What to do about bad language on the internet
Published 1 June 2013
Jacob Eisenstein
Citations334
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
A critical review of the NLP community's response to the landscape of bad language is offered, and a quantitative analysis of the lexical diversity of social media text, and its relationship to other corpora is presented.
Abstract
The rise of social media has brought computational linguistics in ever-closer contact with bad language: text that defies our expectations about vocabulary, spelling, and syntax. This paper surveys the landscape of bad language, and offers a critical review of the NLP community’s response, which has largely followed two paths: normalization and domain adaptation. Each approach is evaluated in the context of theoretical and empirical work on computer-mediated communication. In addition, the paper presents a quantitative analysis of the lexical diversity of social media text, and its relationship to other corpora. 1
Keywords
Computer Science
Information Communication & SocietyCRITICAL QUESTIONS FOR BIG DATA
5,321 Citations2012danah boyd, Kate Crawford
Six provocations are offered to spark conversations about the issues of Big Data: a cultural, technological, and scholarly phenomenon that rests on the interplay of technology, analysis, and mythology that provokes extensive utopian and dystopian rhetoric.
Discourse StudiesIdentity and interaction: a sociocultural linguistic approach
3,676 Citations2005Mary Bucholtz, Kira Hall
Earthquake shakes Twitter users
3,667 Citations2010Takeshi Sakaki, Makoto Okazaki +1 more
This paper investigates the real-time interaction of events such as earthquakes in Twitter and proposes an algorithm to monitor tweets and to detect a target event and produces a probabilistic spatiotemporal model for the target event that can find the center and the trajectory of the event location.
CERN Document Server (European Organization for Nuclear Research)Natural Language Processing with Python
3,449 Citations2009Steven Bird, Ewan Klein +1 more
Incorporating non-local information into information extraction systems by Gibbs sampling
3,035 Citations2005Jenny Rose Finkel, Trond Grenager +1 more
By using simulated annealing in place of Viterbi decoding in sequence models such as HMMs, CMMs, and CRFs, it is possible to incorporate non-local structure while preserving tractable inference.
Feature-rich part-of-speech tagging with a cyclic dependency network
2,851 Citations2003Kristina Toutanova, Dan Klein +2 more
A new part-of-speech tagger is presented that demonstrates the following ideas: explicit use of both preceding and following tag contexts via a dependency network representation, broad use of lexical features, and effective use of priors in conditional loglinear models.
Journal of SociolinguisticsVariation and the indexical field<sup>1</sup>
1,963 Citations2008Penelope Eckert
Cambridge University Press eBooksLanguage and Gender
1,258 Citations2013Penelope Eckert, Sally McConnell‐Ginet
Named Entity Recognition in Tweets: An Experimental Study
1,203 Citations2011Alan Ritter, Sam Clark +1 more
The novel T-ner system doubles F1 score compared with the Stanford NER system, and leverages the redundancy inherent in tweets to achieve this performance, using LabeledLDA to exploit Freebase dictionaries as a source of distant supervision.
ScienceDiurnal and Seasonal Mood Vary with Work, Sleep, and Daylength Across Diverse Cultures
1,176 Citations2011Scott A. Golder, Michael W. Macy
Individual-level diurnal and seasonal mood rhythms in cultures across the globe are identified, using data from millions of public Twitter messages, and it is found that individuals awaken in a good mood that deteriorating as the day progresses.
Social Science Computer ReviewThe Impacts of Emoticons on Message Interpretation in Computer-Mediated Communication
862 Citations2001Joseph B. Walther, Kyle P. D’Addario
The results indicate that emoticons’ contributions were outweighed by verbal content, but a negativity effect appeared such that any negative message aspect—verbal or graphic—shifts message interpretation in the direction of the negative element.
FigsharePart-of-Speech Tagging for Twitter: Annotation, Features, and Experiments
830 Citations2018Kevin Gimpel, Nathan Schneider +8 more
Proceedings of the International AAAI Conference on Web and Social MediaCharacterizing Microblogs with Topic Models
724 Citations2010Daniel Ramage, Susan Dumais +1 more
A scalable implementation of a partially supervised learning model (Labeled LDA) that maps the content of the Twitter feed into dimensions that correspond roughly to substance, style, status, and social characteristics of posts is presented.
Find me if you can
712 Citations2010Lars Bäckström, Eric Sun +1 more
Using user-supplied address data and the network of associations between members of the Facebook social network, an algorithm is introduced that predicts the location of an individual from a sparse set of located users with performance that exceeds IP-based geolocation.
