Search

Despite impressive advances in Natural Language Generation (NLG) and Large Language Models (LLMs), researchers are still unclear about important aspects of NLG evaluation. To substantiate this claim, I examine current classifications of hallucination and omission in data-text NLG, and I propose a logic-based synthesis of these classfications. I conclude by highlighting some remaining limitations of all current thinking about hallucination and by discussing implications for LLMs. [ABSTRACT FROM AUTHOR]

Speakers often produce definite referring expressions that areoverspecified: they tend to include more attributes than neces-sary to distinguish the target referent. The current paper inves-tigates how the occurrence of overspecification is affected byviewing time. We conducted an experiment in which speakerswere asked to refer to target objects in visual domains. Half ofthe speakers had unlimited time to inspect the domains, whileviewing time was limited (1000 ms) for the other half. The re-sults reveal that limited viewing time induces the occurrenceof overspecification. We conjecture that limited viewing timecaused speakers to rely heavily on quick heuristics during at-tribute selection, which urge them to select attributes that areperceptually salient. In the case of unlimited inspection time,speakers seem to rely on a combination of heuristic and moredeliberate selection strategies.

The article discusses models of reference, which focuses on contributions from psycholinguistics and computational linguistics. Topics discussed include information stored by interlocutors, concern of overspecification, how visual perception influences reference, and use of machine learning to offer a model of the choice between different types of referring expression.

A substantial amount of recent work in natural language generation has focused on the generation of ''one-shot'' referring expressions whose only aim is to identify a target referent. Dale and Reiter's Incremental Algorithm (IA) is often thought to be the best algorithm for maximizing the similarity to referring expressions produced by people. We test this hypothesis by eliciting referring expressions from human subjects and computing the similarity between the expressions elicited and the ones generated by algorithms. It turns out that the success of the IA depends substantially on the ''preference order'' (PO) employed by the IA, particularly in complex domains. While some POs cause the IA to produce referring expressions that are very similar to expressions produced by human subjects, others cause the IA to perform worse than its main competitors; moreover, it turns out to be difficult to predict the success of a PO on the basis of existing psycholinguistic findings or frequencies in corpora. We also examine the computational complexity of the algorithms in question and argue that there are no compelling reasons for preferring the IA over some of its main competitors on these grounds. We conclude that future research on the generation of referring expressions should explore alternatives to the IA, focusing on algorithms, inspired by the Greedy Algorithm, which do not work with a fixed PO. [ABSTRACT FROM AUTHOR]

This article explores the role of surface ambiguities in referring expressions, and how the risk of such ambiguities should be taken into account by an algorithm that generates referring expressions, if these expressions are to be optimally effective for a hearer. We focus on the ambiguities that arise when adjectives occur in coordinated structures. The central idea is to use statistical information about lexical co-occurrence to estimate which interpretation of a phrase is most likely for human readers, and to avoid generating phrases where misunderstandings are likely. Various aspects of the problem were explored in three experiments in which responses by human participants provided evidence about which reading was most likely for certain phrases, which phrases were deemed most suitable for particular referents, and the speed at which various phrases were read. We found a preference for ''clear'' expressions to ''unclear'' ones, but if several of the expressions are ''clear,'' then brief expressions are preferred over non-brief ones even though the brief ones are syntactically ambiguous and the non-brief ones are not; the notion of clarity was made precise using Kilgarriff's Word Sketches. We outline an implemented algorithm that generates noun phrases conforming to our hypotheses. [ABSTRACT FROM AUTHOR]

Starting from the assumption that NPs of all kinds can be anaphoric on antecedents in the linguistic context, we work towards a general theory of context-dependent NP meaning. Two complicating factors are that the relation between anaphors and antecedents is by no means unrestricted and that often there is a partial match between anaphor and antecedent. We argue that the presuppositions-as-anaphors approach of van der Sandt provides a natural starting point for our enterprise. Unfortunately, this theory has a number of deficiencies for our purposes, in particular where the treatment of partial matches is concerned. We propose a number of modifications of van der Sandt's formal theory and apply the modified algorithm first to definite NPs and later to NPs of all kinds. The resulting modified version of the presuppositions-as-anaphors theory is argued to be more general, formally more precise, and empirically more adequate than its predecessor. [ABSTRACT FROM PUBLISHER]