Linguists characterize dialects by the presence, absence, and frequency of dozens of interpretable features. Detecting these features in text has applications to social science and dialectology, and can be used to assess the robustness of natural language processing systems to dialect differences. For most dialects, large-scale annotated corpora for these features are unavailable, making it difficult to train recognizers. Linguists typically define dialect features by providing a small number of minimal pairs, which are paired examples distinguished only by whether the feature is present, while holding everything else constant. In this paper, we present two multitask learning architectures for recognizing dialect features, both based on pretrained transformers. We evaluate these models on two test sets of Indian English, annotated for a total of 22 dialect features. We find these models learn to recognize many features with high accuracy; crucially, a few minimal pairs can be nearly as effective for training as thousands of labeled examples. We also demonstrate the downstream applicability of our dialect feature detection model as a dialect density measure and as a dialect classifier.
Natural language processing systems often struggle with out-of-vocabulary (OOV) terms, which do not appear in training data. Blends, such as "innoventor", are one particularly challenging class of OOV, as they are formed by fusing together two or more bases that relate to the intended meaning in unpredictable manners and degrees. In this work, we run experiments on a novel dataset of English OOV blends to quantify the difficulty of interpreting the meanings of blends by large-scale contextual language models such as BERT. We first show that BERT's processing of these blends does not fully access the component meanings, leaving their contextual representations semantically impoverished. We find this is mostly due to the loss of characters resulting from blend formation. Then, we assess how easily different models can recognize the structure and recover the origin of blends, and find that context-aware embedding systems outperform character-level and context-free embeddings, although their results are still far from satisfactory.
In this paper, we propose a new adversarial augmentation method for Neural Machine Translation (NMT). The main idea is to minimize the vicinal risk over virtual sentences sampled from two vicinity distributions, of which the crucial one is a novel vicinity distribution for adversarial sentences that describes a smooth interpolated embedding space centered around observed training sentence pairs. We then discuss our approach, AdvAug, to train NMT models using the embeddings of virtual sentences in sequence-to-sequence learning. Experiments on Chinese-English, English-French, and English-German translation benchmarks show that AdvAug achieves significant improvements over the Transformer (up to 4.9 BLEU points), and substantially outperforms other data augmentation techniques (e.g. back-translation) without using extra corpora.
Dual encoder architectures perform retrieval by encoding documents and queries into dense low-dimensional vectors, and selecting the document that has the highest inner product with the query. We investigate the capacity of this architecture relative to sparse bag-of-words retrieval models and attentional neural networks. We establish new connections between the encoding dimension and the number of unique terms in each document and query, using both theoretical and empirical analysis. We show an upper bound on the encoding size, which may be unsustainably large for long documents. For cross-attention models, we show an upper bound using much smaller encodings per token, but such models are difficult to scale to realistic retrieval problems due to computational cost. Building on these insights, we propose a simple neural model that combines the efficiency of dual encoders with some of the expressiveness of attentional architectures, and explore a sparse-dense hybrid to capitalize on the precision of sparse retrieval. These models outperform strong alternatives in open retrieval.
Collective attention is central to the spread of real world news and the key to understanding how public discussions report emerging topics and breaking news. Most research measures collective attention via activity metrics such as post volume. While useful, this kind of metric obscures the nuanced content side of collective attention, which may reflect how breaking events are perceived by the public. In this work, we conduct a large-scale language analysis of public online discussions of breaking crisis events on Facebook and Twitter. Specifically, we examine how people refer to locations of hurricanes in their discussion with or without contextual information (e.g. writing "San Juan" vs. "San Juan, Puerto Rico") and how such descriptor expressions are added or omitted in correlation with social factors including relative time, audience and additional information requirements. We find that authors' references to locations are influenced by both macro-level factors such as the location's global importance and micro-level social factors like audience characteristics, and there is a decrease in descriptor context use over time at a collective level as well as at an individual-author level. Our results provide insight that can help crisis event analysts to better predict the public's understanding of news events and to determine how to share information during such events.
Diachronic word embeddings offer remarkable insights into the evolution of language and provide a tool for quantifying socio-cultural change. However, while this method identifies words that have semantically shifted, it studies them in isolation; it does not facilitate the discovery of documents that lead or lag with respect to specific semantic innovations. In this paper, we propose a method to quantify the degree of semantic progressiveness in each usage. These usages can be aggregated to obtain scores for each document. We analyze two large collections of documents, representing legal opinions and scientific articles. Documents that are predicted to be semantically progressive receive a larger number of citations, indicating that they are especially influential. Our work thus provides a new technique for identifying lexical semantic leaders and demonstrates a new link between early adoption and influence in a citation network.
In this article we describe our experiences with computational text analysis. We hope to achieve three primary goals. First, we aim to shed light on thorny issues not always at the forefront of discussions about computational text analysis methods. Second, we hope to provide a set of best practices for working with thick social and cultural concepts. Our guidance is based on our own experiences and is therefore inherently imperfect. Still, given our diversity of disciplinary backgrounds and research practices, we hope to capture a range of ideas and identify commonalities that will resonate for many. And this leads to our final goal: to help promote interdisciplinary collaborations. Interdisciplinary insights and partnerships are essential for realizing the full potential of any computational text analysis that involves social and cultural concepts, and the more we are able to bridge these divides, the more fruitful we believe our work will be.
The text of clinical notes can be a valuable source of patient information and clinical assessments. Historically, the primary approach for exploiting clinical notes has been information extraction: linking spans of text to concepts in a detailed domain ontology. However, recent work has demonstrated the potential of supervised machine learning to extract document-level codes directly from the raw text of clinical notes. We propose to bridge the gap between the two approaches with two novel syntheses: (1) treating extracted concepts as features, which are used to supplement or replace the text of the note; (2) treating extracted concepts as labels, which are used to learn a better representation of the text. Unfortunately, the resulting concepts do not yield performance gains on the document-level clinical coding task. We explore possible explanations and future research directions.