While it is commonly accepted that maintaining common ground plays a role in conversational success, little prior research exists connecting conversational grounding to success in task-oriented conversations. We study failures of grounding in the Ubuntu IRC dataset, where participants use text-only communication to resolve technical issues. We find that disruptions in conversational flow often stem from a misalignment in common ground, driven by a divergence in beliefs and assumptions held by participants. These disruptions, which we call conversational friction, significantly correlate with task success. While LLMs can identify overt cases of conversational friction, they struggle with subtler and more context-dependent instances that require pragmatic or domain-specific reasoning.

We present CoGS-NLI, a Natural Language Inference (NLI) evaluation benchmark testing understanding of English phrasal constructions drawn from the Construction Grammar Schematicity (CoGS) corpus. This dataset of 1,500 NLI triples facilitates assessment of constructional understanding in a downstream inference task. We present an evaluation benchmark based on the performance of two language models, where we vary the number and kinds of examples given in the prompt, with and without chain-of-thought prompting. The best-performing model and prompt combination achieves a strong overall accuracy of .94 when provided in-context learning examples with the target phrasal constructions, whereas providing additional general NLI examples hurts performance. This evidences the value of resources explicitly capturing the semantics of phrasal constructions, while our qualitative analysis suggests caveats in assuming this performance indicates a deep understanding of constructional semantics.

We probe large language models’ ability to learn deep form-meaning mappings as defined by construction grammars. We introduce the ConTest-NLI benchmark of 80k sentences covering eight English constructions from highly lexicalized to highly schematic. Our pipeline generates diverse synthetic NLI triples via templating and the application of a model-in-the loop filter. This provides aspects of human validation to ensure challenge and label reliability. Zero-shot tests on leading LLMs reveal a 24% drop in accuracy between naturalistic (88%) and adversarial data (64%), with schematic patterns proving hardest. Fine-tuning on a subset of ConTest-NLI yields up to 9% improvement, yet our results highlight persistent abstraction gaps in current LLMs and offer a scalable framework for evaluating construction informed learning.

In this paper, we apply the lens of Construction Grammar to provide linguistically-grounded evidence for the recently introduced view of LLMs that moves beyond the “stochastic parrot” and “emergent Artificial General Intelligence” extremes. We provide further evidence, this time rooted in linguistic theory, that the capabilities of LLMs are best explained by a process of context-directed extrapolation from their training priors. This mechanism, guided by in-context examples in base models or the prompt in instruction-tuned models, clarifies how LLM performance can exceed stochastic parroting without achieving the scalable, general-purpose reasoning seen in humans. Construction Grammar is uniquely suited to this investigation, as it provides a precise framework for testing the boundary between true generalization and sophisticated pattern-matching on novel linguistic tasks. The ramifications of this framework explaining LLM performance are three-fold: first, there is explanatory power providing insights into seemingly idiosyncratic LLM weaknesses and strengths; second, there are empowering methods for LLM users to improve performance of smaller models in post-training; third, there is a need to shift LLM evaluation paradigms so that LLMs are assessed relative to the prevalence of relevant priors in training data, and Construction Grammar provides a framework to create such evaluation data.

During Human Robot Interactions in disaster relief scenarios, Large Language Models (LLMs) have the potential for substantial physical reasoning to assist in mission objectives. However, these reasoning capabilities are often found only in larger models, which are not currently reasonable to deploy on robotic systems due to size constraints. To meet our problem space requirements, we introduce a dataset and pipeline to create Field Reasoning and Instruction Decoding Agent (FRIDA) models. In our pipeline, domain experts and linguists combine their knowledge to make high-quality, few-shot prompts used to generate synthetic data for fine-tuning. We hand-curate datasets for this few-shot prompting and for evaluation to improve LLM reasoning on both general and disaster-specific objects. We concurrently run an ablation study to understand which kinds of synthetic data most affect performance. We fine-tune several small instruction-tuned models and find that ablated FRIDA models only trained on objects’ physical state and function data outperformed both the FRIDA models trained on all synthetic data and the base models in our evaluation. We demonstrate that the FRIDA pipeline is capable of instilling physical common sense with minimal data.

