A controversial test for Large Language Models concerns the ability to discern possible from impossible language. While some evidence attests to the models' sensitivity to what crosses the limits of grammatically impossible language, this evidence has been contested on the grounds of the soundness of the testing material. We use model-internal representations to tap directly into the way Large Language Models represent the 'grammatical-ungrammatical' distinction. In a novel benchmark, we elicit probabilities from 4 models and compute minimal-pair surprisal differences, juxtaposing probabilities assigned to grammatical sentences to probabilities assigned to (i) lower frequency grammatical sentences, (ii) ungrammatical sentences, (iii) semantically odd sentences, and (iv) pragmatically odd sentences. The prediction is that if string-probabilities can function as proxies for the limits of grammar, the ungrammatical condition will stand out among the conditions that involve linguistic violations, showing a spike in the surprisal rates. Our results do not reveal a unique surprisal signature for ungrammatical prompts, as the semantically and pragmatically odd conditions consistently show higher surprisal. We thus demonstrate that probabilities do not constitute reliable proxies for model-internal representations of syntactic knowledge. Consequently, claims about models being able to distinguish possible from impossible language need verification through a different methodology.

Balanced and efficient information flow is essential for optimizing language generation models. In this work, we propose Entropy-UID, a new token selection method that balances entropy and Uniform Information Density (UID) principles for enhanced efficiency of text generation. Our approach adaptively adjusts token selection by jointly minimizing entropy and surprisal, promoting more even information distribution across generated sequences. Theoretical validation demonstrates that Entropy-UID optimally reduces information spikes while maintaining fluency and coherence. The method has been evulated using information-theoretic metrics on multiple benchmark datasets, including WikiText-2, OpenWebText, and WMT. Experimental results show that Entropy-UID achieves lower surprisal and entropy variance compared to standard GPT-2 and alternative heuristics, leading to more balanced and human-like text generation. Our findings point towards the potential of leveraging information-theoretic constraints to refine token selection strategies in autoregressive language models.

The effect of surprisal on processing difficulty has been a central topic of investigation in psycholinguistics. Here, we use eyetracking data to examine three language processing regimes that are common in daily life but have not been addressed with respect to this question: information seeking, repeated processing, and the combination of the two. Using standard regime-agnostic surprisal estimates we find that the prediction of surprisal theory regarding the presence of a linear effect of surprisal on processing times, extends to these regimes. However, when using surprisal estimates from regime-specific contexts that match the contexts and tasks given to humans, we find that in information seeking, such estimates do not improve the predictive power of processing times compared to standard surprisals. Further, regime-specific contexts yield near zero surprisal estimates with no predictive power for processing times in repeated reading. These findings point to misalignments of task and memory representations between humans and current language models, and question the extent to which such models can be used for estimating cognitively relevant quantities. We further discuss theoretical challenges posed by these results.

Recent studies have shown that as Transformer-based language models become larger and are trained on very large amounts of data, the fit of their surprisal estimates to naturalistic human reading times degrades. The current work presents a series of analyses showing that word frequency is a key explanatory factor underlying these two trends. First, residual errors from four language model families on four corpora show that the inverse correlation between model size and fit to reading times is the strongest on the subset of least frequent words, which is driven by excessively accurate predictions of larger model variants. Additionally, training dynamics reveal that during later training steps, all model variants learn to predict rare words and that larger model variants do so more accurately, which explains the detrimental effect of both training data amount and model size on fit to reading times. Finally, a feature attribution analysis demonstrates that larger model variants are able to accurately predict rare words based on both an effectively longer context window size as well as stronger local associations compared to smaller model variants. Taken together, these results indicate that Transformer-based language models' surprisal estimates diverge from human-like expectations due to the superhumanly complex associations they learn for predicting rare words.

Novelty, akin to gene mutation in evolution, opens possibilities for scholarly advancement. Although peer review remains the gold standard for evaluating novelty in scholarly communication and resource allocation, the vast volume of submissions necessitates an automated measure of scholarly novelty. Adopting a perspective that views novelty as the atypical combination of existing knowledge, we introduce an information-theoretic measure of novelty in scholarly publications. This measure quantifies the degree of 'surprise' perceived by a language model that represents the word distribution of scholarly discourse. The proposed measure is accompanied by face and construct validity evidence; the former demonstrates correspondence to scientific common sense, and the latter is endorsed through alignment with novelty evaluations from a select panel of domain experts. Additionally, characterized by its interpretability, fine granularity, and accessibility, this measure addresses gaps prevalent in existing methods. We believe this measure holds great potential to benefit editors, stakeholders, and policymakers, and it provides a reliable lens for examining the relationship between novelty and academic dynamics such as creativity, interdisciplinarity, and scientific advances.

