We present a targeted, scaled-up comparison of incremental processing in humans and neural language models by collecting by-word reaction time data for sixteen different syntactic test suites across a range of structural phenomena. Human reaction time data comes from a novel online experimental paradigm called the Interpolated Maze task. We compare human reaction times to by-word probabilities for four contemporary language models, with different architectures and trained on a range of data set sizes. We find that across many phenomena, both humans and language models show increased processing difficulty in ungrammatical sentence regions with human and model ‘accuracy’ scores a la Marvin and Linzen (2018) about equal. However, although language model outputs match humans in direction, we show that models systematically under-predict the difference in magnitude of incremental processing difficulty between grammatical and ungrammatical sentences. Specifically, when models encounter syntactic violations they fail to accurately predict the longer reading times observed in the human data. These results call into question whether contemporary language models are approaching human-like performance for sensitivity to syntactic violations.

Transformer-based language models pre-trained on large amounts of text data have proven remarkably successful in learning generic transferable linguistic representations. Here we study whether structural guidance leads to more human-like systematic linguistic generalization in Transformer language models without resorting to pre-training on very large amounts of data. We explore two general ideas. The “Generative Parsing” idea jointly models the incremental parse and word sequence as part of the same sequence modeling task. The “Structural Scaffold” idea guides the language model’s representation via additional structure loss that separately predicts the incremental constituency parse. We train the proposed models along with a vanilla Transformer language model baseline on a 14 million-token and a 46 million-token subset of the BLLIP dataset, and evaluate models’ syntactic generalization performances on SG Test Suites and sized BLiMP. Experiment results across two benchmarks suggest converging evidence that generative structural supervisions can induce more robust and humanlike linguistic generalization in Transformer language models without the need for data intensive pre-training.

The uniform information density (UID) hypothesis posits a preference among language users for utterances structured such that information is distributed uniformly across a signal. While its implications on language production have been well explored, the hypothesis potentially makes predictions about language comprehension and linguistic acceptability as well. Further, it is unclear how uniformity in a linguistic signal—or lack thereof—should be measured, and over which linguistic unit, e.g., the sentence or language level, this uniformity should hold. Here we investigate these facets of the UID hypothesis using reading time and acceptability data. While our reading time results are generally consistent with previous work, they are also consistent with a weakly super-linear effect of surprisal, which would be compatible with UID’s predictions. For acceptability judgments, we find clearer evidence that non-uniformity in information density is predictive of lower acceptability. We then explore multiple operationalizations of UID, motivated by different interpretations of the original hypothesis, and analyze the scope over which the pressure towards uniformity is exerted. The explanatory power of a subset of the proposed operationalizations suggests that the strongest trend may be a regression towards a mean surprisal across the language, rather than the phrase, sentence, or document—a finding that supports a typical interpretation of UID, namely that it is the byproduct of language users maximizing the use of a (hypothetical) communication channel.

Prior work has shown that structural supervision helps English language models learn generalizations about syntactic phenomena such as subject-verb agreement. However, it remains unclear if such an inductive bias would also improve language models’ ability to learn grammatical dependencies in typologically different languages. Here we investigate this question in Mandarin Chinese, which has a logographic, largely syllable-based writing system; different word order; and sparser morphology than English. We train LSTMs, Recurrent Neural Network Grammars, Transformer language models, and Transformer-parameterized generative parsing models on two Mandarin Chinese datasets of different sizes. We evaluate the models’ ability to learn different aspects of Mandarin grammar that assess syntactic and semantic relationships. We find suggestive evidence that structural supervision helps with representing syntactic state across intervening content and improves performance in low-data settings, suggesting that the benefits of hierarchical inductive biases in acquiring dependency relationships may extend beyond English.

