Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
Subjective self-assessment is a frequently used method when it comes to Quality of Experience assessment in general. If correctly used a variety of scales can be applied to assess the quality of cloud gaming for different constructs such as experienced quality or flow. Besides self-assessment, physiological correlates are a promising method to measure the influence of cloud gaming quality on the player without the interruption introduced by the rating task. We present subjective and physiological results and lessons learned from a laboratory study in which 32 participants played a first person shooter game in a cloud gaming setup with varying levels of video quality caused by different video compression bitrates. We found that the video quality influenced the perceived quality, player experience, the valence rating, and the alpha frequency band power. It is shown that 1) subjective methods assess the quality variation and 2) physiological measures capture the influence on the player in terms of a reduced cognitive state. Taken together, test set-ups will benefit from a mixed methods approach in cloud gaming QoE testing.
This chapter uses data extracted from two recently compiled historical corpora of Australian English, the Corpus of Oz Early English (19th century) and AusCorp (20th century) to examine developments in ten morphosyntactic variables over the past two centuries. At the same time comparisons are drawn with British and American English, using data from ARCHER (A Representative Corpus of Historical English Registers). While Australian usage is found to have diverged from that of its British colonial parent, reflecting increasing independence from British linguistic norms, it has shifted towards that of American English – the new centre of gravity of grammatical change in English world-wide – which emerges as the most advanced variety on eight of the ten variables. Keywords: grammar; Australian English; diachronic; corpus; variation
Abstract This paper surveys methods for annotation error detection and correction. Methods can broadly be characterized as to whether they detect inconsistencies with respect to some statistical model based only on the corpus data or whether they detect inconsistencies with respect to a grammatical model, in general, some external information source. Two extended examples are presented, illustrating these different techniques: (1) the variation n ‐gram method, which searches for inconsistences in annotation for identical strings; and (2) a method of ad hoc rule detection, for syntactic annotation, which compares treebank rules to a grammar to determine which are anomalous. Methods for detecting annotation errors have developed much over the last decade, and thus corpus practitioners can benefit greatly from them, while at the same time NLP researchers can learn more about the nuances of the annotation they use and see how error correction methods intersect with NLP techniques.
In this paper we present an approach for the enrichment of WSD knowledge bases with data-driven relations from a gold standard corpus (annotated with word senses, valency information, syntactic analyses, etc.). We focus on Bulgarian as a use case, but our approach is scalable to other languages as well. For the purpose of exploring such methods, the Personalized Page Rank algorithm was used. The reported results show that the addition of new knowledge improves the accuracy of WSD with approximately 10.5%.
International audience
Bidirectional Long Short-Term Memory Recurrent Neural Network (BLSTM-RNN) has been shown to be very effective for tagging sequential data, e.g. speech utterances or handwritten documents. While word embedding has been demoed as a powerful representation for characterizing the statistical properties of natural language. In this study, we propose to use BLSTM-RNN with word embedding for part-of-speech (POS) tagging task. When tested on Penn Treebank WSJ test set, a state-of-the-art performance of 97.40 tagging accuracy is achieved. Without using morphological features, this approach can also achieve a good performance comparable with the Stanford POS tagger.
We are proposing an extension of the recursive neural network that makes use of a variant of the long short-term memory architecture. The extension allows information low in parse trees to be stored in a memory register (the 'memory cell') and used much later higher up in the parse tree. This provides a solution to the vanishing gradient problem and allows the network to capture long range dependencies. Experimental results show that our composition outperformed the traditional neural-network composition on the Stanford Sentiment Treebank.
We study non-deterministic oracles for training non-projective beam search parsers with swap transitions. We map out the spurious ambiguities of the transition system and present two non-deterministic oracles as well as a static oracle that minimizes the number of swaps. An evaluation on 10 treebanks reveals that the difference between static and non-deterministic oracles is generally insignificant for beam search parsers but that non-deterministic oracles can improve the accuracy of greedy parsers that use swap transitions.
For many students, coming to learn mathematics is as much about the pedagogical relay through which concepts are conveyed as it is about the mathematics per se. This relay comprises social, cultural and linguistic norms as well as the mathematical discourse. In this study, I outline the practices of one remote school and how the teaching practices scaffold Indigenous learners whose home language is different from the language of instruction (Standard Australian English) as they come to learn mathematics. Through various strategies, teachers have created positive learning environments that celebrate the home languages of the students while supporting the transition into Standard Australian English and the discourse of mathematics.
