To Augment or Not to Augment? A Comparative Study on Text Augmentation Techniques for Low-Resource NLP

Data-hungry deep neural networks have established themselves as the defacto standard for many NLP tasks including the traditional sequence tagging ones. Despite their state-of-the-art performance on high-resource languages, they still fall behind of their statistical counter-parts in low-resource scenarios. One methodology to counter attack this problem is text augmentation, i.e., generating new synthetic training data points from existing data. Although NLP has recently witnessed a load of textual augmentation techniques, the field still lacks a systematic performance analysis on a diverse set of languages and sequence tagging tasks. To fill this gap, we investigate three categories of text augmentation methodologies which perform changes on the syntax (e.g., cropping sub-sentences), token (e.g., random word insertion) and character (e.g., character swapping) levels.We systematically compare the methods on part-of-speech tagging, dependency parsing and semantic role labeling for a diverse set of language families using various models including the architectures that rely on pretrained multilingual contextualized language models such as mBERT. Augmentation most significantly improves dependency parsing, followed by part-of-speech tagging and semantic role labeling. We find the experimented techniques to be effective on morphologically rich languages in general rather than analytic languages such as Vietnamese. Our results suggest that the augmentation techniques can further improve over strong baselines based on mBERT, especially for dependency parsing. We identify the character-level methods as the most consistent performers, while synonym replacement and syntactic augmenters provide inconsistent improvements. Finally, we discuss that the results most heavily depend on the task, language pair (e.g., syntactic-level techniques mostly benefit higher-level tasks and morphologically richer languages), and the model type (e.g., token-level augmentation provide significant improvements for BPE, while character-level ones give generally higher scores for char and mBERT based models).

Neural Unsupervised Semantic Role Labeling

The task of semantic role labeling ( SRL ) is dedicated to finding the predicate-argument structure. Previous works on SRL are mostly supervised and do not consider the difficulty in labeling each example which can be very expensive and time-consuming. In this article, we present the first neural unsupervised model for SRL. To decompose the task as two argument related subtasks, identification and clustering, we propose a pipeline that correspondingly consists of two neural modules. First, we train a neural model on two syntax-aware statistically developed rules. The neural model gets the relevance signal for each token in a sentence, to feed into a BiLSTM, and then an adversarial layer for noise-adding and classifying simultaneously, thus enabling the model to learn the semantic structure of a sentence. Then we propose another neural model for argument role clustering, which is done through clustering the learned argument embeddings biased toward their dependency relations. Experiments on the CoNLL-2009 English dataset demonstrate that our model outperforms the previous state-of-the-art baseline in terms of non-neural models for argument identification and classification.

Artists, Artworks, Aesthetics, Cognition

Skepticism about neuroaesthetics emerges from a contrast between aesthetic and cognitivist theories of art. Neuroaesthetics represents an aesthetic approach to understanding art. Aesthetic approaches identify the defining features of artworks by their aesthetic features and the affective profile of the experiences they engender. Cognitivist theories, in contrast, define artworks as communicative devices intentionally designed to convey some point, purpose, or meaning. In the article under discussion, the author argues that the conflict between these two views is overblown. He introduces a diagnostic recognition framework for understanding art grounded in a biased competition theory of selective attention. The framework defines artworks as attentional engines intentionally designed to orient perceivers to diagnostic features, including aesthetic features, that carry information about their point, purpose, or meaning. The artistic salience of aesthetic features of a work on this account, consistent with a cognitivist approach, lies in the semantic role they play in the expression of the work’s point, purpose, or meaning.

Predicados de comunicação em português europeu: nominalizações e nomes autónomos

This paper provides an overview of the verbal and noun predicates involving the concept of communication and their distribution in the lexicon‑grammar of European Portuguese. Two key concepts are used: (i) the agent‑speaker semantic role (and other related roles, such as message, and addressee), associated with the subject syntactic slot of these predicates; and (ii) the possibility of the verb to enter a verbum dicendi construction, i.e., introducing direct speech

Subjects in Austronesian

In LFG, grammatical functions are primitives of the theory and treated as both fundamental and universal. However, there is a long standing debate in the wider literature as to whether grammatical functions should be considered universal or language specific/construction-specific notions. Western Austronesian languages have played a large role in this debate on account of their unusual verbal morphology and the split in typical subject properties between the actor semantic role and the argument privileged by the verbal morphology. In this chapter, Hemmings addresses the debate in relation to empirical data from the Kelabit language of Northern Sarawak. She argues that the Kelabit data provides a number of arguments for treating the privileged argument as subject, and the actor as an object in non-actor voice constructions. This has important implications for the treatment of subjects crosslinguistically, Western Austronesian verbal morphology and linking theories.