Humans can learn structural properties about a word from minimal experience, and deploy their learned syntactic representations uniformly in different grammatical contexts. We assess the ability of modern neural language models to reproduce this behavior in English and evaluate the effect of structural supervision on learning outcomes. First, we assess few-shot learning capabilities by developing controlled experiments that probe models’ syntactic nominal number and verbal argument structure generalizations for tokens seen as few as two times during training. Second, we assess invariance properties of learned representation: the ability of a model to transfer syntactic generalizations from a base context (e.g., a simple declarative active-voice sentence) to a transformed context (e.g., an interrogative sentence). We test four models trained on the same dataset: an n-gram baseline, an LSTM, and two LSTM-variants trained with explicit structural supervision. We find that in most cases, the neural models are able to induce the proper syntactic generalizations after minimal exposure, often from just two examples during training, and that the two structurally supervised models generalize more accurately than the LSTM model. All neural models are able to leverage information learned in base contexts to drive expectations in transformed contexts, indicating that they have learned some invariance properties of syntax.

Previous studies investigating the syntactic abilities of deep learning models have not targeted the relationship between the strength of the grammatical generalization and the amount of evidence to which the model is exposed during training. We address this issue by deploying a novel word-learning paradigm to test BERT’s few-shot learning capabilities for two aspects of English verbs: alternations and classes of selectional preferences. For the former, we fine-tune BERT on a single frame in a verbal-alternation pair and ask whether the model expects the novel verb to occur in its sister frame. For the latter, we fine-tune BERT on an incomplete selectional network of verbal objects and ask whether it expects unattested but plausible verb/object pairs. We find that BERT makes robust grammatical generalizations after just one or two instances of a novel word in fine-tuning. For the verbal alternation tests, we find that the model displays behavior that is consistent with a transitivity bias: verbs seen few times are expected to take direct objects, but verbs seen with direct objects are not expected to occur intransitively.

In this work, we analyze how human gaze during reading comprehension is conditioned on the given reading comprehension question, and whether this signal can be beneficial for machine reading comprehension. To this end, we collect a new eye-tracking dataset with a large number of participants engaging in a multiple choice reading comprehension task. Our analysis of this data reveals increased fixation times over parts of the text that are most relevant for answering the question. Motivated by this finding, we propose making automated reading comprehension more human-like by mimicking human information-seeking reading behavior during reading comprehension. We demonstrate that this approach leads to performance gains on multiple choice question answering in English for a state-of-the-art reading comprehension model.

Contemporary autoregressive language models (LMs) trained purely on corpus data have been shown to capture numerous features of human incremental processing. However, past work has also suggested dissociations between corpus probabilities and human next-word predictions. Here we evaluate several state-of-the-art language models for their match to human next-word predictions and to reading time behavior from eye movements. We then propose a novel method for distilling the linguistic information implicit in human linguistic predictions into pre-trained LMs: Cloze Distillation. We apply this method to a baseline neural LM and show potential improvement in reading time prediction and generalization to held-out human cloze data.

While state-of-the-art neural network models continue to achieve lower perplexity scores on language modeling benchmarks, it remains unknown whether optimizing for broad-coverage predictive performance leads to human-like syntactic knowledge. Furthermore, existing work has not provided a clear picture about the model properties required to produce proper syntactic generalizations. We present a systematic evaluation of the syntactic knowledge of neural language models, testing 20 combinations of model types and data sizes on a set of 34 English-language syntactic test suites. We find substantial differences in syntactic generalization performance by model architecture, with sequential models underperforming other architectures. Factorially manipulating model architecture and training dataset size (1M-40M words), we find that variability in syntactic generalization performance is substantially greater by architecture than by dataset size for the corpora tested in our experiments. Our results also reveal a dissociation between perplexity and syntactic generalization performance.

