as to how these structures might intercommunicate. This is therefore a model of cortical-subcortical interactive processing, and Crosson presents it as an alternative to the "classical" models of aphasia which emphasise cortical processing and subcortical interconnection. The prime examples of these classic models are Wernicke (1874) and Lichtheim (1885), and Crosson's chief criticism of them is that they fail to explain the paradoxical but demonstrable role played by language comprehension processes during the act of language output. These functions are kept separate in the classical model (in Wernicke's and Broca's areas respectively), but in fact interact closely and continually. The main points of Crosson's argument are then as follows:
* To understand the overall process of language processing, it is important to understand how information flows between the various subsystems involved, including (and often critically) subcortical ones.
* Language is monitored for accuracy twice, once internally prior to its utterance, and then again during its utterance.
* Initial language "formulation" is a function of the inferior frontal gyrus, the frontal, parietal, and temporal opercula, and the insula. Between them, these centres provide all the conceptual, word-finding, and syntactic processes required during sentence building.
* The thalamus allows this encoding to be monitored prior to its actual execution. Crosson calls this first type of monitoring "preverbal semantic monitoring". The circuit goes through the ventral anterior nucleus and the pulvinar, to the temporoparietal cortex, which is where the actual monitoring takes place. The breakdown of this monitoring is what allows the aphasic phenomenon known as jargoning .
* The ventral anterior thalamus is one of several structures known to be involved in cortical arousal. It receives afferents from the brainstem reticular formation, and it can excite, in turn, much of the frontal cortex. Artificial electrical stimulation of the ventral anterior thalamus can elicit jargon (over-excitation), whilst damage to the fibres ascending from it as they pass upwards through the internal capsule can elicit dysfluencies (under-excitation).
* But the ventral anterior thalamus is itself controlled by the basal ganglia, specifically, the globus pallidus. Lesioning the globus pallidus disinhibits the thalamus, leading to overactivation of the cortex, resulting in the inclusion of "extraneous material" in the final utterance. Stimulation of the globus pallidus, on the other hand, raises the level of thalamic inhibition and interrupts ongoing language.
* The release of the final response is actioned by the intervention of the head of the caudate nucleus. This "exerts an inhibitory influence over the inhibitory mechanisms of the globus pallidus" (p278). This releases the planned output for motor programming and overt delivery.
* The second type of monitoring - "phonological monitoring" - is then possible, and is enabled by information flowing along the arcuate fasciculus. The temporoparietal cortex monitors what is currently being motor programmed, and interrupts that process if it detects an error.
Here is the same argument, expressed diagrammatically:
Crosson's (1985) Model of Subcortical Language Function: Here is Crosson's view of dominant hemisphere speech production, shown in a horizontal section of the left hemisphere at the level of the thalamus. This is the sequence of events involved when turning an idea into a spoken sentence (we recommend tracing out the following pathway step by step with your forefinger, so as not to get lost) .....
1. Following initial formulation in frontal cortex (FOR), information is passed across .....
2. ..... via the ventral anterior thalamus (VA) .....
3. ..... and the pulvinar nucleus (PUL) .....
4. ..... to the temporoparietal "decoding" cortex (DEC), for preverbal semantic monitoring to take place. If no errors are detected, the information is then forwarded to the caudate nucleus (CA), instructing it .....
5. ..... to tell the globus pallidus (GP) .....
6. ..... to disinhibit the ventral anterior thalamus (VA) enough to .....
7. ..... allow the motor programming frontal cortex (MP) .....
8. ..... to issue the necessary speech muscle commands. [The diagram shows activation of MP, but not the resulting pyramidal tract activation.]
9. As speech proceeds, phonological monitoring takes place back in temporoparietal cortex (DEC), based upon information passed directly to it along the arcuate fasciculus (AF). It is this process (and possibly this pathway) which has become non-operational in cases of jargonaphasia.
10. Requests for phonological corrections, if and when necessary, are passed back down the arcuate fasciculus to request reprogramming.
11. When a phrase has been successfully spoken, an acknowledgement of success is passed to the caudate nucleus (CA), so that it can repeat the process with the next intended output phrase.
If this diagram fails to load automatically, it may be accessed separately at
Developed from a black and white original in Crosson (1985:281; Figure 4). Pathway numbering (yellow) added to fit accompanying explanation. This version Copyright © 2002, Derek J. Smith.
Note how all the subprocesses need to go on simultaneously, if the resulting speech output is to flow smoothly. Thus:
"While one segment of language is being executed in speech, the next segment is being programmed for motor execution, the ensuing segment is being verified for semantic content, and another segment is in the process of being formulated" (p280).
Little is known, however, about the mechanisms by which one "segment" of a sentence is "kept in abeyance" (p276) while the immediately preceding segment undergoes monitoring. It is no wonder, therefore, that language takes time to develop during infancy, and calls for "increasingly complex interactions between motor and sensory systems" as it does so (p271). Crosson also points out that the unit of semantic verification will get progressively bigger as language skill develops. In children [and adult second language learners?], it is probably at the level of the single word or morpheme, whilst in adults it is at phrase level or higher.
That is the end of our commentary. Now do the following exercise .....
Exercise to Consolidate
The above exercise will give you the essence of the language output process as a series of interlocking subprocesses all working simultaneously. In the example nothing went wrong, but you may care to experiment with both semantic and phonological errors at your leisure. You will find that you need to give a lot of thought to the mechanisms by which failures of semantic monitoring (at Step #4) or phonological monitoring (at Step #9) are (a) detected, and (b) manage to stop all the other processes in their tracks. A possibly useful metaphor here is the "bucket chain", for it is no good people piling bucket after bucket on you if you have snagged yourself on the one you already have. There are also similarities to the design of microprocessors, where messages are regularly cascaded from one processing module to the next, but each has the power to "issue an interrupt" to the one before. The biological equivalent of such interrupts is rarely discussed in cognitive science. There are also similarities in telecommunications, where each station in a network is given a direct line back to the one before it, so that it can request retransmissions as necessary, or even change the former's transmission speed to match its own processing speed. For more on conversational repair in general, click here, and for more on the telecommunications metaphor, click here or here.
Crosson, B. (1985). Subcortical functions in language: A working model. Brain and Language, 25:257-292.
Smith, D.J. (1997). Neuroanatomy for Students of Communication. Cardiff: UWIC.