Universal Dependencies is a project that seeks to develop cross-linguistically consistent treebank annotation for many languages, with the goal of facilitating multilingual parser development, cross-lingual learning, and parsing research from a language typology perspective. The annotation scheme is based on (universal) Stanford dependencies (de Marneffe et al., 2006, 2008, 2014), Google universal part-of-speech tags (Petrov et al., 2012), and the Interset interlingua for morphosyntactic tagsets (Zeman, 2008). This is the second release of UD Treebanks, Version 1.1.
Using functional near-infrared spectroscopy, the present study investigated how listening to differently valenced music is associated with changes in hemoglobin concentrations in the prefrontal cortex area, indicating changes in neural activity. Thirty healthy people (15 men; M age = 24.8 yr., SD = 2.4; 15 women; M age = 25.2 yr., SD = 3.1) participated. Prefrontal cortex activation, emotional responses (heart rate variability), and self-reported affective ratings were measured while listening to calm and motivational music. The songs were presented in a random counterbalanced order and separated by periods of white noise. Mixed-model repeated-measures analysis of variance (ANOVA) evaluated the relationships for main effects and interactions. The results showed that music was associated with increased activation of the prefrontal cortex area. For both sexes, listening to the motivational song was associated with higher vagal withdrawal (lower HR) than the calm song. As expected, participants rated the motivational song with greater affective valence and higher arousal. Effects persisted longer in men than in women. These findings suggest that both the characteristics of music and sex differences may significantly affect the results of emotional neuroimaging in samples of young adults.
In the present study, we raised the question of whether valence information of natural emotional sounds can be extracted rapidly and unintentionally. In a first experiment, we collected explicit valence ratings of brief natural sound segments. Results showed that sound segments of 400 and 600 ms duration-and with some limitation even sound segments as short as 200 ms-are evaluated reliably. In a second experiment, we introduced an auditory version of the affective Simon task to assess automatic (i.e. unintentional and fast) evaluations of sound valence. The pattern of results indicates that affective information of natural emotional sounds can be extracted rapidly (i.e. after a few hundred ms long exposure) and in an unintentional fashion.
The usage of phrasemes evidences not only their variability, transformations and modifications, but also the most frequent forms of their realization (phraseme-types) and frequency (phraseme-tokens), i.e. phrasemes’ flexibility. In this paper, selected Lithuanian idiomatic predicate phrasemes are analysed in the Corpus of Contemporary Lithuanian Language, in the Phraseological Dictionary and in the lexical database of the Dictionary of Lithuanian Phrases. The results of comparison show that the corpus research can give rich evidence about the morphological flexibility of phrasemes. This information can help to improve representation of phrasemes in the phraseological dictionaries of Lithuanian, in order to make them more usage-based and more usage-oriented.
Large-scale data resources needed for progress toward natural language understanding are not yet widely available and typically require considerable expense and expertise to create. This paper addresses the problem of developing scalable approaches to annotating semantic frames and explores the viability of crowdsourcing for the task of frame disambiguation. We present a novel supervised crowdsourcing paradigm that incorporates insights from human computation research designed to accommodate the relative complexity of the task, such as exemplars and real-time feedback. We show that non-experts can be trained to perform accurate frame disambiguation, and can even identify errors in gold data used as the training exemplars. Results demonstrate the efficacy of this paradigm for semantic annotation requiring an intermediate level of expertise. 1 The semantic bottleneck Behind every great success in speech and language lies a great corpus—or at least a very large one. Advances in speech recognition, machine translation and syntactic parsing can be traced to the availability of large-scale annotated resources (Wall Street Journal, Europarl and Penn Treebank, respectively) providing crucial supervised input to statistically learned models. Semantically annotated resources have been comparatively harder to come by: representing meaning poses myriad philosophical, theoretical and practical challenges, particularly for general purpose resources that can be applied to diverse domains. If these challenges can be addressed, however, semantic resources hold significant potential for fueling progress beyond shallow syntax and toward deeper language understanding. This paper explores the feasibility of developing scalable methodologies for semantic annotation, inspired by three strands of work. First, frame semantics, and its instantiation in the Berkeley FrameNet project (Fillmore and Baker, 2010), offers a principled approach to representing meaning. FrameNet is a lexicographic resource that captures syntactic and semantic generalizations that go beyond surface form and part of speech, famously including the relationships among words like buy, sell, purchase and price. These rich structural relations provide