Effect of Lexical-Semantic Cues during Real-Time Sentence Processing in Aphasia

Abstract

Using a visual world eye-tracking paradigm, we investigated the real-time auditory sentence processing of neurologically unimpaired listeners and individuals with aphasia. We examined whether lexical-semantic cues provided as adjectives of a target noun modulate the encoding and retrieval dynamics of a noun phrase during the processing of complex, non-canonical sentences. We hypothesized that the real-time processing pattern of sentences containing a semantically biased lexical cue (e.g., the venomous snake) would be different than sentences containing unbiased adjectives (e.g., the voracious snake). More specifically, we predicted that the presence of a biased lexical cue would facilitate (1) lexical encoding (i.e., boosted lexical access) of the target noun, snake, and (2) on-time syntactic retrieval or dependency linking (i.e., increasing the probability of on-time lexical retrieval at post-verb gap site) for both groups. For unimpaired listeners, results revealed a difference in the time course of gaze trajectories to the target noun (snake) during lexical encoding and syntactic retrieval in the biased compared to the unbiased condition. In contrast, for the aphasia group, the presence of biased adjectives did not affect the time course of processing the target noun. Yet, at the post-verb gap site, the presence of a semantically biased adjective influenced syntactic re-activation. Our results extend the cue-based parsing model by offering new and valuable insights into the processes underlying sentence comprehension of individuals with aphasia.

Keywords: aphasia; eye tracking; real-time sentence processing; semantic cue; syntax.

Figure A1

Gaze proportion differences toward N3 between conditions and groups. Solid lines represent observed data and dashed lines represent the GCA model fit.

Figure A2

Gaze proportion differences to N1 and N2 between groups. Solid lines represent observed data and dashed lines represent the GCA model fit.

Figure 1

An illustration of the overall pattern of lexical processing in an ongoing sentence that involves an activation and de-activation phase. Activation is represented by increase in gaze proportion toward the heard item in the sentence while de-activation is represented by reduction in gaze proportion over time.

Figure 2

Example of a visual world eye-tracking paradigm. The speaker represents the auditory sentence.

Figure 3

Specified windows of interest. The arrows represent when in the sentence a prespecified window starts and ends. The windows of analysis were overlapping as we wanted to capture the full morphology of the gaze pattern toward a targeted image. In these windows, we capture the activation (gazes toward) and deactivation (gazes away) parts of lexical processing. Here, we divided our sentence into four analysis windows to capture processing patterns via the gaze dynamics to the three images of the nouns that were mentioned in the sentence (here N1 represents the illustration of eagle, N2 the snake, N3 the bear).

Figure 4

Mean gaze over time toward N1 (first noun, solid salmon line), N2 (second noun, solid green line), N3 (third noun, solid blue line), and a distractor image (solid purple line) averaged across conditions for each group which begins at the auditory onset of the sentence (N1), N4 (unrelated noun, purple line). Shaded areas represent 95% confidence intervals within subject. The dotted salmon line represents mean offset N1; the dotted green line represents mean offset N2; the dotted blue line represents mean offset N3; the dotted black line represents mean offset of the verb.

Figure 5

The plot captures the part of the sentence as “the eagle saw the/adjective/snake”. This plot demonstrates the gaze proportion differences to N1 between conditions and groups. Solid lines represent observed data and dashed lines represent the GCA model fit. The graphic representation of the model is showing the quadratic fit, yet the significant results for the condition effect were observed at the linear term.

Figure 6

The plot captures the following part of the sentence “/adjective/snake that the bear”. This plot demonstrates the gaze proportion differences to N2 between conditions and groups. Solid lines represent observed data and dashed lines represent the GCA model fit. The graphic representation of the model is showing the quadratic fit, yet the marginal results for the group effect were observed at the intercept level, in addition to the interaction effect which was significant at the linear term.

Figure 7

Averaged gaze proportions to N1, N2, and N3 between groups and conditions. This is the raw data observation of gaze toward N1, N2, and N3 in the verb-frame window. The gray shaded ribbons around the lines represent standard errors. The red boxes indicate the window in which the effect is expected.