The few studies looking systematically into the neurophysiological and neuropsychological components of available therapies for chronic aphasia are highly heterogeneous in nature. Results from these studies have, unsurprisingly, indicated heterogeneous results, such as dissimilar neural outcomes associated with neuropsychological gains. There is, therefore, no consensus of how a successful therapy- that is, one that produces a measurable language gain in either production or comprehension -impacts the functional language networks of the brain in a specific type of aphasia population.

A recent study has shown that inner speech (the imagination of speech) involves networks and areas dissociable from those implicated in speech production. Further, behavioural analysis has shown an interesting discrepancy between inner speech and overt speech (also called speech production) in a small chronic aphasia population: some participants elicited poor inner speech coupled with relatively intact overt speech, while others elicited relatively intact inner speech coupled with poor overt speech. This unexplored discrepancy implies that inner speech and speech production are dissociable, though share similar networks.

This discrepancy, and the notion that these speech components share a similar network, drives this study's hypothesis that improvement in speech production after rehabilitation might be facilitated by an intact inner speech network. Much as good athletes visualise their performance before the actual event in order to increase their chances of success, so too might intact inner speech facilitate speech production, helping to visualise the word in order to increase the success of produced speech.

By studying a specific component of speech-inner speech-in a relatively homogeneous population of chronic expressive aphasics, the present study provides an explicit, critical means of understanding neurophysiological (as assessed by functional magnetic resonance imaging) and neuropsychological (as assessed by language batteries and personal questionnaires/interviews) changes occurring during speech therapy.

As a secondary objective, this study will explore the effectiveness, feasibility and adherence to an at-home computerised aphasia software delivered via a portable tablet.

15 million people worldwide have a stroke each year, with 152,000 in the United Kingdom. Recent estimates suggest that roughly 33% of patients suffering a stroke develop aphasia, a loss or impairment of language function caused by brain damage, which can have a significant impact on all aspects of an individual's life, as well as that of their carers. Aphasia can often be long-term, or chronic, affecting patients at least a year or more after their initial stroke.

Few studies have systematically investigated the effects of rehabilitation on brain mechanisms recruited to support recovery in stroke. Studies in this area are highly heterogeneous. The heterogeneity largely stems from 'lesion-related or language deficit-related differences in the patients studied'. Participants across and sometimes within the few studies conducted in this area vary with regard to the type of aphasia or time-following-stroke. These studies also boast differences including the type and dosage of treatment, the type of scanning task used to evaluate the desired effects of treatment, and the type of data analysis employed. Results from these studies have, unsurprisingly, indicated dissimilar neural outcomes associated with neuropsychological gains, such as increased right hemisphere (contra-lateral) involvement, or, in contrast, increased peri-lesional activation. There is, therefore, no consensus of how 'successful therapies' (that is, those that elicit some kind of language gain, either in comprehension or production) impact the language networks of the brain.

"It is well known that individuals with aphasia differ greatly with often varying language patterns and associated lesions, and even study participants carefully selected for their deficit patterns are seldom, if ever, homogeneous" . People with aphasia will, and do, differ markedly. Given this predicament, it becomes necessary to systematically control the other parts of the study, which includes limiting the imaging tasks and analysis to one component of the language system and using a powerful design, such as a crossover with two therapeutic conditions.

This study therefore aims to use systematic methodology to add information to the diminutive body of literature concerning chronic aphasia rehabilitation by exploring a specific component of the language network, inner speech, and its potential influence on speech production (neurophysiologically and neuropsychologically).

A previous study has shown that inner speech involves networks and areas dissociable from those implicated in speech production, such as the left inferior frontal gyrus, especially the pars opercularis and the supramarginal gyrus. Further, behavioural analysis has shown an interesting interaction of inner and overt speech in a small chronic aphasic population, whereas some chronic stroke patients showed poor inner speech coupled with good overt speech, while others showed good inner speech coupled with poor overt speech. This finding implies an unexplored relationship between the two networks: conceivably, that improvement in speech production during rehabilitation might be facilitated by an intact inner speech network.

There are hundreds of aphasia therapies on the market today. In general, two types of therapy exist: impairment-based and communication-based. Impairment-based therapies are those that specifically target increasing the ability of components of the language system, such as naming, reading, writing and sentence structure, and comprise most of the therapies on the market. Communication-based therapies are more informal, aiming to stimulate conversation by any means. The most utilised therapies for aphasia include constraint-induced therapy, which involves constraining the participant to using only words and not gestures in their communication, thus hoping to free the individual of non-speech compensatory strategies; melodic intonation therapy, based on the observation that people with aphasia have a better success rate if singing words rather than just saying words, uses melody as a crucial component for relearning speech; and various phonological cueing or naming therapies, which use repetition, semantic and phonological cueing based on specific anomia (naming) deficits. Aphasia therapies are so prevalent because, unlike drug therapies, they carry very little risk. This does not mean, however, that each aphasia therapy on the market is effective for all types of aphasia deficits. The scientific community lacks understanding of these therapies in several facets: understanding which treatments produce language gains in specific populations (i.e., chronic vs. acute individuals, fluent vs. non-fluent aphasia types); how language gains map onto changes in neurological function; and the trajectory of language gains over time, neuropsychologically and neurologically.

Utilising an at-home computerised aphasia rehabilitation program, this study will explore whether inner speech can assist patients in restoring access to spoken language, therefore resulting in improvement of language production (as assessed by neuropsychological examinations) and instigating changes in functional networks (as assessed by functional magnetic resonance imaging).

The computerised therapy was chosen because of its detailed clinical output system, its ability to be personalised to each individual, and its ability to adapt difficulty levels to the needs of the user. A successful at-home therapy program may provide a means to combat the lacking resources for continued rehabilitation outside of acute, hospital settings. As a secondary outcome, this study will investigate the success, feasibility and adherence to this software by collecting qualitative patient feedback and by analysing the software's quantitative outputs, such as exercise completion and number of times attempted, total time used and overall performance on the exercises.

Inclusion Criteria:

• Left hemisphere stroke
• Clinical presentation of first ever stroke
• Nonfluent/Expressive Aphasia: impairment in language production and spared language comprehension
• Age >18 years
• Adequate co-operation for scanning
• Right handed before stroke as tested with the Edinburgh inventory
• Native British-English speakers (this is due to the nature of the fMRI task and inner speech battery, which rely upon words that are rhymes or homophones in the British English language)
• No history of neurological or psychiatric disorders
• No current specific cognitive deficit other than the language deficit
• No contra-indication to MRI scan as indicated by the WBIC protocol
• Patients able to lie flat in the scanner for 2 hours
• Consent obtained prior to initiating the study from the patient, in accordance with Local Research Ethics Committee guidelines
• Stroke and subsequent aphasia having been present for more than 12 months (ie, chronic)

Exclusion Criteria:

For successful fMRI scans (relevant for all participants):

• Women with any chance of pregnancy
• Claustrophobia
• Any contra-indication to MRI as indicated by the WBIC protocol
• Concomitant medical disorder that means the patient is unable to lie flat comfortably in the scanner for a maximum of 2 hours (e.g. poorly controlled or severe respiratory disease or severe joint disease)

All recruited patients:

• History of significant pre-morbid cognitive impairment
• Alcohol or illicit drug abuse
• Severe deafness or visual impairment
• History of significant neurological disease (e.g. epilepsy, multiple sclerosis)
• Major organ failure that may complicate imaging studies (e.g. significant cardiac or liver disease)

Of those patients recruited, further exclusion from crossover study:

• Demonstration of intact inner speech with good overt speech
• Demonstration of poor inner speech with poor overt speech