Misinformation detection remains today a challenging task for both annotators and computer systems. While there are many known markers of misinformation—e.g., logical fallacies, propaganda techniques, and improper use of sources—labeling these markers in practice has been shown to produce low agreement as it requires annotators to make several subjective judgments and rely on their own knowledge, external to the text, which may vary between annotators. In this work, we address these challenges with a collection of linguistically-inspired litmus tests. We annotate a schema of 25 logical fallacies, each of which is defined with rigorous tests applied during annotation. Our annotation methodology results in a comparatively high IAA on this task: Cohen’s kappa in the range .69-.86. We release a corpus of 12 documents from various domains annotated with fallacy labels. Additionally, we experiment with a large language model baseline showing that the largest, most advanced models struggle on this challenging task, achieving an F1-score with our gold standard of .08 when excluding non-fallacious examples, compared to human performance of .59-.73. However, we find that prompting methodologies requiring the model to work through our litmus tests improves performance. Our work contributes a robust fallacy annotation schema and annotated corpus, which advance capabilities in this critical research area.

Frame semantics provides an explanation for how we make use of conceptual frames, which encapsulate background knowledge and associations, to more completely understand the meanings of words within a context. Unfortunately, FrameNet, the only widely available implementation of frame semantics, is limited in both scale and coverage. Therefore, we introduce a novel mechanism for generating task-specific frames using large language models (LLMs), which we call Generative FrameNet. We demonstrate its effectiveness on a task that is highly relevant in the current landscape of LLMs: the interpretable storage and retrieval of factual information. Specifically, Generative Frames enable the extension of Retrieval-Augmented Generation (RAG), providing an interpretable framework for reducing inaccuracies in LLMs. We conduct experiments to demonstrate the effectiveness of this method both in terms of retrieval effectiveness as well as the relevance of the automatically generated frames and frame relations. Expert analysis shows that Generative Frames capture a more suitable level of semantic specificity than the frames from FrameNet. Thus, Generative Frames capture a notion of frame semantics that is closer to Fillmore’s originally intended definition, and offer potential for providing data-driven insights into Frame Semantics theory. Our results also show that this novel mechanism of Frame Semantic-based interpretable retrieval improves RAG for question answering with LLMs—outperforming a GPT-4 based baseline by up to 8 points. We provide open access to our data, including prompts and Generative FrameNet.

For human-robot dialogue in a search-and-rescue scenario, a strong knowledge of the conditions and objects a robot will face is essential for effective interpretation of natural language instructions. In order to utilize the power of large language models without overwhelming the limited storage capacity of a robot, we propose PropBank-Powered Data Creation. PropBank-Powered Data Creation is an expert-in-the-loop data generation pipeline which creates training data for disaster-specific language models. We leverage semantic role labeling and Rich Event Ontology resources to efficiently develop seed sentences for fine-tuning a smaller, targeted model that could operate onboard a robot for disaster relief. We developed 32 sentence templates, which we used to make 2 seed datasets of 175 instructions for earthquake search and rescue and train derailment response. We further leverage our seed datasets as evaluation data to test our baseline fine-tuned models.

We evaluate the ability of large language models (LLMs) to provide PropBank semantic role label annotations across different realizations of the same verbs in transitive, intransitive, and middle voice constructions. In order to assess the meta-linguistic capabilities of LLMs as well as their ability to glean such capabilities through in-context learning, we evaluate the models in a zero-shot setting, in a setting where it is given three examples of another verb used in transitive, intransitive, and middle voice constructions, and finally in a setting where it is given the examples as well as the correct sense and roleset information. We find that zero-shot knowledge of PropBank annotation is almost nonexistent. The largest model evaluated, GPT-4, achieves the best performance in the setting where it is given both examples and the correct roleset in the prompt, demonstrating that larger models can ascertain some meta-linguistic capabilities through in-context learning. However, even in this setting, which is simpler than the task of a human in PropBank annotation, the model achieves only 48% accuracy in marking numbered arguments correctly. To ensure transparency and reproducibility, we publicly release our dataset and model responses.