FigshareImproved Part-of-Speech Tagging for Online Conversational Text with Word Clusters
682 Citations2018Olutobi Owoputi, Brendan O’Connor +4 more
This work systematically evaluates the use of large-scale unsupervised word clustering and new lexical features to improve tagging accuracy on Twitter and achieves state-of-the-art tagging results on both Twitter and IRC POS tagging tasks.
Journal of English LinguisticsMobility, Indexicality, and the Enregisterment of “Pittsburghese”
667 Citations2006Barbara Johnstone, Jennifer Andrus +1 more
Classifying latent user attributes in twitter
650 Citations2010Delip Rao, David Yarowsky +2 more
A novel investigation of stacked-SVM-based classification algorithms over a rich set of original features, applied to classifying these four user attributes, as distinct from the other primarily spoken genres previously studied in the user-property classification literature.
Communication TheoryFunctions of the Nonverbal in CMC: Emoticons and Illocutionary Force
645 Citations2010Eli Dresner, Susan C. Herring
FigshareA Latent Variable Model for Geographic Lexical Variation
607 Citations2018Jacob Eisenstein, Brendan O’Connor +2 more
A multi-level generative model that reasons jointly about latent topics and geographical regions is presented, which recovers coherent topics and their regional variants, while identifying geographic areas of linguistic consistency.
Empirical Methods in Natural Language ProcessingDiscriminating Gender on Twitter
566 Citations2011John D. Burger, John C. Henderson +2 more
The construction of a large, multilingual dataset labeled with gender is described and statistical models for determining the gender of uncharacterized Twitter users are investigated, and several different classifier types are explored.
Lexical Normalisation of Short Text Messages: Makn Sens a #twitter
455 Citations2011Bo Han, Timothy Baldwin
This paper targets out-of-vocabulary words in short text messages and proposes a method for identifying and normalising ill-formed words, which achieves state- of-the-art performance over an SMS corpus and a novel dataset based on Twitter.
National Research Council Canada (Government of Canada)Unsupervised Modeling of Twitter Conversations
429 Citations2010Alan Ritter, Colin Cherry +1 more
This work proposes the first unsupervised approach to the problem of modeling dialogue acts in an open domain, trained on a corpus of noisy Twitter conversations, and addresses the challenge of evaluating the emergent model with a qualitative visualization and an intrinsic conversation ordering task.
American SpeechLINGUISTIC RUIN? LOL! INSTANT MESSAGING AND TEEN LANGUAGE
404 Citations2008Sali A. Tagliamonte, Derek Denis
Recognizing Named Entities in Tweets
372 Citations2011Xiaohua Liu, Shaodian Zhang +2 more
This work proposes to combine a K-Nearest Neighbors classifier with a linear Conditional Random Fields model under a semi-supervised learning framework to tackle the challenges of Named Entities Recognition for tweets.
Computer Speech & LanguageNormalization of non-standard words
340 Citations2001Richard Sproat, Alan W. Black +4 more
A taxonomy of NSWs was developed on the basis of four rather distinct text types, and several general techniques including n-gram language models, decision trees and weighted finite-state transducers were investigated, demonstrating that a systematic treatment can lead to better results than have been obtained by the ad hoc treatments that have typically been used in the past.
Journal of the American Society for Information Science and TechnologyHip and trendy: Characterizing emerging trends on Twitter
323 Citations2011Mor Naaman, Hila Becker +1 more
This study develops a taxonomy of the trends present in the data of Twitter messages from one geographic area, and identifies important dimensions according to which trends can be categorized, as well as the key distinguishing features of trends that can be derived from their associated messages.
Journal of Computer-Mediated CommunicationFrom Statistical Panic to Moral Panic: The Metadiscursive Construction and Popular Exaggeration of New Media Language in the Print Media
312 Citations2006Crispin Thurlow
This article critically reviews an international corpus of 101 print-media accounts of language-use in technologies such as instant messaging and text messaging to demonstrate the most recurrent metadiscursive themes in mediatized depictions of technologically or computer-mediated discourse (CMD).
Annotating named entities in Twitter data with crowdsourcing
285 Citations2010Tim Finin, William Murnane +4 more
This paper describes the experience using both Amazon Mechanical Turk (MTurk) and Crowd-Flower to collect simple named entity annotations for Twitter status updates, and describes how to use MTurk to collect judgements on the quality of "word clouds."
