Matthew W. Lowder

Language structure in the brain: A fixation-related fMRI study of syntactic surprisal in reading.

How is syntactic analysis implemented by the human brain during language comprehension? The current study combined methods from computational linguistics, eyetracking, and fMRI to address this question. Subjects read passages of text presented as paragraphs while their eye movements were recorded in an MRI scanner. We parsed the text using a probabilistic context-free grammar to isolate syntactic difficulty. Syntactic difficulty was quantified as syntactic surprisal, which is related to the expectedness of a given words syntactic category given its preceding context. We compared words with high and low syntactic surprisal values that were equated for length, frequency, and lexical surprisal, and used fixation-related (FIRE) fMRI to measure neural activity associated with syntactic surprisal for each fixated word. We observed greater neural activity for high than low syntactic surprisal in two predicted cortical regions previously identified with syntax: left inferior frontal gyrus (IFG) and less robustly, left anterior superior temporal lobe (ATL). These results support the hypothesis that left IFG and ATL play a central role in syntactic analysis during language comprehension. More generally, the results suggest a broader cortical network associated with syntactic prediction that includes increased activity in bilateral IFG and insula, as well as fusiform and right lingual gyri.

Complex Sentence Processing: A Review of Theoretical Perspectives on the Comprehension of Relative Clauses

A major goal of psycholinguistics is to gain a better understanding of how syntactically complex sentences are processed. Pursuit of this goal has frequently focused on the contrast between objectand subject-extracted relative clauses (RCs). Although a large body of literature demonstrates that comprehension is more difficult for object RCs than for subject RCs, the proposed explanations for this processing asymmetry are diverse and hotly debated. This article reviews theoretical accounts of RC processing in terms of whether they characterize the critical differences in comprehension difficulty as arising from memory processes, interpretive processes, or processes tuned to the frequency with which different types of language are encountered. Understanding the cognitive mechanisms that enable us to process and comprehend syntactically complex sentences is a central goal of psycholinguistics. The contrast between subject-extracted and object-extracted relative clauses (RCs) has provided an empirically rich test bed for pursuing this goal. In a subject-extracted RC (SRC), as in (1), the head noun phrase (NP) serves as the subject of the RC, whereas in an object-extracted RC (ORC), as in (2), the head NP serves as the object of the RC. According to standard linguistic accounts, both SRCs and ORCs contain a phonologically empty placeholder—or gap—that is co-indexed with the head NP (the senator in the examples below). In (1), this gap (denoted by D) appears in the subject position of the embedded verb (e.g., the senator bothered the reporter), whereas in (2), this gap appears in the object position of the embedded verb (e.g., the reporter bothered the senator). In order to understand the sentence, the listener or reader must use information from the filler to interpret the gap, which would otherwise lead to an ungrammatical sentence.

The sentence-composition effect: Processing of complex sentences depends on the configuration of common noun phrases versus unusual noun phrases

In 2 experiments, the authors used an eye tracking while reading methodology to examine how different configurations of common noun phrases versus unusual noun phrases (NPs) influenced the difference in processing difficulty between sentences containing object- and subject-extracted relative clauses. Results showed that processing difficulty was reduced when the head NP was unusual relative to the embedded NP, as manipulated by lexical frequency. When both NPs were common or both were unusual, results showed strong effects of both commonness and sentence structure, but no interaction. In contrast, when 1 NP was common and the other was unusual, results showed the critical interaction. These results provide evidence for a sentence-composition effect analogous to the list-composition effect that has been well documented in memory research, in which the pattern of recall for common versus unusual items is different, depending on whether items are studied in a pure or mixed list context. This work represents an important step in integrating the list-memory and sentence-processing literatures and provides additional support for the usefulness of studying complex sentence processing from the perspective of memory-based models.

It's Hard to Offend the College: Effects of Sentence Structure on Figurative-Language Processing.

Previous research has given inconsistent evidence about whether familiar metonyms are more difficult to process than literal expressions. In 2 eye-tracking-while-reading experiments, we tested the hypothesis that the difficulty associated with processing metonyms would depend on sentence structure. Experiment 1 examined comprehension of familiar place-for-institution metonyms (e.g., college) when they were an argument of the main verb and showed that they are more difficult to process in a figurative context (e.g., offended the college) than in a literal context (e.g., photographed the college). Experiment 2 demonstrated that when they are arguments of the main verb, familiar metonyms are more difficult to process than frequency-and-length-matched nouns that refer to people (e.g., offended the leader), but that this difficulty was reduced when the metonym appeared as part of an adjunct phrase (e.g., offended the honor of the college). The results support the view that figurative-language processing is moderated by sentence structure. When the metonym was an argument of the verb, the results were consistent with the pattern predicted by the indirect-access model of figurative-language comprehension. In contrast, when the metonym was part of an adjunct phrase, the results were consistent with the pattern predicted by the direct-access model.

Natural forces as agents: reconceptualizing the animate-inanimate distinction.

