Author

Aaron Steven White

@aaronstevenwhite.io

computational semanticist. into modular synths, film photography, and rum. @anglebas.in is my alter ego.

Posts

Publications

Subscriptions

Readers

Computational Semantics++Apr 7, 2026

Machine-readable attitudes

User intent declarations can be viewed as propositional attitudes (permission, prohibition, desire, intention, belief, etc.) over structured descriptions of data use. Treating them that way gives you composable building blocks from existing theory and lens-based translations between community vocabularies that make explicit what each translation cannot carry through.

atprotobluesky

Aaron Steven White on ChiveAug 1, 2025

Cross-Document Event-Keyed Summarization

Event-keyed summarization (EKS) requires summarizing a specific event described in a document given the document text and an event representation extracted from it. In this work, we extend EKS to the cross-document setting (CDEKS), in which summaries must synthesize information from accounts of the same event as given by multiple sources. We introduce SEAMuS (Summaries of Events Across Multiple Sources), a high-quality dataset for CDEKS based on an expert reannotation of the FAMuS dataset for cross-document argument extraction. We present a suite of baselines on SEAMuS—covering both smaller, fine-tuned models, as well as zero- and few-shot prompted LLMs—along with detailed ablations and a human evaluation study, showing SEAMuS to be a valuable benchmark for this new task.

Aaron Steven White on ChiveApr 28, 2025

Dynamic modeling of EEG responses to natural speech reveals earlier processing of predictable words

In recent years, it has become clear that EEG indexes the comprehension of natural, narrative speech. One particularly compelling demonstration of this fact can be seen by regressing EEG responses to speech against measures of how individual words in that speech linguistically relate to their preceding context. This approach produces a so-called temporal response function that displays a centro-parietal negativity reminiscent of the classic N400 component of the event-related potential. One shortcoming of previous implementations of this approach is that they have typically assumed a linear, time-invariant relationship between the linguistic speech features and the EEG responses. In other words, the analysis typically assumes that the response has the same shape and timing for every word – and only varies (linearly) in terms of its amplitude. In the present work, we relax this assumption under the hypothesis that responses to individual words may be processed more rapidly when they are predictable. Specifically, we introduce a framework wherein the standard linear temporal response function can be modulated in terms of its amplitude, latency, and temporal scale based on the predictability of the current and prior words. We use the proposed approach to model EEG recorded from a set of participants who listened to an audiobook narrated by a single talker, and a separate set of participants who attended to one of two concurrently presented audiobooks. We show that expected words are processed faster – evoking lower amplitude N400-like responses with earlier peaks – and that this effect is driven both by the word’s own predictability and the predictability of the immediately preceding word. Additional analysis suggests that this finding is not simply explained based on how quickly words can be disambiguated from their phonetic neighbors. As such, our study demonstrates that the timing and amplitude of brain responses to words in natural speech depend on their predictability. By accounting for these effects, our framework also improves the accuracy with which neural responses to natural speech can be modeled.

Aaron Steven White on ChiveJan 24, 2025

Modeling the prompt in inference judgment tasks

We show that when analyzing data from inference judgment tasks, it can be important to incorporate into one's data analysis regime an explicit representation of the semantics of the natural language prompt used to guide participants on the task. To demonstrate this, we conduct two experiments within an existing experimental paradigm focused on measuring factive inferences, while manipulating the prompt participants receive in small but semantically potent ways. In statistical model comparisons couched within the framework of probabilistic dynamic semantics, we find that probabilistic models structured, in part, by the semantics of the prompt fit better to data collected using that prompt than models that ignore the semantics of the prompt.

Aaron Steven White on ChiveJan 24, 2025

Cause, Make, and Force as Graded Causatives

We investigate the semantics of the causal verbs cause, make, and force as used in the construction X {caused/made/forced} Y (to) Z. The predominant approach to analyzing verbs of causing has been to argue that they convey some version of SUFFICIENCY, but it has also been suggested that INTENTION or possible ALTERNATIVES may also factor into the semantics of the verbs. Using sequences of tic-tac-toe states as experimental stimuli, we measure the three possible contributing factors in each stimuli and ask participants whether each verb is appropriate for describing the sequence. We find experimental support for a differentiating semantics of these verbs, in which no single predictor is the sole factor in when each verb is appropriate.