A phrase-based statistical model for SMS text normalization
280 Citations2006AiTi Aw, Min Zhang +2 more
This paper views the task of SMS normalization as a translation problem from the SMS language to the English language and proposes to adapt a phrase-based statistical MT model for the task, which can largely boost SMS translation performance.
Simple supervised document geolocation with geodesic grids
215 Citations2011Benjamin Wing, Jason Baldridge
This work investigates automatic geolocation (i.e. identification of the location, expressed as latitude/longitude coordinates) of documents and describes several simple supervised methods for document geolocated using only the document's raw text as evidence.
Cooooooooooooooollllllllllllll!!!!!!!!!!!!!! Using Word Lengthening to Detect Sentiment in Microblogs
213 Citations2011Samuel Brody, Nicholas Diakopoulos
An automatic method which leverages word lengthening to adapt a sentiment lexicon specifically for Twitter and similar social messaging networks, and evaluates its method by comparison to human judgments, and analyzes its strengths and weaknesses.
International Journal on Document Analysis and Recognition (IJDAR)Investigation and modeling of the structure of texting language
211 Citations2007Monojit Choudhury, Rahul Saraf +4 more
The nature and type of compressions used in SMS texts are investigated, and a Hidden Markov Model based word-model for TL is developed, which results in a 35% reduction of the relative word level error rates.
First MondayMining the Blogosphere: Age, gender and the varieties of self-expression
211 Citations2007Shlomo Argamon, Moshe Koppel +2 more
It is found that a number of stylistic and content-based indicators are significantly affected by both age and gender, and that the main difference between older and younger bloggers, and between male and female bloggers, lies in the extent to which their discourse is outer- or inner-directed.
DSpace@MIT (Massachusetts Institute of Technology)Event Discovery in Social Media Feeds
182 Citations2011Edward Benson, Aria Haghighi +1 more
A graphical model is developed that addresses record extraction from social streams such as Twitter by learning a latent set of records and a record-message alignment simultaneously, resulting in a set of canonical records that are consistent with aligned messages.
Automatically Constructing a Normalisation Dictionary for Microblogs
181 Citations2012Bo Han, Paul Cook +1 more
This paper proposes a method for constructing a dictionary of lexical variants of known words that facilitates lexical normalisation via simple string substitution and shows that a dictionary-based approach achieves state-of-the-art performance for both F-score and word error rate on a standard dataset.
Age Prediction in Blogs: A Study of Style, Content, and Online Behavior in Pre- and Post-Social Media Generations
175 Citations2011Kathleen McKeown, Sara Rosenthal
It is found that the birth dates of students in college at the time when social media such as AIM, SMS text messaging, MySpace and Facebook first became popular, enable accurate age prediction.
Summarizing Microblogs Automatically
170 Citations2010Beaux Sharifi, Mark-Anthony Hutton +1 more
An algorithm is developed that takes a trending phrase or any phrase specified by a user, collects a large number of posts containing the phrase, and provides an automatically created summary of the posts related to the term.
Journal of SociolinguisticsLanguage variation on Internet Relay Chat: A social network approach
168 Citations2001John C. Paolillo
Examination of linguistic variation on an Internet Relay Chat channel with respect to the hypothesis that standard variants tend to be associated with weak social network ties, while vernacular variants are associated with strong network ties reveals a structured relationship between tie strength and several linguistic variants.
Overview of the 2012 Shared Task on Parsing the Web
166 Citations2012Slav Petrov, Ryan McDonald
A shared task on parsing web text from the Google Web Treebank to build a single parsing system that is robust to domain changes and can handle noisy text that is commonly encountered on the web is described.
North American Chapter of the Association for Computational LinguisticsAutomatic Domain Adaptation for Parsing
160 Citations2010David McClosky, Eugene Charniak +1 more
The resulting system proposes linear combinations of parsing models trained on the source corpora that outperforms all non-oracle baselines including the best domain-independent parsing model.
Expressing Social Relationships on the Blog through Links and Comments.
142 Citations2007Noor Ali-Hasan, Lada A. Adamic
A study of bloggers’ online and real life relationships in three blog communities: Kuwait Blogs, Dallas/Fort Worth blogs, and United Arab Emirates Blogs finds different characteristics for different kinds of links.
Journal of Literacy ResearchR u txting? Is the Use of Text Speak Hurting Your Literacy?
