Sentences containing multiple semantic operators with overlapping scope often create ambiguities in interpretation, known as scope ambiguities. These ambiguities offer rich insights into the interaction between semantic structure and world knowledge in language processing. Despite this, there has been little research into how modern large language models treat them. In this paper, we investigate how different versions of certain autoregressive language models—GPT-2, GPT-3/3.5, Llama 2, and GPT-4—treat scope ambiguous sentences, and compare this with human judgments. We introduce novel datasets that contain a joint total of almost 1,000 unique scope-ambiguous sentences, containing interactions between a range of semantic operators, and annotated for human judgments. Using these datasets, we find evidence that several models (i) are sensitive to the meaning ambiguity in these sentences, in a way that patterns well with human judgments, and (ii) can successfully identify human-preferred readings at a high level of accuracy (over 90% in some cases).1

Many utterances convey meanings that go beyond the literal meaning of a sentence. One class of such meanings is scalar implicatures, a phenomenon by which a speaker conveys the negation of a more informative utterance by producing a less informative utterance. This paper introduces a Natural Language Inference (NLI) dataset designed to investigate the ability of language models to interpret utterances with scalar implicatures. Our dataset is comprised of text extracted from the C4 English text corpus and annotated with both crowd-sourced and expert annotations. We evaluate NLI models based on DeBERTa to investigate 1) whether NLI models can learn to predict pragmatic inferences involving gradable adjectives and 2) whether models generalize to utterances involving unseen adjectives. We find that fine-tuning NLI models on our dataset significantly improves their performance to derive scalar implicatures, both for in-domain and for out-of domain examples. At the same time, we find that the investigated models still perform considerably worse on examples with scalar implicatures than on other types of NLI examples, highlighting that pragmatic inferences still pose challenges for current models.

Automatic generation of code from natural language descriptions has emerged as one of the main use cases of large language models (LLMs). This has also led to a proliferation of datasets to track progress in the reliability of code generation models, including domains such as programming challenges and common data science tasks. However, existing datasets primarily target the use of code generation models to aid expert programmers in writing code. In this work, we consider a domain of code generation which is more frequently used by users without sophisticated programming skills: translating English descriptions to spreadsheet formulas that can be used to do everyday data processing tasks. We extract naturalistic instructions from StackOverflow posts and manually verify and standardize the corresponding spreadsheet formulas. We use this dataset to evaluate an off-the-shelf code generation model (GPT 3.5 text-davinci-003) as well as recently proposed pragmatic code generation procedures and find that Code Reviewer reranking (Zhang et al., 2022) performs best among the evaluated methods but still frequently generates formulas that differ from human-generated ones.

Keeping track of how states of entities change as a text or dialog unfolds is a key prerequisite to discourse understanding. Yet, there have been few systematic investigations into the ability of large language models (LLMs) to track discourse entities. In this work, we present a task probing to what extent a language model can infer the final state of an entity given an English description of the initial state and a series of state-changing operations. We use this task to first investigate whether Flan-T5, GPT-3 and GPT-3.5 can track the state of entities, and find that only GPT-3.5 models, which have been pretrained on large amounts of code, exhibit this ability. We then investigate whether smaller models pretrained primarily on text can learn to track entities, through finetuning T5 on several training/evaluation splits. While performance degrades for more complex splits, we find that even when evaluated on a different set of entities from training or longer operation sequences, a finetuned model can perform non-trivial entity tracking. Taken together, these results suggest that language models can learn to track entities but pretraining on text corpora alone does not make this capacity surface.

Scalar inferences (SI) are a signature example of how humans interpret language based on unspoken alternatives. While empirical studies have demonstrated that human SI rates are highly variable—both within instances of a single scale, and across different scales—there have been few proposals that quantitatively explain both cross- and within-scale variation. Furthermore, while it is generally assumed that SIs arise through reasoning about unspoken alternatives, it remains debated whether humans reason about alternatives as linguistic forms, or at the level of concepts. Here, we test a shared mechanism explaining SI rates within and across scales: context-driven expectations about the unspoken alternatives. Using neural language models to approximate human predictive distributions, we find that SI rates are captured by the expectedness of the strong scalemate as an alternative. Crucially, however, expectedness robustly predicts cross-scale variation only under a meaning-based view of alternatives. Our results suggest that pragmatic inferences arise from context-driven expectations over alternatives, and these expectations operate at the level of concepts.1