an attractive foundation for work in deeper natural language understanding and inference, as attested by the breadth of applications at the Workshop in Honor of Chuck Fillmore at ACL 2014 (Petruck and de Melo, 2014). But FrameNet was not designed to support scalable language technologies; indeed, it is perhaps a paradigm example of a hand-curated knowledge resource, one that has required significant expertise, training, time and expense to create and that remains under development. Second, the task of automatic semantic role labeling (ASRL) (Gildea and Jurafsky, 2002) serves as an applied counterpart to the ideas of frame semantics. Recent progress has demonstrated the viability of training automated models using frameannotated data (Das et al., 2013; Das et al., 2010; Johansson and Nugues, 2006). Results based on FrameNet data have been limited by its incomplete
This study examines what effect mindfulness has on anxiety and memory levels in comparison to a suppression and control group. Participants underwent natural, suppression, and mindful conditions while being shown a series of positive and negative images and rating how happy/unhappy (valence) they felt and how excited/calm (arousal). After answering a series of questionnaires, participants were tested on their recall levels. The results revealed that arousal and valence ratings of the pictures were not significantly different across conditions, and neither was the recall of the participants. The results of this experiment do not align with previous research and may be due to a few limitations within the study. Therefore, more studies will need to be conducted and further research will need to be completed.
Faced with the need to access information in various languages about the cultural heritage of Italy, or to translate documents concerning this heritage, the “Lessico multilingue dei Beni Culturali” research group of Florence University has proposed creating a multilingual website in seven languages for the development of a cultural heritage lexicon dedicated to the city of Florence. In its pilot phase, this website will consist of a sample number of comparative lexical databases and a multilingual dictionary. The database texts on which the dictionary editors will base their work have been chosen for the purpose of combining the classic description of Florence’s cultural heritage with its contemporary reinterpretation: a database of the translations of Vasari's Lives, a key work for describing this heritage, and a database of the translations of tour guides of Florence. The examples offered for the Russian language illustrate the need to reserve an important role in the lexical card of the multilingual Portal not only to the grammatical description of the terms, but also to their graphic form, to their definition, if possible with a reflection on the etymology that takes the origins and the history of the word into account. Quotations from the original literature as well as from translations are equally important. This project can constitute an important advance in helping both the general and the specialized public to improve their understanding of cultural heritage issues.
This dissertation examines the tensions at work in contemporary French cultural politics between, on the one hand, homogenizing/assimilating hegemonic tendencies and, on the other hand, performances of heterogeneity/disharmony especially in literature but also in other artistic forms such as music. “Regional literature/culture” and “banlieue (ghetto) literature/culture” are studied as two major phenomena that performatively go against France’s “state monolingualism,” here understood as much as a “one-language policy” as the enforcement of “one discourse about Frenchness” (mono-logos). I rely on Jacques Rancière’s notions of “archipolitics” and “aesthetic regimes” to suggest that May 1968 has constituted an epistemological shift which has made it possible for alternate French discourses to emerge and become perceptible. Literatures displaying such discourses (either regional-related or immigration-related or both) are termed “accented literatures,” with “accent” being defined both as “variation from the linguistic norm” and “variation from the discursive norm.” These “accented literatures” become a distinctive trait of “democracy,” or “agonistic community,” allowing space for disharmonic representations of the “French” “nation.” Regarding regional (Alsatian) literature, I focus on André Weckmann’s literary use of magical surrealism and of a dialogic “Germanic French language”; regarding immigrant identity and banlieue literature, I first explicate the profound implications, for banlieue literature as a whole, of the “two-generation theory” developed by Algeria-born French rapper and writer Mounsi, with Azouz Begag’s literary production as a case study. Then turning to Abd al Malik, a French rapper/writer/filmmaker of Congolese origin, I pinpoint his concept of “pacific, new French revolution” as an ultimate form of accentuation of French discourse, scrutinizing the ways in which his art performs Frenchness as well as Islam. Because the notion of “accent” is closely linked to those of “prestige” and “legitimacy vs. lack thereof,” this dissertation eventually leads to a redefinition of “legitimate culture” in France. As a practical consequence of these literary-political debates, I advocate for the teaching, within the French public school system, of both regional languages and immigrant languages such as Arabic as a way to address identity challenges specific to the contemporary postcolonial era.