Large Language Models (LLMs) have been developed without a theoretical framework, yet we posit that evaluating and improving LLMs will benefit from the development of theoretical frameworks that enable comparison of the structures of human language and the model of language built up by LLMs through the processing of text. In service of this goal, we develop the Construction Grammar Schematicity (“CoGS”) corpus of 10 distinct English constructions, where the constructions vary with respect to schematicity, or in other words the level to which constructional slots require specific, fixed lexical items, or can be filled with a variety of elements that fulfill a particular semantic role of the slot. Our corpus constructions are carefully curated to range from substantive, frozen constructions (e.g., Let-alone) to entirely schematic constructions (e.g., Resultative). The corpus was collected to allow us to probe LLMs for constructional information at varying levels of abstraction. We present our own probing experiments using this corpus, which clearly demonstrate that even the largest LLMs are limited to more substantive constructions and do not exhibit recognition of the similarity of purely schematic constructions. We publicly release our dataset, prompts, and associated model responses.

We introduce the Situated Corpus Of Understanding Transactions (SCOUT), a multi-modal collection of human-robot dialogue in the task domain of collaborative exploration. The corpus was constructed from multiple Wizard-of-Oz experiments where human participants gave verbal instructions to a remotely-located robot to move and gather information about its surroundings. SCOUT contains 89,056 utterances and 310,095 words from 278 dialogues averaging 320 utterances per dialogue. The dialogues are aligned with the multi-modal data streams available during the experiments: 5,785 images and 30 maps. The corpus has been annotated with Abstract Meaning Representation and Dialogue-AMR to identify the speaker’s intent and meaning within an utterance, and with Transactional Units and Relations to track relationships between utterances to reveal patterns of the Dialogue Structure. We describe how the corpus and its annotations have been used to develop autonomous human-robot systems and enable research in open questions of how humans speak to robots. We release this corpus to accelerate progress in autonomous, situated, human-robot dialogue, especially in the context of navigation tasks where details about the environment need to be discovered.

To collaborate effectively in physically situated tasks, robots must be able to ground concepts in natural language to the physical objects in the environment as well as their own capabilities. We describe the implementation and the demonstration of a system architecture that sup- ports tasking robots using natural language. In this architecture, natural language instructions are first handled by a dialogue management component, which provides feedback to the user and passes executable instructions along to an Abstract Meaning Representation (AMR) parser. The parse distills the action primitives and parameters of the instructed behavior in the form of a directed a-cyclic graph, passed on to the grounding component. We find AMR to be an efficient formalism for grounding the nodes of the graph using a Distributed Correspondence Graph. Thus, in our approach, the concepts of language are grounded to entities in the robot’s world model, which is populated by its sensors, thereby enabling grounded natural language communication. The demonstration of this system will allow users to issue navigation commands in natural language to direct a simulated ground robot (running the Robot Operating System) to various landmarks observed by the user within a simulated environment.

NLP systems have shown impressive performance at answering questions by retrieving relevant context. However, with the increasingly large models, it is impossible and often undesirable to constrain models’ knowledge or reasoning to only the retrieved context. This leads to a mismatch between the information that the models access to derive the answer and the information that is available to the user to assess the model predicted answer. In this work, we study how users interact with QA systems in the absence of sufficient information to assess their predictions. Further, we ask whether adding the requisite background helps mitigate users’ over-reliance on predictions. Our study reveals that users rely on model predictions even in the absence of sufficient information needed to assess the model’s correctness. Providing the relevant background, however, helps users better catch model errors, reducing over-reliance on incorrect predictions. On the flip side, background information also increases users’ confidence in their accurate as well as inaccurate judgments. Our work highlights that supporting users’ verification of QA predictions is an important, yet challenging, problem.

