Temporal dynamics of linguistic processes are reorganized in aphasics' cortex: an EEG mapping study

Abstract

Brain lesions are known to elicit reorganization of function in representational cortex. Using linguistic function as an example, we show that (a) injury-related reorganization may also be observed in language-related cortex and (b) this reorganization not only appears in cortical space but also in the dynamic flow of activity. The present study investigated cortical reorganization in a group of 10 nonfluent aphasics who demonstrated partial recovery of linguistic functions. Compared with controls, linguistic functions were organized in an atypical manner, both in terms of spatial structures involved and in the time course of the linguistic processes, from word reading to late stages of word encoding in working memory. For this purpose, event-related potentials were recorded in a two-stimulus design comprising phonological and semantic tasks. Subjects were asked to judge whether two words, separated by a 2-s interval, rhymed (phonological task) or were semantically associated. During word reading of the phonological task, controls showed negativity/activation over occipital sites, whereas patients displayed negativity at left-medial orbitofrontal locations anterior to the common sites of lesion. During the subsequent 2-s interval associated with word encoding, the two groups showed a reversed pattern: significant left–right anterior asymmetry prevailed in controls, whereas lateralization was absent in patients. Aphasics displayed maximum positivity/inhibition over the left frontal regions, at the typical site of lesion. Compared with controls, patients exhibited significant disinhibition (decreased positivity) of right frontal areas and greater activation of left temporal sites. These results suggest that the concept of language plasticity should include, in addition to spatial aspects of linguistic reorganization, the reorganized temporal dynamics associated with recovery of impaired functions.

Introduction

Research on brain system plasticity focuses on alterations of functional organization, i.e., reorganization in representational cortex. In humans, several experiments have demonstrated functional reorganization of somatosensory areas in healthy subjects—e.g., string players—with highly specialized use of hands (Elbert et al., 1995), in Braille readers (Sterr et al., 1998), and also in patients, e.g., affected by focal hand dystonia (Elbert et al., 1998) or amputees suffering from phantom limb pain Elbert et al., 1994, Flor et al., 1995. In the human brain, motor (Stefan et al., 2000) and parietal cortices (Bavelier et al., 2001) have also shown capacity for plastic changes. Compared with sensory and motor functioning, higher level processes such as those involved in language perception and production have been shown to involve neural networks relatively more widespread in the cortex, mainly within the left hemisphere, but also in subcortical structures and right cerebral cortex Cabeza and Nyberg, 1997, Fiez and Petersen, 1998, Sterr et al., 1998.

Recently, several investigations have been performed in order to clarify whether and how cortical reorganization occurs in aphasic patients. Imaging of cerebral blood flow using PET and, less frequently, fMRI, has demonstrated reorganization of linguistic function both within the left hemisphere Karbe et al., 1998, Warburton et al., 1999, also in the proximity of the damaged area (Belin et al., 1996), and in the right hemisphere, in regions more or less homologous with those damaged in the left hemisphere Buckner et al., 1996, Calvert et al., 2000 fMRI; Cappa et al., 1997, Gold and Kertesz, 2000, Lazar et al., 2000, Müller et al., 1999a, Müller et al., 1999b, Musso et al., 1999, Ohyama et al., 1996, Weiller et al., 1995. In most of these investigations, language-related activity has been found in both hemispheres, although the relative contribution of one hemisphere over the other and the location of language-activated areas was quite variable across studies.

Compared with PET and fMRI research, fewer studies have used EEG or evoked potentials for locating inter- or intrahemispheric reorganization of language in patients after recovery. A number of recent ERP studies found significant differences in linguistic processing (lexical, semantic, and syntactic) between aphasics and controls Friederici et al., 1999, Hagoort et al., 1996, Swaab et al., 1997. Another set of studies used electrophysiological methods to locate linguistic activation in aphasics: some studies pointed to greater activation of right hemisphere in aphasics Moore, 1986, Selinger et al., 1989, Thomas et al., 1997, but there is also evidence of left frontal Cohen et al., 2001, Thomas et al., 1997 and temporal (Dobel et al., 2002) cortex activation in these patients.

The partial inconsistency of the results obtained so far probably depends on the heterogeneity of the patients studied, the extent and location of their lesions, and the variety of experimental designs, tasks, and stimuli used. Nevertheless, part of this inconsistency may be due to the fact that different cortical maps are dynamically activated by specific functional features (e.g., lexical, semantic, syntactic, phonological) of linguistic stimuli which follow one another across time. With brain-imaging methods such as PET and fMRI, many cortical regions seem to be simultaneously coactivated, whereas instead they may infact be activated sequentially (Elbert and Keil, 2000). There is currently evidence of phenomena of dynamic cortical reorganization in which different somatosensory maps may be dynamically switched and modulated Braun et al., 2000, Braun et al., 2002.

The main innovation of the present study is an attempt to measure both spatial aspects of cortical plasticity and the temporal dimension of the reorganizational processes associated with the functional recovery of language in aphasics. We hypothesized the existence of spatial reorganization of linguistic processes in aphasics' cortex, depending on the different temporal phases of linguistic processes. In particular, we tried to distinguish the cortical networks activated by word reading (the early stage of word processing) from networks involved in verbal working memory and phonological/semantic word elaboration (late stage). The use of evoked potentials and their high time resolution was therefore mandatory to test the above hypothesis.

The second aim of the present experiment was to overcome some methodological limits of past investigations. A number of contrivances were adopted. First, in order to control the phenotypic variability of patients' lesions, a relatively homogenous sample of nonfluent Broca aphasics was selected. Second, to avoid psychophysical and word-category systematic effects, the same sample of linguistic stimuli was used in different tasks. We have shown in previous cross-linguistic validation that the same words activate different neural networks, depending on the task in which they are used Angrilli et al., 2000, Elbert et al., 1999. Third, the adoption of an eye artifact modeling method termed MSEC Berg and Scherg, 1991, Berg and Scherg, 1994 yielded more precise and reliable scalp mapping of electrical brain activity, which avoided the typical underestimation of frontal lobe activity produced by standard eye movement correction methods.