Abstract The two main classes of grammars are (a) hand-crafted grammars, which are developed by language experts, and (b) data-driven grammars, which are extracted from annotated corpora. This paper introduces a statistical method for mapping the elementary structures of a data-driven grammar onto the elementary structures of a hand-crafted grammar in order to combine their advantages. The idea is employed in the context of Lexicalized Tree-Adjoining Grammars (LTAG) and tested on two LTAGs of English: the hand-crafted LTAG developed in the XTAG project, and the data-driven LTAG, which is automatically extracted from the Penn Treebank and used by the MICA parser. We propose a statistical model for mapping any elementary tree sequence of the MICA grammar onto a proper elementary tree sequence of the XTAG grammar. The model has been tested on three subsets of the WSJ corpus that have average lengths of 10, 16, and 18 words, respectively. The experimental results show that full-parse trees with average F 1 -scores of 72.49, 64.80, and 62.30 points could be built from 94.97%, 96.01%, and 90.25% of the XTAG elementary tree sequences assigned to the subsets, respectively. Moreover, by reducing the amount of syntactic lexical ambiguity of sentences, the proposed model significantly improves the efficiency of parsing in the XTAG system.
One avenue for supporting the continued use and revitalization of endangered languages in the current, pervasively computerized world is the creation of computational models of the often rich and complex morphology of these languages. Such computational models can be used as a basis for creating a suite of reader’s and writer’s tools, including e.g. (1) an intelligent electronic dictionary that combines the computational model and a lexical database allowing for linking any inflected form with the appropriate dictionary entry, as well as the generation of word paradigms, (2) an intelligent computer-aided language learning application (ICALL) that allows for the dynamic generation of large numbers of exercises combining the entire core vocabulary (up to several thousand of the most common words) and a substantially smaller set of exercise templates, and (3) a spell-checker that supports adherence with one or more existing orthographical conventions, and thus the production of good-quality texts. Importantly, these tools can be made publicly available over the Internet and integrated as part of general software applications such as web browsers and word processors, to be used with little or no cost by any speakers or language-learners in the respective communities as well as any researchers, anywhere – instead of remaining on an individual researcher’s computer drive or on a library bookshelf. Based on our recent experiences on trying out various practical approaches in developing computational morphological models for Plains Cree and Northern Haida, using Finite-State Transducer (FST) technology (Beesley & Karttunen, 2003), once one gains access both to (a) a comprehensive set of full word paradigms, for every possible paradigm type, and (b) an accompanying extensive electronic lexical resource with coding indicating the relevant paradigm type, we have been able to create surprisingly rapidly, potentially within only several months, initial but already full-fledged FST models that can be readily adapted into the aforementioned software tools (1-3). Nevertheless, these first versions will certainly benefit from further work, where one cannot do without the active participation of the language community. However, we will demonstrate how, when a researcher or community linguist pays careful attention in their lexical documentation work on the systematic and detailed coding of the morphological characteristics of the vocabulary in some structured electronic format (e.g. when using software such as ToolBox), they will at the same time facilitate the rapid initial development of computational tools which will make benefits of their work available to the entire community. References Beesley, Kenneth R. and Lauri Karttunen. 2003. Finite State Morphology. Stanford (CA): CSLI Publications.
Wide-coverage resources for lexicalized grammars have been obtained by converting the existing treebanks into collections of derivations. Additional annotations to the source treebank can be used to improve these derivations. A treebank annotation called the NTT treebank was used for this paper to improve a CCGbank for Japanese. The source treebank of the CCGbank itself is created by automatically converting chunk-dependencies, but the CCGbank contains errors caused by noisier phrase structures and a lack of linguistic information, which is difficult to represent in chunk-dependency. The NTT treebank provides cleaner trees and functional and semantic information, e.g., coordinations and predicate-argument structures. The effect of the improvement process is empirically evaluated in terms of the changes in the dependency relations extracted from the resulting derivations.