Language models have shown great promise in common-sense related tasks. However, it remains unseen how they would perform in the context of physically situated human-robot interactions, particularly in disaster-relief sce- narios. In this paper, we develop a language model evaluation dataset with more than 800 cloze sentences, written to probe for the func- tion of over 200 objects. The sentences are divided into two tasks: an “easy” task where the language model has to choose between vo- cabulary with different functions (Task 1), and a “challenge” where it has to choose between vocabulary with the same function, yet only one vocabulary item is appropriate given real world constraints on functionality (Task 2). Dis- tilBERT performs with about 80% accuracy for both tasks. To investigate how annotator variability affected those results, we developed a follow-on experiment where we compared our original results with wrong answers chosen based on embedding vector distances. Those results showed increased precision across docu- ments but a 15% decrease in accuracy. We con- clude that language models do have a strong knowledge basis for object reasoning, but will require creative fine-tuning strategies in order to be successfully deployed.

We evaluate an annotation schema for labeling logical fallacy types, originally developed for a crowd-sourcing annotation paradigm, now using an annotation paradigm of two trained linguist annotators. We apply the schema to a variety of different genres of text relating to the COVID-19 pandemic. Our linguist (as opposed to crowd-sourced) annotation of logical fallacies allows us to evaluate whether the annotation schema category labels are sufficiently clear and non-overlapping for both manual and, later, system assignment. We report inter-annotator agreement results over two annotation phases as well as a preliminary assessment of the corpus for training and testing a machine learning algorithm (Pattern-Exploiting Training) for fallacy detection and recognition. The agreement results and system performance underscore the challenging nature of this annotation task and suggest that the annotation schema and paradigm must be iteratively evaluated and refined in order to arrive at a set of annotation labels that can be reproduced by human annotators and, in turn, provide reliable training data for automatic detection and recognition systems.

We adopt, evaluate, and improve upon a two-step natural language understanding (NLU) pipeline that incrementally tames the variation of unconstrained natural language input and maps to executable robot behaviors. The pipeline first leverages Abstract Meaning Representation (AMR) parsing to capture the propositional content of the utterance, and second converts this into “Dialogue-AMR,” which augments standard AMR with information on tense, aspect, and speech acts. Several alternative approaches and training datasets are evaluated for both steps and corresponding components of the pipeline, some of which outperform the original. We extend the Dialogue-AMR annotation schema to cover a different collaborative instruction domain and evaluate on both domains. With very little training data, we achieve promising performance in the new domain, demonstrating the scalability of this approach.

Generative language models trained on large, diverse corpora can answer questions about a passage by generating the most likely continuation of the passage followed by a question/answer pair. However, accuracy rates vary depending on the type of question asked. In this paper we keep the passage fixed, and test with a wide variety of question types, exploring the strengths and weaknesses of the GPT-3 language model. We provide the passage and test questions as a challenge set for other language models.

This paper examines how Abstract Meaning Representation (AMR) can be utilized for finding answers to research questions in medical scientific documents, in particular, to advance the study of UV (ultraviolet) inactivation of the novel coronavirus that causes the disease COVID-19. We describe the development of a proof-of-concept prototype tool, InfoForager, which uses AMR to conduct a semantic search, targeting the meaning of the user question, and matching this to sentences in medical documents that may contain information to answer that question. This work was conducted as a sprint over a period of six weeks, and reveals both promising results and challenges in reducing the user search time relating to COVID-19 research, and in general, domain adaption of AMR for this task.

This paper describes a schema that enriches Abstract Meaning Representation (AMR) in order to provide a semantic representation for facilitating Natural Language Understanding (NLU) in dialogue systems. AMR offers a valuable level of abstraction of the propositional content of an utterance; however, it does not capture the illocutionary force or speaker’s intended contribution in the broader dialogue context (e.g., make a request or ask a question), nor does it capture tense or aspect. We explore dialogue in the domain of human-robot interaction, where a conversational robot is engaged in search and navigation tasks with a human partner. To address the limitations of standard AMR, we develop an inventory of speech acts suitable for our domain, and present “Dialogue-AMR”, an enhanced AMR that represents not only the content of an utterance, but the illocutionary force behind it, as well as tense and aspect. To showcase the coverage of the schema, we use both manual and automatic methods to construct the “DialAMR” corpus—a corpus of human-robot dialogue annotated with standard AMR and our enriched Dialogue-AMR schema. Our automated methods can be used to incorporate AMR into a larger NLU pipeline supporting human-robot dialogue.