Scalar implicature (SI) arises when a speaker uses an expression (e.g., “some”) that is semantically compatible with a logically stronger alternative on the same scale (e.g., “all”), leading the listener to infer that they did not intend to convey the stronger meaning. Prior work has demonstrated that SI rates are highly variable across scales, raising the question of what factors determine the SI strength for a particular scale. Here, we test the hypothesis that SI rates depend on the listener’s confidence in the underlying scale, which we operationalize as uncertainty over the distribution of possible alternatives conditioned on the context. We use a T5 model fine-tuned on a text infilling task to estimate this distribution. We find that scale uncertainty predicts human SI rates, measured as entropy over the sampled alternatives and over latent classes among alternatives in sentence embedding space. Furthermore, we do not find a significant effect of the surprisal of the strong scalemate. Our results suggest that pragmatic inferences depend on listeners’ context-driven uncertainty over alternatives.

Understanding longer narratives or participating in conversations requires tracking of discourse entities that have been mentioned. Indefinite noun phrases (NPs), such as ‘a dog’, frequently introduce discourse entities but this behavior is modulated by sentential operators such as negation. For example, ‘a dog’ in ‘Arthur doesn’t own a dog’ does not introduce a discourse entity due to the presence of negation. In this work, we adapt the psycholinguistic assessment of language models paradigm to higher-level linguistic phenomena and introduce an English evaluation suite that targets the knowledge of the interactions between sentential operators and indefinite NPs. We use this evaluation suite for a fine-grained investigation of the entity tracking abilities of the Transformer-based models GPT-2 and GPT-3. We find that while the models are to a certain extent sensitive to the interactions we investigate, they are all challenged by the presence of multiple NPs and their behavior is not systematic, which suggests that even models at the scale of GPT-3 do not fully acquire basic entity tracking abilities.

Relations between words are governed by hierarchical structure rather than linear ordering. Sequence-to-sequence (seq2seq) models, despite their success in downstream NLP applications, often fail to generalize in a hierarchy-sensitive manner when performing syntactic transformations—for example, transforming declarative sentences into questions. However, syntactic evaluations of seq2seq models have only observed models that were not pre-trained on natural language data before being trained to perform syntactic transformations, in spite of the fact that pre-training has been found to induce hierarchical linguistic generalizations in language models; in other words, the syntactic capabilities of seq2seq models may have been greatly understated. We address this gap using the pre-trained seq2seq models T5 and BART, as well as their multilingual variants mT5 and mBART. We evaluate whether they generalize hierarchically on two transformations in two languages: question formation and passivization in English and German. We find that pre-trained seq2seq models generalize hierarchically when performing syntactic transformations, whereas models trained from scratch on syntactic transformations do not. This result presents evidence for the learnability of hierarchical syntactic information from non-annotated natural language text while also demonstrating that seq2seq models are capable of syntactic generalization, though only after exposure to much more language data than human learners receive.

Understanding language requires grasping not only the overtly stated content, but also making inferences about things that were left unsaid. These inferences include presuppositions, a phenomenon by which a listener learns about new information through reasoning about what a speaker takes as given. Presuppositions require complex understanding of the lexical and syntactic properties that trigger them as well as the broader conversational context. In this work, we introduce the Naturally-Occurring Presuppositions in English (NOPE) Corpus to investigate the context-sensitivity of 10 different types of presupposition triggers and to evaluate machine learning models’ ability to predict human inferences. We find that most of the triggers we investigate exhibit moderate variability. We further find that transformer-based models draw correct inferences in simple cases involving presuppositions, but they fail to capture the minority of exceptional cases in which human judgments reveal complex interactions between context and triggers.

Universal Dependencies is an open community effort to create cross-linguistically consistent treebank annotation for many languages within a dependency-based lexicalist framework. The annotation consists in a linguistically motivated word segmentation; a morphological layer comprising lemmas, universal part-of-speech tags, and standardized morphological features; and a syntactic layer focusing on syntactic relations between predicates, arguments and modifiers. In this paper, we describe version 2 of the universal guidelines (UD v2), discuss the major changes from UD v1 to UD v2, and give an overview of the currently available treebanks for 90 languages.