BACKGROUND: Migraine is a disorder of periodic disabling headache. Facilitated cortical responsivity has been suggested as one predisposing factor. Although the underlying mechanisms of migraine attack onsets are not fully understood, facilitated cortical responsivity has been suggested as one predisposing factor. Here, we investigate if enhanced cortical responsivity is reflected in altered event-related potentials during processing of complex pictures. METHOD: Altogether, 16 migraine patients and 16 healthy volunteers participated in this study. Each patient had a diagnosed migraine and was headache- and medication-free for the study. Participants watched positive, negative and neutral pictures from the international affective picture system. An electroencephalogram was recorded during picture presentation. Afterwards, participants were asked to rate the pictures for valence and arousal. RESULTS: Migraine patients showed significantly more negative-going early event-related potential components from 100 ms to 180 ms to all picture categories over occipital regions as well as more positive-going late potentials over central regions. Patients and controls did not differ in valence and arousal ratings for the international-affective picture system. DISCUSSION: Patients with migraine seem to react cortically more intensely to all kinds of pictorial stimuli, regardless of emotional content. This facilitated processing may be related to the high cortical responsivity shown in various other event-related potential studies and might contribute to the recurring intense headache attacks.
The paper aims to evaluate the semantic -terminological effect on the use of the Latin lexicon that are shaped by a particular author, given the complete documentation of the occurrences of his lemmata (Index Thomisticus), further supplied by on-going syntactic annotation, but nevertheless requiring proper, theoretically well-founded semantic analysis and lexicographic production. We propose linking the rich theoretical tradition which has produced the Prague Dependency Treebank (PDT) and the correlated Tectogrammatical Annotation (TGA) to the objectives of the Bicultural Thomistic Lexicon (LTB). To do this, we will examine and discuss Havrnek's contribution to the shaping of standard language, especially through the distinctive, creative procedure he calls intellectualisation. At the same time, we will outline the effect determined by the use -established in all its extensions -of an expression (lexical item or noun phrase). Given that the user -Thomas Aquinas -was an author who profoundly influenced subsequent thought, the creation or codification of a terminological value of the item will be investigated.
Information presented in graphical form can better perceive the data for analysis and further processing. Large selection of software packages for graphics peculiarities associated with the presentation of graphical information and the need to solve specific tasks. The use of software for image processing in different sectors of human activity raises the task review the relevant analytical software. For the analysis of selected software products best known for working with raster, vector and fractal graphics. An analytical review of software products and graphics imaging rating of set the intensity of their use. The classification software and presented its component structure. Based on the developed classifications created software selection rule. Results of the work make it possible to identify the components of an existing software needs further improvement and development and is the basis for new software. Graphic editor selection formula is reduced to coincidence the relevant user-defined classifications. Prospects for future research is the improvement and development of individual instruments photo editor.
Unconditioned stimulus(US) devaluation has been put forward as an effective measure to decrease fear response. Previous studies mostly focused on investigating the impact of US devaluation in the test phase after fear acquisition. As widely recognized, exposure therapy based on the theory of extinction training is a frequently used method for the treatment of mental disorders. Hence, we made an attempt to implement this program in the extinction training to examine if it could improve the therapeutic effect. In addition, to further understand the mechanisms of US devaluation, evaluative conditioning was explored.An experiment was designed to test the impact of reduction in US intensity on conditioned fear extinction.All participants were subjected to a fear conditioning experiment consisted of acquisition, US devaluation and extinction phases while subjective US-expectancy and skin conductance response(SCR)were rated online. In the experiment, the intensity of US was decreased after acquisition for one group(devaluation) and held constant for another group(control). Two simple geometrical figures served as CS+ and CS?, and a 1-sec female vocal stimulus(i.e., scream) as US. Each CS+ was paired with a US during the acquisition phase, and in the subsequent devaluation phrase, subjects were only exposed to the intensity-changed US for three times. To measure evaluative conditioning, participants were required to rate CS-valence at the end of each conditioning phase.The results show that US-expectancy to CS+ was not significantly different between two groups, which seemed to reflect a similar mode of fear extinction. However, the SCR to CS+ of devaluation groupwas significantly lower than that of control group, suggesting an efficient promoted process of fear extinction by US revaluation. The effect of US devaluation was also found in evaluative conditioning. Compared to control group,devaluation group showed more positive valence ratings for CS+.The results suggest that US devaluation did promote the extinction according to the SCR. Through the US devaluation, the explicit awareness of CS-US bond did not change; but the SCR, representing a procedural fear memory formed by implicit learning, had been influenced, which declared a separation between different fear response indexes. From the evaluating conditioning results, we inferred that US might have served as a transporter, to transform the negative valence to CS. After devaluation, individuals might restate the cognition of CS fear valence, leading to the promotion of extinction. The results suggest that the modulation of US intensity may provide a new perspective for exposure therapy. Based on relevant literature review, the knowledge of the internal mechanisms of US devaluation that influence the conditioned fear extinction has acquired great advance. In the future, the treatment on mental disorders should be more focused on the behavioral therapy based on extinction training.