We detail refinements made to Abstract Meaning Representation (AMR) that make the representation more suitable for supporting a situated dialogue system, where a human remotely controls a robot for purposes of search and rescue and reconnaissance. We propose 36 augmented AMRs that capture speech acts, tense and aspect, and spatial information. This linguistic information is vital for representing important distinctions, for example whether the robot has moved, is moving, or will move. We evaluate two existing AMR parsers for their performance on dialogue data. We also outline a model for graph-to-graph conversion, in which output from AMR parsers is converted into our refined AMRs. The design scheme presented here, though task-specific, is extendable for broad coverage of speech acts using AMR in future task-independent work.

Commonsense, real-world knowledge about the events that entities or “things in the world” are typically involved in, as well as part-whole relationships, is valuable for allowing computational systems to draw everyday inferences about the world. Here, we focus on automatically extracting information about (1) the events that typically bring about certain entities (origins), (2) the events that are the typical functions of entities, and (3) part-whole relationships in entities. These correspond to the agentive, telic and constitutive qualia central to the Generative Lexicon. We describe our motivations and methods for extracting these qualia relations from the Suggested Upper Merged Ontology (SUMO) and show that human annotators overwhelmingly find the information extracted to be reasonable. Because ontologies provide a way of structuring this information and making it accessible to agents and computational systems generally, efforts are underway to incorporate the extracted information to an ontology hub of Natural Language Processing semantic role labeling resources, the Rich Event Ontology.

We propose a method to improve human activity recognition in video by leveraging semantic information about the target activities from an expert-defined linguistic resource, VerbNet. Our hypothesis is that activities that share similar event semantics, as defined by the semantic predicates of VerbNet, will be more likely to share some visual components. We use a deep convolutional neural network approach as a baseline and incorporate linguistic information from VerbNet through multi-task learning. We present results of experiments showing the added information has negligible impact on recognition performance. We discuss how this may be because the lexical semantic information defined by VerbNet is generally not visually salient given the video processing approach used here, and how we may handle this in future approaches.

In this paper, we explore the challenges of building a computational lexicon for Moroccan Darija (MD), an Arabic dialect spoken by over 32 million people worldwide but which only recently has begun appearing frequently in written form in social media. We raise the question of what belongs in such a lexicon and start by describing our work building traditional word-level lexicon entries with their English translations. We then discuss challenges in translating idiomatic MD text that led to creating multi-word expression lexicon entries whose meanings could not be fully derived from the individual words. Finally, we provide a preliminary exploration of constructions to be considered for inclusion in an MD constructicon by translating examples of English constructions and examining their MD counterparts.

This paper describes the construction and annotation of a corpus of verbal MWEs for English, as part of the PARSEME Shared Task 1.1 on automatic identification of verbal MWEs. The criteria for corpus selection, the categories of MWEs used, and the training process are discussed, along with the particular issues that led to revisions in edition 1.1 of the annotation guidelines. Finally, an overview of the characteristics of the final annotated corpus is presented, as well as some discussion on inter-annotator agreement.

This paper identifies stylistic differences in instruction-giving observed in a corpus of human-robot dialogue. Differences in verbosity and structure (i.e., single-intent vs. multi-intent instructions) arose naturally without restrictions or prior guidance on how users should speak with the robot. Different styles were found to produce different rates of miscommunication, and correlations were found between style differences and individual user variation, trust, and interaction experience with the robot. Understanding potential consequences and factors that influence style can inform design of dialogue systems that are robust to natural variation from human users.

In this paper we describe a new lexical semantic resource, The Rich Event On-tology, which provides an independent conceptual backbone to unify existing semantic role labeling (SRL) schemas and augment them with event-to-event causal and temporal relations. By unifying the FrameNet, VerbNet, Automatic Content Extraction, and Rich Entities, Relations and Events resources, the ontology serves as a shared hub for the disparate annotation schemas and therefore enables the combination of SRL training data into a larger, more diverse corpus. By adding temporal and causal relational information not found in any of the independent resources, the ontology facilitates reasoning on and across documents, revealing relationships between events that come together in temporal and causal chains to build more complex scenarios. We envision the open resource serving as a valuable tool for both moving from the ontology to text to query for event types and scenarios of interest, and for moving from text to the ontology to access interpretations of events using the combined semantic information housed there.