We present STARC (Structured Annotations for Reading Comprehension), a new annotation framework for assessing reading comprehension with multiple choice questions. Our framework introduces a principled structure for the answer choices and ties them to textual span annotations. The framework is implemented in OneStopQA, a new high-quality dataset for evaluation and analysis of reading comprehension in English. We use this dataset to demonstrate that STARC can be leveraged for a key new application for the development of SAT-like reading comprehension materials: automatic annotation quality probing via span ablation experiments. We further show that it enables in-depth analyses and comparisons between machine and human reading comprehension behavior, including error distributions and guessing ability. Our experiments also reveal that the standard multiple choice dataset in NLP, RACE, is limited in its ability to measure reading comprehension. 47% of its questions can be guessed by machines without accessing the passage, and 18% are unanimously judged by humans as not having a unique correct answer. OneStopQA provides an alternative test set for reading comprehension which alleviates these shortcomings and has a substantially higher human ceiling performance.

Targeted syntactic evaluations have yielded insights into the generalizations learned by neural network language models. However, this line of research requires an uncommon confluence of skills: both the theoretical knowledge needed to design controlled psycholinguistic experiments, and the technical proficiency needed to train and deploy large-scale language models. We present SyntaxGym, an online platform designed to make targeted evaluations accessible to both experts in NLP and linguistics, reproducible across computing environments, and standardized following the norms of psycholinguistic experimental design. This paper releases two tools of independent value for the computational linguistics community: 1. A website, syntaxgym.org, which centralizes the process of targeted syntactic evaluation and provides easy tools for analysis and visualization; 2. Two command-line tools, ‘syntaxgym‘ and ‘lm-zoo‘, which allow any user to reproduce targeted syntactic evaluations and general language model inference on their own machine.

Word embeddings trained on large-scale historical corpora can illuminate human biases and stereotypes that perpetuate social inequalities. These embeddings are often trained in separate vector space models defined according to different attributes of interest. In this paper, we introduce a single, unified dynamic embedding model that learns attribute-specific word embeddings and apply it to a novel dataset—talk radio shows from around the US—to analyze perceptions about refugees. We validate our model on a benchmark dataset and apply it to two corpora of talk radio shows averaging 117 million words produced over one month across 83 stations and 64 cities. Our findings suggest that dynamic word embeddings are capable of identifying nuanced differences in public discourse about contentious topics, suggesting their usefulness as a tool for better understanding how the public perceives and engages with different issues across time, geography, and other dimensions.

Work using artificial languages as training input has shown that LSTMs are capable of inducing the stack-like data structures required to represent context-free and certain mildly context-sensitive languages — formal language classes which correspond in theory to the hierarchical structures of natural language. Here we present a suite of experiments probing whether neural language models trained on linguistic data induce these stack-like data structures and deploy them while incrementally predicting words. We study two natural language phenomena: center embedding sentences and syntactic island constraints on the filler–gap dependency. In order to properly predict words in these structures, a model must be able to temporarily suppress certain expectations and then recover those expectations later, essentially pushing and popping these expectations on a stack. Our results provide evidence that models can successfully suppress and recover expectations in many cases, but do not fully recover their previous grammatical state.

We investigate the extent to which the behavior of neural network language models reflects incremental representations of syntactic state. To do so, we employ experimental methodologies which were originally developed in the field of psycholinguistics to study syntactic representation in the human mind. We examine neural network model behavior on sets of artificial sentences containing a variety of syntactically complex structures. These sentences not only test whether the networks have a representation of syntactic state, they also reveal the specific lexical cues that networks use to update these states. We test four models: two publicly available LSTM sequence models of English (Jozefowicz et al., 2016; Gulordava et al., 2018) trained on large datasets; an RNN Grammar (Dyer et al., 2016) trained on a small, parsed dataset; and an LSTM trained on the same small corpus as the RNNG. We find evidence for basic syntactic state representations in all models, but only the models trained on large datasets are sensitive to subtle lexical cues signaling changes in syntactic state.