This paper focuses on improving a specific opinion spam detection task, deceptive spam. In addition to traditional word form and other shallow syntactic features, we introduce two types of deep level linguistic features. The first type of features are derived from a shallow discourse parser trained on Penn Discourse Treebank (PDTB), which can capture inter-sentence information. The second type is based on the relationship between sentiment analysis and spam detection. The experimental results over the benchmark dataset demonstrate that both of the proposed deep features achieve improved performance over the baseline.
The task of recognizing events from video has attracted a lot of attention in recent years. However, due to the complex nature of user-defined events, the use of purely audio- visual content analysis without domain knowledge has been found to be grossly inadequate. In this paper, we propose to construct a semantic-visual knowledge base to encode the rich event-centric concepts and their relationships from the well- established lexical databases, including FrameNet, as well as the concept-specific visual knowledge from ImageNet. Based on this semantic-visual knowledge bases, we design an effective system for video event recognition. Specifically, in order to narrow the semantic gap between the high-level complex events and low-level visual representations, we utilize the event-centric semantic concepts encoded in the knowledge base as the intermediate-level event representation, which offers both human-perceivable and machine-interpretable semantic clues for event recognition. In addition, in order to leverage the abundant ImageNet images, we propose a robust transfer learning model to learn the noise- resistant concept classifiers for videos. Extensive experiments on various real-world video datasets demonstrate the superiority of our proposed system as compared to the state-of-the-art approaches.
We present a new treebank of English and French technical forum content which has been annotated for grammatical errors and phrase structure. This double annotation allows us to empirically measure the effect of errors on parsing performance. While it is slightly easier to parse the corrected versions of the forum sentences, the errors are not the main factor in making this kind of text hard to parse.
Congenital amusia is a neurodevelopmental disorder characterized by impaired pitch processing. Although pitch simultaneities are among the fundamental building blocks of Western tonal music, affective responses to simultaneities such as isolated dyads varying in consonance/dissonance or chords varying in major/minor quality have rarely been studied in amusic individuals. Thirteen amusics and thirteen matched controls enculturated to Western tonal music provided pleasantness ratings of sine-tone dyads and complex-tone dyads in piano timbre as well as perceived happiness/sadness ratings of sine-tone triads and complex-tone triads in piano timbre. Acoustical analyses of roughness and harmonicity were conducted to determine whether similar acoustic information contributed to these evaluations in amusics and controls. Amusic individuals' pleasantness ratings indicated sensitivity to consonance and dissonance for complex-tone (piano timbre) dyads and, to a lesser degree, sine-tone dyads, whereas controls showed sensitivity when listening to both tone types. Furthermore, amusic individuals showed some sensitivity to the happiness-major association in the complex-tone condition, but not in the sine-tone condition. Controls rated major chords as happier than minor chords in both tone types. Linear regression analyses revealed that affective ratings of dyads and triads by amusic individuals were predicted by roughness but not harmonicity, whereas affective ratings by controls were predicted by both roughness and harmonicity. We discuss affective sensitivity in congenital amusia in view of theories of affective responses to isolated chords in Western listeners.
Neuroimaging while participants listen to audiobooks provides a rich data source for theories of incremental parsing. We compare nested regression models of these data. These mixed-effects models incorporate linguistic predictors at various grain sizes ranging from part-of-speech bigrams, through surprisal on context-free treebank grammars, to incremental node counts in trees that are derived by Minimalist Grammars. The fine-grained structures make an independent contribution over and above coarser predictors. However, this result only obtains with time courses from anterior temporal lobe (aTL). In analogous time courses from inferior frontal gyrus, only n-grams improve upon a non-syntactic baseline. These results support the idea that aTL does combinatoric processing during naturalistic story comprehension, processing that bears a systematic relationship to linguistic structure.