Robot-directed communication is variable, and may change based on human perception of robot capabilities. To collect training data for a dialogue system and to investigate possible communication changes over time, we developed a Wizard-of-Oz study that (a) simulates a robot’s limited understanding, and (b) collects dialogues where human participants build a progressively better mental model of the robot’s understanding. With ten participants, we collected ten hours of human-robot dialogue. We analyzed the structure of instructions that participants gave to a remote robot before it responded. Our findings show a general initial preference for including metric information (e.g., move forward 3 feet) over landmarks (e.g., move to the desk) in motion commands, but this decreased over time, suggesting changes in perception.

Recent efforts have focused on expanding the annotation coverage of PropBank from verb relations to adjective and noun relations, as well as light verb constructions (e.g., make an offer, take a bath). While each new relation type has presented unique annotation challenges, ensuring consistent and comprehensive annotation of light verb constructions has proved particularly challenging, given that light verb constructions are semi-productive, difficult to define, and there are often borderline cases. This research describes the iterative process of developing PropBank annotation guidelines for light verb constructions, the current guidelines, and a comparison to related resources.

This research focuses on expanding PropBank, a corpus annotated with predicate argument structures, with new predicate types; namely, noun, adjective and complex predicates, such as Light Verb Constructions. This effort is in part inspired by a sister project to PropBank, the Abstract Meaning Representation project, which also attempts to capture “who is doing what to whom” in a sentence, but does so in a way that abstracts away from syntactic structures. For example, alternate realizations of a ‘destroying’ event in the form of either the verb ‘destroy’ or the noun ‘destruction’ would receive the same Abstract Meaning Representation. In order for PropBank to reach the same level of coverage and continue to serve as the bedrock for Abstract Meaning Representation, predicate types other than verbs, which have previously gone without annotation, must be annotated. This research describes the challenges therein, including the development of new annotation practices that walk the line between abstracting away from language-particular syntactic facts to explore deeper semantics, and maintaining the connection between semantics and syntactic structures that has proven to be very valuable for PropBank as a corpus of training data for Natural Language Processing applications.

This paper gives guidelines of how to create and update Propbank frameset files using a dedicated editor, Cornerstone. Propbank is a corpus in which the arguments of each verb predicate are annotated with their semantic roles in relation to the predicate. Propbank annotation also requires the choice of a sense ID for each predicate. Thus, for each predicate in Propbank, there exists a corresponding frameset file showing the expected predicate argument structure of each sense related to the predicate. Since most Propbank annotations are based on the predicate argument structure defined in the frameset files, it is important to keep the files consistent, simple to read as well as easy to update. The frameset files are written in XML, which can be difficult to edit when using a simple text editor. Therefore, it is helpful to develop a user-friendly editor such as Cornerstone, specifically customized to create and edit frameset files. Cornerstone runs platform independently, is light enough to run as an X11 application and supports multiple languages such as Arabic, Chinese, English, Hindi and Korean.

This paper gives guidelines of how to annotate Propbank instances using a dedicated editor, Jubilee. Propbank is a corpus in which the arguments of each verb predicate are annotated with their semantic roles in relation to the predicate. Propbank annotation also requires the choice of a sense ID for each predicate. Jubilee facilitates this annotation process by displaying several resources of syntactic and semantic information simultaneously: the syntactic structure of a sentence is displayed in the main frame, the available senses with their corresponding argument structures are displayed in another frame, all available Propbank arguments are displayed for the annotators choice, and example annotations of each sense of the predicate are available to the annotator for viewing. Easy access to each of these resources allows the annotator to quickly absorb and apply the necessary syntactic and semantic information pertinent to each predicate for consistent and efficient annotation. Jubilee has been successfully adapted to many Propbank projects in several universities. The tool runs platform independently, is light enough to run as an X11 application and supports multiple languages such as Arabic, Chinese, English, Hindi and Korean.