A prominent theory of affective response to music revolves around the concepts of surprisal and expectation. In prior work, this idea has been operationalized in the form of probabilistic models of music which allow for precise computation of song (or note-by-note) probabilities, conditioned on a 'training set' of prior musical or cultural experiences. To date, however, these models have been limited to compute exact probabilities through hand-crafted features or restricted to linear models which are likely not sufficient to represent the complex conditional distributions present in music. In this work, we propose to use modern deep probabilistic generative models in the form of a Diffusion Model to compute an approximate likelihood of a musical input sequence. Unlike prior work, such a generative model parameterized by deep neural networks is able to learn complex non-linear features directly from a training set itself. In doing so, we expect to find that such models are able to more accurately represent the 'surprisal' of music for human listeners. From the literature, it is known that there is an inverted U-shaped relationship between surprisal and the amount human subjects 'like' a given song. In this work we show that pre-trained diffusion models indeed yield musical surprisal values which exhibit a negative quadratic relationship with measured subject 'liking' ratings, and that the quality of this relationship is competitive with state of the art methods such as IDyOM. We therefore present this model a preliminary step in developing modern deep generative models of music expectation and subjective likability.

This paper investigates the use of word surprisal, a measure of the predictability of a word in a given context, as a feature to aid speech synthesis prosody. We explore how word surprisal extracted from large language models (LLMs) correlates with word prominence, a signal-based measure of the salience of a word in a given discourse. We also examine how context length and LLM size affect the results, and how a speech synthesizer conditioned with surprisal values compares with a baseline system. To evaluate these factors, we conducted experiments using a large corpus of English text and LLMs of varying sizes. Our results show that word surprisal and word prominence are moderately correlated, suggesting that they capture related but distinct aspects of language use. We find that length of context and size of the LLM impact the correlations, but not in the direction anticipated, with longer contexts and larger LLMs generally underpredicting prominent words in a nearly linear manner. We demonstrate that, in line with these findings, a speech synthesizer conditioned with surprisal values provides a minimal improvement over the baseline with the results suggesting a limited effect of using surprisal values for eliciting appropriate prominence patterns.

A sharp tension exists about the nature of human language between two opposite parties: those who believe that statistical surface distributions, in particular using measures like surprisal, provide a better understanding of language processing, vs. those who believe that discrete hierarchical structures implementing linguistic information such as syntactic ones are a better tool. In this paper, we show that this dichotomy is a false one. Relying on the fact that statistical measures can be defined on the basis of either structural or non-structural models, we provide empirical evidence that only models of surprisal that reflect syntactic structure are able to account for language regularities.

Bayesian hierarchical models are well-suited to analyzing the often noisy data from electroencephalography experiments in cognitive neuroscience: these models provide an intuitive framework to account for structures and correlations in the data, and they allow a straightforward handling of uncertainty. In a typical neurolinguistic experiment, event-related potentials show only very small effect sizes and frequentist approaches to data analysis fail to establish the significance of some of these effects. Here, we present a Bayesian approach to analyzing event-related potentials using as an example data from an experiment which relates word surprisal and neural response. Our model is able to estimate the effect of word surprisal on most components of the event-related potential and provides a richer description of the data. The Bayesian framework also allows easier comparison between estimates based on surprisal values calculated using different language models.

More predictable words are easier to process - they are read faster and elicit smaller neural signals associated with processing difficulty, most notably, the N400 component of the event-related brain potential. Thus, it has been argued that prediction of upcoming words is a key component of language comprehension, and that studying the amplitude of the N400 is a valuable way to investigate the predictions that we make. In this study, we investigate whether the linguistic predictions of computational language models or humans better reflect the way in which natural language stimuli modulate the amplitude of the N400. One important difference in the linguistic predictions of humans versus computational language models is that while language models base their predictions exclusively on the preceding linguistic context, humans may rely on other factors. We find that the predictions of three top-of-the-line contemporary language models - GPT-3, RoBERTa, and ALBERT - match the N400 more closely than human predictions. This suggests that the predictive processes underlying the N400 may be more sensitive to the surface-level statistics of language than previously thought.