Aaron Steven White on ChiveNov 1, 2024

Event-Keyed Summarization

We introduce event-keyed summarization (EKS), a novel task that marries traditional summarization and document-level event extraction, with the goal of generating a contextualized summary for a specific event, given a document and an extracted event structure. We introduce a dataset for this task, MUCSUM, consisting of summaries of all events in the classic MUC-4 dataset, along with a set of baselines that comprises both pretrained LM standards in the summarization literature, as well as larger frontier models. We show that ablations that reduce EKS to traditional summarization or structure-to-text yield inferior summaries of target events and that MUCSUM is a robust benchmark for this task. Lastly, we conduct a human evaluation of both reference and model summaries, and provide some detailed analysis of the results.

Aaron Steven White on ChiveJun 1, 2024

FAMuS: Frames Across Multiple Sources

Understanding event descriptions is a central aspect of language processing, but current approaches focus overwhelmingly on single sentences or documents. Aggregating information about an event across documents can offer a much richer understanding. To this end, we present FAMuS, a new corpus of Wikipedia passages that report on some event, paired with underlying, genre-diverse (non-Wikipedia) source articles for the same event. Events and (cross-sentence) arguments in both report and source are annotated against FrameNet, providing broad coverage of different event types. We present results on two key event understanding tasks enabled by FAMuS: source validation—determining whether a document is a valid source for a target report event—and cross-document argument extraction—full-document argument extraction for a target event from both its report and the correct source article.

Aaron Steven White on ChiveMar 1, 2024

MultiMUC: Multilingual Template Filling on MUC-4

We introduce MultiMUC, the first multilingual parallel corpus for template filling, comprising translations of the classic MUC-4 template filling benchmark into five languages: Arabic, Chinese, Farsi, Korean, and Russian. We obtain automatic translations from a strong multilingual machine translation system and manually project the original English annotations into each target language. For all languages, we also provide human translations for key portions of the dev and test splits. Finally, we present baselines on MultiMUC both with state-of-the-art template filling models for MUC-4 and with ChatGPT. We release MUC-4 and the supervised baselines to facilitate further work on document-level information extraction in multilingual settings.

Aaron Steven White on ChiveJan 1, 2024

Probabilistic dynamic semantics

We introduce the framework of Probabilistic Dynamic Semantics (PDS), which we use to seamlessly integrate dynamic semantic analyses of lexical and discourse phenomena in the Montagovian tradition with probabilistic models of linguistic inference datasets. We show how PDS provides a general standpoint from which to understand uncertainty and context sensitivity in discourse and briefly illustrate applications to gradable adjective semantics and the veridicality inferences of clause-embedding predicates.

Aaron Steven White on ChiveJan 1, 2024

Generating event descriptions under syntactic and semantic constraints

With the goal of supporting scalable lexical semantic annotation, analysis, and theorizing, we conduct a comprehensive evaluation of different methods for generating event descriptions under both syntactic constraints — e.g. desired clause structure — and semantic constraints — e.g. desired verb sense. We compare three different methods — (i) manual generation by experts; (ii) sampling from a corpus annotated for syntactic and semantic information; and (iii) sampling from a language model (LM) conditioned on syntactic and semantic information — along three dimensions of the generated event descriptions: (a) naturalness, (b) typicality, and (c) distinctiveness. We find that all methods reliably produce natural, typical, and distinctive event descriptions, but that manual generation continues to produce event descriptions that are more natural, typical, and distinctive than the automated generation methods. We conclude that the automated methods we consider produce event descriptions of sufficient quality for use in downstream annotation and analysis insofar as the methods used for this annotation and analysis are robust to a small amount of degradation in the resulting event descriptions.

Aaron Steven White on ChiveJan 1, 2024

Small Models Are (Still) Effective Cross-Domain Argument Extractors

Effective ontology transfer has been a major goal of recent work on event argument extraction (EAE). Two methods in particular — question answering (QA) and template infilling (TI) — have emerged as promising approaches to this problem. However, detailed explorations of these techniques' ability to actually enable this transfer are lacking. In this work, we provide such a study, exploring zero-shot transfer using both techniques on six major EAE datasets at both the sentence and document levels. Further, we challenge the growing reliance on LLMs for zero-shot extraction, showing that vastly smaller models trained on an appropriate source ontology can yield zero-shot performance superior to that of GPT-3.5 or GPT-4.

Aaron Steven White on ChiveDec 1, 2023

On Event Individuation for Document-Level Information Extraction

As information extraction (IE) systems have grown more adept at processing whole documents, the classic task of template filling has seen renewed interest as a benchmark for document-level IE. In this position paper, we call into question the suitability of template filling for this purpose. We argue that the task demands definitive answers to thorny questions of event individuation — the problem of distinguishing distinct events — about which even human experts disagree. Through an annotation study and error analysis, we show that this raises concerns about the usefulness of template filling metrics, the quality of datasets for the task, and the ability of models to learn it. Finally, we consider possible solutions.