State-of-the-art LSTM language models trained on large corpora learn sequential contingencies in impressive detail, and have been shown to acquire a number of non-local grammatical dependencies with some success. Here we investigate whether supervision with hierarchical structure enhances learning of a range of grammatical dependencies, a question that has previously been addressed only for subject-verb agreement. Using controlled experimental methods from psycholinguistics, we compare the performance of word-based LSTM models versus Recurrent Neural Network Grammars (RNNGs) (Dyer et al. 2016) which represent hierarchical syntactic structure and use neural control to deploy it in left-to-right processing, on two classes of non-local grammatical dependencies in English—Negative Polarity licensing and Filler-Gap Dependencies—tested in a range of configurations. Using the same training data for both models, we find that the RNNG outperforms the LSTM on both types of grammatical dependencies and even learns many of the Island Constraints on the filler-gap dependency. Structural supervision thus provides data efficiency advantages over purely string-based training of neural language models in acquiring human-like generalizations about non-local grammatical dependencies.

The goal of this tutorial is to bring the fields of computational linguistics and computational cognitive science closer: we will introduce different stages of language acquisition and their parallel problems in NLP. As an example, one of the early challenges children face is mapping the meaning of word labels (such as “cat”) to their referents (the furry animal in the living room). Word learning is similar to the word alignment problem in machine translation. We explain the current computational models of language acquisition, their limitations, and how the insights from these models can be incorporated into NLP applications. Moreover, we discuss how we can take advantage of the cognitive science of language in computational linguistics: for example, by designing cognitively-motivated evaluations task or buildings language-learning inductive biases into our models.

What information from an act of sentence understanding is robustly represented in the human brain? We investigate this question by comparing sentence encoding models on a brain decoding task, where the sentence that an experimental participant has seen must be predicted from the fMRI signal evoked by the sentence. We take a pre-trained BERT architecture as a baseline sentence encoding model and fine-tune it on a variety of natural language understanding (NLU) tasks, asking which lead to improvements in brain-decoding performance. We find that none of the sentence encoding tasks tested yield significant increases in brain decoding performance. Through further task ablations and representational analyses, we find that tasks which produce syntax-light representations yield significant improvements in brain decoding performance. Our results constrain the space of NLU models that could best account for human neural representations of language, but also suggest limits on the possibility of decoding fine-grained syntactic information from fMRI human neuroimaging.

Neural language models have achieved state-of-the-art performances on many NLP tasks, and recently have been shown to learn a number of hierarchically-sensitive syntactic dependencies between individual words. However, equally important for language processing is the ability to combine words into phrasal constituents, and use constituent-level features to drive downstream expectations. Here we investigate neural models’ ability to represent constituent-level features, using coordinated noun phrases as a case study. We assess whether different neural language models trained on English and French represent phrase-level number and gender features, and use those features to drive downstream expectations. Our results suggest that models use a linear combination of NP constituent number to drive CoordNP/verb number agreement. This behavior is highly regular and even sensitive to local syntactic context, however it differs crucially from observed human behavior. Models have less success with gender agreement. Models trained on large corpora perform best, and there is no obvious advantage for models trained using explicit syntactic supervision.

We present a novel approach for determining learners’ second language proficiency which utilizes behavioral traces of eye movements during reading. Our approach provides stand-alone eyetracking based English proficiency scores which reflect the extent to which the learner’s gaze patterns in reading are similar to those of native English speakers. We show that our scores correlate strongly with standardized English proficiency tests. We also demonstrate that gaze information can be used to accurately predict the outcomes of such tests. Our approach yields the strongest performance when the test taker is presented with a suite of sentences for which we have eyetracking data from other readers. However, it remains effective even using eyetracking with sentences for which eye movement data have not been previously collected. By deriving proficiency as an automatic byproduct of eye movements during ordinary reading, our approach offers a potentially valuable new tool for second language proficiency assessment. More broadly, our results open the door to future methods for inferring reader characteristics from the behavioral traces of reading.