Research spanning multiple domains of psychology has demonstrated preferential processing of animate as compared to inanimate entities--a pattern that is commonly explained as due to evolutionarily adaptive behavior. Forces of nature represent a class of entities that are semantically inanimate but which behave as if they are animate in that they possess the ability to initiate movement and cause actions. We report an eye-tracking experiment demonstrating that natural forces are processed like animate entities during online sentence processing: they are easier to integrate with action verbs than instruments, and this effect is mediated by sentence structure. The results suggest that many cognitive and linguistic phenomena that have previously been attributed to animacy may be more appropriately attributed to perceived agency. To the extent that this is so, the cognitive potency of animate entities may not be due to vigilant monitoring of the environment for unpredictable events as argued by evolutionary psychologists but instead may be more adequately explained as reflecting a cognitive and linguistic focus on causal explanations that is adaptive because it increases the predictability of events.

Word recognition during reading: The interaction between lexical repetition and frequency

Memory studies utilizing long-term repetition priming have generally demonstrated that priming is greater for low-frequency than for high-frequency words and that this effect persists if words intervene between the prime and the target. In contrast, word-recognition studies utilizing masked short-term repetition priming have typically shown that the magnitude of repetition priming does not differ as a function of word frequency and does not persist across intervening words. We conducted an eyetracking-while-reading experiment to determine which of these patterns more closely resembles the relationship between frequency and repetition during the natural reading of a text. Frequency was manipulated using proper names that were either high-frequency (e.g., Stephen) or low-frequency (e.g., Dominic). The critical name was later repeated in the sentence, or a new name was introduced. First-pass reading times and skipping rates on the critical name revealed robust repetition-by-frequency interactions, such that the magnitude of the repetition-priming effect was greater for low-frequency than for high-frequency names. In contrast, measures of later processing showed effects of repetition that did not depend on lexical frequency. These results are interpreted within a framework that conceptualizes eye-movement control as being influenced in different ways by lexical- and discourse-level factors.

Effects of animacy and noun-phrase relatedness on the processing of complex sentences

Previous work has suggested that syntactically complex object-extracted relative clauses are easier to process when the head noun phrase (NP1) is inanimate and the embedded noun phrase (NP2) is animate, as compared with the reverse animacy configuration, with differences in processing difficulty beginning as early as NP2 (e.g., The article that the senator . . . vs. The senator that the article . . .). Two eye-tracking-while-reading experiments were conducted to better understand the source of this effect. Experiment 1 showed that having an inanimate NP1 facilitated processing even when NP2 was held constant. Experiment 2 manipulated both animacy of NP1 and the degree of semantic relatedness between the critical NPs. When NP1 and NP2 were paired arbitrarily, the early animacy effect emerged at NP2. When NP1 and NP2 were semantically related, this effect disappeared, with effects of NP1 animacy emerging in later processing stages for both the related and arbitrary conditions. The results indicate that differences in the animacy of NP1 influence early processing of complex sentences only when the critical NPs share no meaningful relationship.

Individual differences in the perceptual span during reading: Evidence from the moving window technique

We report the results of an eye tracking experiment that used the gaze-contingent moving window technique to examine individual differences in the size of readers’ perceptual span. Participants read paragraphs while the size of the rightward window of visible text was systematically manipulated across trials. In addition, participants completed a large battery of individual-difference measures representing two cognitive constructs: language ability and oculomotor processing speed. Results showed that higher scores on language ability measures and faster oculomotor processing speed were associated with faster reading times and shorter fixation durations. More interestingly, the size of readers’ perceptual span was modulated by individual differences in language ability but not by individual differences in oculomotor processing speed, suggesting that readers with greater language proficiency are more likely to have efficient mechanisms to extract linguistic information beyond the fixated word.

Focus takes time: structural effects on reading

Previous eye-tracking work has yielded inconsistent evidence regarding whether readers spend more or less time encoding focused information compared with information that is not focused. We report the results of an eye-tracking experiment that used syntactic structure to manipulate whether a target word was linguistically defocused, neutral, or focused, while controlling for possible oculomotor differences across conditions. As the structure of the sentence made the target word increasingly more focused, reading times systematically increased. We propose that the longer reading times for linguistically focused words reflect deeper encoding, which explains previous findings showing that readers have better subsequent memory for focused versus defocused information.

The Manuscript That We Finished: Structural Separation Reduces the Cost of Complement Coercion.

Two eye-tracking experiments examined the effects of sentence structure on the processing of complement coercion, in which an event-selecting verb combines with a complement that represents an entity (e.g., began the memo). Previous work has demonstrated that these expressions impose a processing cost, which has been attributed to the need to type-shift the entity into an event in order for the sentence to be interpretable (e.g., began writing the memo). Both experiments showed that the magnitude of the coercion cost was reduced when the verb and complement appeared in separate clauses (e.g., The memo that was begun by the secretary; What the secretary began was the memo) compared with when the constituents appeared together in the same clause. The moderating effect of sentence structure on coercion is similar to effects that have been reported for the processing of 2 other types of semantically complex expressions (inanimate subject-verb integration and metonymy). We propose that sentence structure influences the depth at which complex semantic relationships are computed. When the constituents that create the need for a complex semantic interpretation appear in a single clause, readers experience processing difficulty stemming from the need to detect or resolve the semantic mismatch. In contrast, the need to engage in additional processing is reduced when the expression is established across a clause boundary or other structure that deemphasizes the complex relationship.