Memorizing All for Implicit Discourse Relation Recognition

Implicit discourse relation recognition is a challenging task due to the absence of the necessary informative clues from explicit connectives. An implicit discourse relation recognizer has to carefully tackle the semantic similarity of sentence pairs and the severe data sparsity issue. In this article, we learn token embeddings to encode the structure of a sentence from a dependency point of view in their representations and use them to initialize a baseline model to make it really strong. Then, we propose a novel memory component to tackle the data sparsity issue by allowing the model to master the entire training set, which helps in achieving further performance improvement. The memory mechanism adequately memorizes information by pairing representations and discourse relations of all training instances, thus filling the slot of the data-hungry issue in the current implicit discourse relation recognizer. The proposed memory component, if attached with any suitable baseline, can help in performance enhancement. The experiments show that our full model with memorizing the entire training data provides excellent results on PDTB and CDTB datasets, outperforming the baselines by a fair margin.

Mehrsprachigkeit in transmodaler Kommunikation

The paper is concerned with codeswitching in transmodal WhatsApp messenger chats. Based on a corpus of text and audio postings from a group of German-Lebanese cousins that is complemented by ethnographic interviews, the study shows that language alternations can be associated with particular metapragmatic or indexical functions in the different modalities. In audio postings, switching between German and Arabic contextualises varying discourse relations. Also, the cousins use Arabic discourse markers (such as ya'ne, ‘it means’) frequently to structure their talk. In contrast, when they switch to Arabic in their text-postings – using Arabizi, a CMC-register in the Arabic-speaking world – this recurrently establishes a playful or ironic frame for ritual teasings. The final section discusses these transmodal and multilingual practices as multi-layered identity positionings vis-à-vis a monolingual society, their multilingual family and networked communities.

Constructing the Corpus of Chinese Textual ‘Run-on’ Sentences (CCTRS)

Chinese is a discourse-oriented language. “Run-on” sentences (liushui ju) are a typical and prevalent form of discourse in Chinese. These sentences show the capacity of the Chinese language for organizing loose structures into an effective and coherent discourse. Despite their widespread use in Chinese, previous studies have only explored “run-on” sentences by using small-scale examples. In order to carry out a quantitative investigation of “run-on” sentences, we need to establish a corpus. The present study selects 500 “run-on” sentences and annotates them on the levels of discourse, syntax and semantics. We mainly adopt PDTB (Penn Discourse Treebank) styles in the discourse annotations but we also borrow some features from RST (rhetorical structure theory). We find that the distribution of the frequency of discourse relations in the data extracted from this corpus follows the power law. The preliminary results reveal that semantic leaps in “run-on” sentences are closely related to the use of the topic chain and the animacy and the span of discourse relations. This corpus can thus aid in carrying out further computational and cognitive studies of Chinese discourse.

A Structure Self-Aware Model for Discourse Parsing on Multi-Party Dialogues

Conversational discourse structures aim to describe how a dialogue is organized, thus they are helpful for dialogue understanding and response generation. This paper focuses on predicting discourse dependency structures for multi-party dialogues. Previous work adopts incremental methods that take the features from the already predicted discourse relations to help generate the next one. Although the inter-correlations among predictions considered, we find that the error propagation is also very serious and hurts the overall performance. To alleviate error propagation, we propose a Structure Self-Aware (SSA) model, which adopts a novel edge-centric Graph Neural Network (GNN) to update the information between each Elementary Discourse Unit (EDU) pair layer by layer, so that expressive representations can be learned without historical predictions. In addition, we take auxiliary training signals (e.g. structure distillation) for better representation learning. Our model achieves the new state-of-the-art performances on two conversational discourse parsing benchmarks, largely outperforming the previous methods.

Dialogue Discourse-Aware Graph Model and Data Augmentation for Meeting Summarization

Meeting summarization is a challenging task due to its dynamic interaction nature among multiple speakers and lack of sufficient training data. Existing methods view the meeting as a linear sequence of utterances while ignoring the diverse relations between each utterance. Besides, the limited labeled data further hinders the ability of data-hungry neural models. In this paper, we try to mitigate the above challenges by introducing dialogue-discourse relations. First, we present a Dialogue Discourse-Dware Meeting Summarizer (DDAMS) to explicitly model the interaction between utterances in a meeting by modeling different discourse relations. The core module is a relational graph encoder, where the utterances and discourse relations are modeled in a graph interaction manner. Moreover, we devise a Dialogue Discourse-Aware Data Augmentation (DDADA) strategy to construct a pseudo-summarization corpus from existing input meetings, which is 20 times larger than the original dataset and can be used to pretrain DDAMS. Experimental results on AMI and ICSI meeting datasets show that our full system can achieve SOTA performance. Our codes and outputs are available at https://github.com/xcfcode/DDAMS/.

Discourse Relations and Connectives in Higher Text Structure

The present article investigates possibilities and limits of local (shallow) analysis of discourse coherence with respect to the phenomena of global coherence and higher composition of texts. We study corpora annotated with local discourse relations in Czech and partly in English to try and find clues in the local annotation indicating a higher discourse structure. First, we classify patterns of subsequent or overlapping pairs of local relations, and hierarchies formed by nested local relations. Special attention is then given to relations crossing paragraph boundaries and their semantic types, and to paragraph-initial discourse connectives. In the third part, we examine situations in which annotators incline to marking a large argument (larger than one sentence) of a discourse relation even with a minimality principle annotation rule in place. Our analyses bring (i) new linguistic insights regarding coherence signals in local and higher contexts, e.g. detection and description of hierarchies of local discourse relations up to 5 levels in Czech and English, description of distribution differences in semantic types in cross-paragraph and other settings, identification of Czech connectives only typical for higher structures, or the detection of prevalence of large left-sided arguments in locally annotated data; (ii) as another type of contribution, some new reflections on methodologies of the approaches under scrutiny.

Lexical and Structural Cues to Discourse Processing in First and Second Language

Discourse connectives are lexical items like “but” and “so” that are well-known to influence the online processing of the discourse relations they convey. Yet, discourse relations like causality or contrast can also be signaled by other means than connectives, such as syntactic structures. So far, the influence of these alternative signals for discourse processing has been comparatively under-researched. In particular, their processing in a second language remains entirely unexplored. In a series of three self-paced reading experiments, we compare the reading patterns of contrastive relations by native French-speakers and non-native speakers of French with English as a first language. We focus on the effect of syntactic parallelism and how it interacts with different types of connectives. We test whether native and non-native readers equally recruit parallelism to process contrast in combination with or without a connective (Experiment 1), with a frequent vs. infrequent connective (Experiment 2) and with an ambiguous vs. unambiguous connective (Experiment 3), thus varying the explicitness and ease of retrieval of the contrast relation. Our results indicate that parallelism plays an important role for both groups of readers, but that it is a more prominent cue for non-native speakers, while its effect is modulated by task difficulty for native participants.