Pragmatic inferences often subtly depend on the presence or absence of linguistic features. For example, the presence of a partitive construction (of the) increases the strength of a so-called scalar inference: listeners perceive the inference that Chris did not eat all of the cookies to be stronger after hearing “Chris ate some of the cookies” than after hearing the same utterance without a partitive, “Chris ate some cookies”. In this work, we explore to what extent neural network sentence encoders can learn to predict the strength of scalar inferences. We first show that an LSTM-based sentence encoder trained on an English dataset of human inference strength ratings is able to predict ratings with high accuracy (r = 0.78). We then probe the model’s behavior using manually constructed minimal sentence pairs and corpus data. We first that the model inferred previously established associations between linguistic features and inference strength, suggesting that the model learns to use linguistic features to predict pragmatic inferences.

One of the first steps in the utterance interpretation pipeline of many task-oriented conversational AI systems is to identify user intents and the corresponding slots. Since data collection for machine learning models for this task is time-consuming, it is desirable to make use of existing data in a high-resource language to train models in low-resource languages. However, development of such models has largely been hindered by the lack of multilingual training data. In this paper, we present a new data set of 57k annotated utterances in English (43k), Spanish (8.6k) and Thai (5k) across the domains weather, alarm, and reminder. We use this data set to evaluate three different cross-lingual transfer methods: (1) translating the training data, (2) using cross-lingual pre-trained embeddings, and (3) a novel method of using a multilingual machine translation encoder as contextual word representations. We find that given several hundred training examples in the the target language, the latter two methods outperform translating the training data. Further, in very low-resource settings, multilingual contextual word representations give better results than using cross-lingual static embeddings. We also compare the cross-lingual methods to using monolingual resources in the form of contextual ELMo representations and find that given just small amounts of target language data, this method outperforms all cross-lingual methods, which highlights the need for more sophisticated cross-lingual methods.

Sentences with gapping, such as Paul likes coffee and Mary tea, lack an overt predicate to indicate the relation between two or more arguments. Surface syntax representations of such sentences are often produced poorly by parsers, and even if correct, not well suited to downstream natural language understanding tasks such as relation extraction that are typically designed to extract information from sentences with canonical clause structure. In this paper, we present two methods for parsing to a Universal Dependencies graph representation that explicitly encodes the elided material with additional nodes and edges. We find that both methods can reconstruct elided material from dependency trees with high accuracy when the parser correctly predicts the existence of a gap. We further demonstrate that one of our methods can be applied to other languages based on a case study on Swedish.

Clickbait has become a nuisance on social media. To address the urging task of clickbait detection, we constructed a new corpus of 38,517 annotated Twitter tweets, the Webis Clickbait Corpus 2017. To avoid biases in terms of publisher and topic, tweets were sampled from the top 27 most retweeted news publishers, covering a period of 150 days. Each tweet has been annotated on 4-point scale by five annotators recruited at Amazon’s Mechanical Turk. The corpus has been employed to evaluate 12 clickbait detectors submitted to the Clickbait Challenge 2017. Download: https://webis.de/data/webis-clickbait-17.html Challenge: https://clickbait-challenge.org

We evaluate two cross-lingual techniques for adding enhanced dependencies to existing treebanks in Universal Dependencies. We apply a rule-based system developed for English and a data-driven system trained on Finnish to Swedish and Italian. We find that both systems are accurate enough to bootstrap enhanced dependencies in existing UD treebanks. In the case of Italian, results are even on par with those of a prototype language-specific system.

The Conference on Computational Natural Language Learning (CoNLL) features a shared task, in which participants train and test their learning systems on the same data sets. In 2017, the task was devoted to learning dependency parsers for a large number of languages, in a real-world setting without any gold-standard annotation on input. All test sets followed a unified annotation scheme, namely that of Universal Dependencies. In this paper, we define the task and evaluation methodology, describe how the data sets were prepared, report and analyze the main results, and provide a brief categorization of the different approaches of the participating systems.

Many shallow natural language understanding tasks use dependency trees to extract relations between content words. However, strict surface-structure dependency trees tend to follow the linguistic structure of sentences too closely and frequently fail to provide direct relations between content words. To mitigate this problem, the original Stanford Dependencies representation also defines two dependency graph representations which contain additional and augmented relations that explicitly capture otherwise implicit relations between content words. In this paper, we revisit and extend these dependency graph representations in light of the recent Universal Dependencies (UD) initiative and provide a detailed account of an enhanced and an enhanced++ English UD representation. We further present a converter from constituency to basic, i.e., strict surface structure, UD trees, and a converter from basic UD trees to enhanced and enhanced++ English UD graphs. We release both converters as part of Stanford CoreNLP and the Stanford Parser.