Long-range neural synchronization supports fast and efficient reading: EEG correlates of processing expected words in sentences

Abstract

Word reading is heavily influenced by the information provided by previous context. In this study, we analyzed the neurophysiological bases of sentence reading through the EEG activity elicited during reading the same word embedded in differently constraining contexts: a) a low-constraining context; b) a high-constraining semantic compositional context; c) a high-constraining collocational context in which the item was in final position of a multi-word fixed-order expression. Cloze-probability of the two high-constraining contexts was equated. Before reading the target word we observed increased EEG gamma phase synchronization for the high-constraining compositional context and increased EEG theta synchronization for the collocational context (both compared to the low-constraining condition). After reading the target word we observed increased frontal positive EEG evoked activity (~220ms) for the high-constraining compositional context but an even earlier (~120ms) effect for the high-constraining collocational condition that was distributed over the scalp. A positive correlation was found only between the increased theta synchronization and the early EEG effect for the high-constraining collocational condition. Results indicate that long-range frontal-occipital interactions in the theta band - indexing working memory operations - support early visual-orthographic analysis of an incoming stimulus (such as the expected word); gamma-phase synchronization better represents binding operations between feed-forward activation and matching feedback. These data suggest that internal linguistic knowledge stored in long-term memory - if unambiguously pre-activated - supports the low-level perceptual processes involved in reading.

Figure 1

Cloze-probability values (with corresponding standard error bars computed over items) for the three experimental conditions along the sentence segment of interest starting from two words before the target word (TW−2) until the target word (TW). Conditional probability of TW when participants had to report TW−1 is also indicated (TW if TW−1).

Figure 2

Upper panel: Event Related Potentials (ERPs) elicited time-locked to the reading of the word preceding the target (TW−1) until 800 ms post-target word (TW) onset for each experimental condition. On the right we report a detail of the early ERP effects time-locked to TW in the electrode F4. Negative voltages are plotted up. Lower panel: voltage maps of the ERPs elicited by the TW in the two high-constraining conditions minus the low-constraining condition. We distinguish the distribution of the earlier effects from the later ERP effects (distinguished in time-windows of 100 ms shifting for 50 ms in time).

Figure 3

Phase synchrony charts and scalp topography. (A) The time-frequency maps per experimental condition. Experimental conditions are indicated at the top of the maps. Frequency range and time are indicated in the y and x axis of the maps respectively. Color bars at the right side of the maps show the magnitude of the phase-locking value (in standard deviation units). Vertical lines indicate the onset of TW−1, TW and TW+1 respectively. The black rectangles in the maps indicate the time-frequency intervals that are significantly different between conditions [p<0.05]. (B) Topographical significant differences of phase synchrony between experimental conditions. Experimental conditions compared are indicated at the top of the scalps. Oscillation frequency and temporal window are indicated at the top and bottom of each scalp respectively. Time 0 indicates the onset of the TW. Black lines connect pairs of electrodes displaying significantly larger synchronization [p_FDRcorr<0.002].

Figure 4

Correlation maps reflecting Pearson correlation coefficients (−1 to 1) both between increased pre-TW Theta synchronization for the HC-Collocation condition and post-TW ERP voltage differences (HC-Collocation minus LC-Control) in specific time intervals of interest for each electrode (left maps) and between increased pre-TW Gamma synchronization for the HC-Composition condition and post-TW ERP voltage difference (HC-Composition minus LC-Control) in specific time intervals of interest for each electrode (right maps). Significant correlations are indicated by white asterisks.