Search

Introduction: Considerable changes in the cortical representation of language processing can occur following stroke. Both left and right hemisphere regions are thought to support language recovery; however the relative contributions of each hemisphere and neural mechanisms mediating this process are not well understood (Meltzer et al., 2013; Price & Crinion, 2005; Saur et al., 2006; Thompson, 2000; 2010). It is generally thought that recovery of function in perilesional areas offers the best prognosis for clinical improvement (Heiss, 2003; Heiss & Thiel, 2006). Therefore, assessing the functionality of these areas is essential to targeting interventions. One potential biomarker of perilesional function and dysfunction relates to abnormalities in spontaneous neural activity (Meinzer et al., 2007; Poza et al., 2007). Perilesional tissue produces a large amount of high amplitude slow-wave activity, which can be quantified using algorithms that examine frequency spectra or complexity (e.g. multiscale entropy: MSE). This slowed spontaneous activity is a marker of subtle neural damage associated with the long-term effects of stroke beyond the primary infarct zone, and may be an indicator of “functional lesion” extent. Many questions remain unanswered regarding its potential associations with 1) cognitive dysfunction, 2) anatomical damage such as white matter disconnection, and 3) decreased blood flow (hypoperfusion). Method: In the present study we used magnetoencephalography (MEG) to clarify the roles of dorsal and ventral language pathways in processing of semantic and syntactic information in healthy adults and participants with stroke-induced aphasia. In addition, we used functional and structural imaging to understand the roles of perilesional and contralesional activity in recovery from post-stroke aphasia, and to explore the potential of right hemisphere activity to support recovery. Using resting MEG and ASL, we developed a paradigm to identify perilesional cortex that is structurally intact, but not functioning optimally. We mapped the distribution of spontaneous slow-wave activity, its relationship to task-related activation during language comprehension, and decreased blood flow (hypoperfusion) in a group of stroke participants. Results: We found that in healthy controls, activation of a temporo-frontal “ventral network” was involved in semantic processing, and a fronto-parietal “dorsal network” was associated with syntax (Figure 1A). Patients with aphasia activated homologous RH areas and dorsal LH regions; however they failed to activate ventral cortex adjacent to the lesion (Figure 1B). In addition, these perilesional regions consistently produced slow-wave activity, indicating that they are dysfunctional even though the tissue was not infarcted (Figure 1 C and D). Furthermore, abnormalities in spontaneous neural activity were associated with less task-related activation, and hypoperfusion in the affected regions These results suggest that the reduced task-related responses in perilesional tissue and the degree of RH recruitment during language processing, are related to abnormal slowing of neural activity, and to reduced blood flow (hypoperfusion). Reversal of such abnormalities may be a fruitful target for interventions such as noninvasive brain stimulation. Conclusion: Our results provide a basis for identifying a relationship between functional, structural and vascular markers of intact/impaired language function and recovery in patients with post-stroke aphasia.

Recently, cross-situational word learning (XSL) has gained attention as a viable mechanism for learning the meaning of words. XSL refers to the tracking of wordreferent pairs over a series of different exposures (Siskind, 1996; Yu & Smith, 2007). However, it has recently been suggested that word learning likely functions by the extraction of information from several cues, such as perceptual statistical, social, and linguistic (Hirsh-Pasek, Golinkoff, & Hollich, 2000). In this work, I measured the contribution of co-occurrence frequencies (statistical cues) and syntactic-semantic (linguistic cues) links in a sentential XSL experiment that contained both nouns and verbs. Participants were exposed to three different learning conditions that varied in the mode of presentation. Participants either learned nouns and verbs together in one phase (unstaggered learning), or learned in two phases, initially learning the subset of nouns (staggered-with-nouns) or verbs (staggered-with-verbs) first. With unstaggered learning, participants could use primarily only XSL, while with staggered learning, participants could use both XSL and syntactic bootstrapping. Results demonstrated that the combination of syntactic bootstrapping and XSL significantly facilitated word learning. The results support an integrative and comprehensive model of word learning (e.g. Hirsh-Pasek et al., 2000). In the general discussion, the relationship between an integrative view of language and a domain-general account of language is discussed.