Assessing Relative Linguistic Impairment With Model-Based Item Selection

Area of Science:

• Neuroscience
• Psycholinguistics
• Clinical Neurology

Background:

• Anomia, a common deficit after left-hemisphere stroke, impairs word retrieval.
• Existing tests may not precisely differentiate lexical and sublexical processing deficits.

Purpose of the Study:

• To introduce and psychometrically validate a novel picture naming test to assess specific naming impairments.
• To develop the relative linguistic impairment (RLI) score to quantify lexical and sublexical processing deficits.

Main Methods:

• Constructed two item sets targeting lexical and sublexical word retrieval using a computational model.
• Administered the RLI test to 91 chronic left-hemisphere stroke patients with anomia.
• Assessed RLI score psychometric properties and predictive validity for language behaviors, lesion location, and brain activation.

Main Results:

• RLI scores demonstrated adequate psychometric properties and clinical significance.
• RLI scores predicted spontaneous speech fluency beyond overall accuracy.
• Dissociations were found between RLI performance and other language tests; sublexical RLI correlated with apraxia of speech and perisylvian lesions, while lexical RLI correlated with deep white matter lesions.
• Functional brain imaging revealed RLI-related activation shifts in dorsal and ventral networks during naming.

Conclusions:

• The RLI assessment is a psychometrically sound clinical tool for evaluating anomia.
• RLI provides insights into the neural underpinnings of word retrieval deficits in stroke survivors.