Speakers often have more than one way to express the same meaning. What general principles govern speaker choice in the face of optionality when near semantically invariant alternation exists? Studies have shown that optional reduction in language is sensitive to contextual predictability, such that more predictable a linguistic unit is, the more likely it is to get reduced. Yet it is unclear whether these cases of speaker choice are driven by audience design versus toward facilitating production. Here we argue that for a different optionality phenomenon, namely classifier choice in Mandarin Chinese, Uniform Information Density and at least one plausible variant of availability-based production make opposite predictions regarding the relationship between the predictability of the upcoming material and speaker choices. In a corpus analysis of Mandarin Chinese, we show that the distribution of speaker choices supports the availability-based production account and not the Uniform Information Density.

RNN language models have achieved state-of-the-art perplexity results and have proven useful in a suite of NLP tasks, but it is as yet unclear what syntactic generalizations they learn. Here we investigate whether state-of-the-art RNN language models represent long-distance filler–gap dependencies and constraints on them. Examining RNN behavior on experimentally controlled sentences designed to expose filler–gap dependencies, we show that RNNs can represent the relationship in multiple syntactic positions and over large spans of text. Furthermore, we show that RNNs learn a subset of the known restrictions on filler–gap dependencies, known as island constraints: RNNs show evidence for wh-islands, adjunct islands, and complex NP islands. These studies demonstrates that state-of-the-art RNN models are able to learn and generalize about empty syntactic positions.

Simple reference games are of central theoretical and empirical importance in the study of situated language use. Although language provides rich, compositional truth-conditional semantics to facilitate reference, speakers and listeners may sometimes lack the overall lexical and cognitive resources to guarantee successful reference through these means alone. However, language also has rich associational structures that can serve as a further resource for achieving successful reference. Here we investigate this use of associational information in a setting where only associational information is available: a simplified version of the popular game Codenames. Using optimal experiment design techniques, we compare a range of models varying in the type of associative information deployed and in level of pragmatic sophistication against human behavior. In this setting we find that listeners’ behavior reflects direct bigram collocational associations more strongly than word-embedding or semantic knowledge graph-based associations and that there is little evidence for pragmatically sophisticated behavior on the part of either speakers or listeners. More generally, we demonstrate the effective use of simple tasks to derive insights into the nature of complex linguistic phenomena.

We use the noisy-channel theory of human sentence comprehension to develop an incremental processing cost model that unifies and extends key features of expectation-based and memory-based models. In this model, which we call noisy-context surprisal, the processing cost of a word is the surprisal of the word given a noisy representation of the preceding context. We show that this model accounts for an outstanding puzzle in sentence comprehension, language-dependent structural forgetting effects (Gibson and Thomas, 1999; Vasishth et al., 2010; Frank et al., 2016), which are previously not well modeled by either expectation-based or memory-based approaches. Additionally, we show that this model derives and generalizes locality effects (Gibson, 1998; Demberg and Keller, 2008), a signature prediction of memory-based models. We give corpus-based evidence for a key assumption in this derivation.

We propose a non-parametric Bayesian model for learning and weighting symbolically-defined constraints to populate a log-linear model. The model jointly infers a vector of binary constraint values for each candidate output and likely definitions for these constraints, combining observations of the output classes with a (potentially infinite) grammar over potential constraint definitions. We present results on a small morphophonological system, English regular plurals, as a test case. The inferred constraints, based on a grammar of articulatory features, perform as well as theoretically-defined constraints on both observed and novel forms of English regular plurals. The learned constraint values and definitions also closely resemble standard constraints defined within phonological theory.

With syntactically annotated corpora becoming increasingly available for a variety of languages and grammatical frameworks, tree query tools have proven invaluable to linguists and computer scientists for both data exploration and corpus-based research. We provide a combined engine for tree query (Tregex) and manipulation (Tsurgeon) that can operate on arbitrary tree data structures with no need for preprocessing. Tregex remedies several expressive and implementational limitations of existing query tools, while Tsurgeon is to our knowledge the most expressive tree manipulation utility available.