133 Citations2009Michelle Drouin, Claire Davis
An unsupervised model for text message normalization
132 Citations2009Paul Cook, Suzanne Stevenson
An unsupervised noisy-channel model for text message normalization is constructed on a sample of creative, non-standard text message word forms to determine frequent word formation processes in texting language.
FigshareDiscovering Sociolinguistic Associations with Structured Sparsity
124 Citations2018Jacob Eisenstein, Noah A. Smith +1 more
A method to discover robust and interpretable sociolinguistic associations from raw geotagged text data is presented, using aggregate demographic statistics about the authors' geographic communities to solve a multi-output regression problem between demographics and lexical frequencies.
Insertion, Deletion, or Substitution? Normalizing Text Messages without Pre-categorization nor Supervision
108 Citations2011Fei Liu, Fuliang Weng +2 more
This paper proposes a unified letter transformation approach that requires neither pre-categorization nor human supervision and significantly outperformed the state-of-the-art deletion-based abbreviation system and the jazzy spell checker.
Joint Inference of Named Entity Recognition and Normalization for Tweets
106 Citations2012Xiaohua Liu, Ming Zhou +3 more
A novel graphical model is proposed to simultaneously conduct NER and NEN on multiple tweets to address the problem of named entity normalization for tweets, which introduces a binary random variable for each pair of words with the same lemma across similar tweets.
Predicting response to political blog posts with topic models
102 Citations2009Tae Yano, William W. Cohen +1 more
The models jointly describe the generation of the primary documents as well as the authorship and the contents of the blog community's verbal reactions to each post (comments) and evaluate their model on a novel comment prediction task.
Contextual Bearing on Linguistic Variation in Social Media
98 Citations2011Stephan Gouws, Donald Metzler +2 more
The empirical study investigates properties of lexical transformations as observed within Twitter microtexts and reveals that different populations of users exhibit different amounts of shortened English terms and different shortening styles.
Journal of Computer-Mediated CommunicationThe Virtual Speech Community: Social Network and Language Variation on IRC
97 Citations2006John C. Paolillo
Language Identification for Creating Language-Specific Twitter Collections
92 Citations2012Shane Bergsma, Paul McNamee +3 more
This work annotates and releases a large collection of tweets in nine languages, focusing on confusable languages using the Cyrillic, Arabic, and Devanagari scripts, the first publicly-available collection of LID-annotated tweets in non-Latin scripts and should become a standard evaluation set for LID systems.
Proceedings of the International AAAI Conference on Web and Social MediaDude, srsly?: The Surprisingly Formal Nature of Twitter's Language
81 Citations2021Yuheng Hu, Kartik Talamadupula +1 more
Twitter's language is surprisingly more conservative, and less informal than SMS and online chat; Twitter users appear to be developing linguistically unique styles; and Twitter has less variation of affect than other more formal mediums.
DSpace@MIT (Massachusetts Institute of Technology)Content Models with Attitude
76 Citations2011Christina Sauper, Aria Haghighi +1 more
A probabilistic topic model for jointly identifying properties and attributes of social media review snippets and captures aggregate user sentiments towards these properties that enables discovery of highly rated or inconsistent properties of a product.
Dublin City University Open Access Institutional Repository (Dublin City University)From News to Comment: Resources and Benchmarks for Parsing the Language of Web 2.0
71 Citations2011Jennifer Foster, Özlem Çetinoğlu +5 more
It is found that the Wall-Street-Journal-trained statistical parsers have a particular problem with tweets and that a substantial part of this problem is related to POS tagging accuracy.
An Exploration of Observable Features Related to Blogger Age.
63 Citations2006John D. Burger, John C. Henderson
This paper offers an initial exploratory data analysis of candidate features for blogger age prediction by analyzing the text and metadata of blog entries for evidence of blogger age.
Unsupervised Mining of Lexical Variants from Noisy Text
53 Citations2011Stephan Gouws, Dirk Hovy +1 more
A novel unsupervised method for extracting domain-specific lexical variants given a large volume of text and yields a 20% reduction in word error rate over an existing state-of-the-art approach.
Phonological Factors in Social Media Writing
42 Citations2013Jacob Eisenstein
Examples of the phonological variable of consonant cluster reduction in Twitter suggest that when social media writing transcribes phonological properties of speech, it is not merely a case of inventing orthographic transcriptions, but rather, social media displays influence from structural properties ofThe phonological system.
GENDER IN TWITTER: STYLES, STANCES, AND SOCIAL NETWORKS
40 Citations2012David Bamman, Jacob Eisenstein +1 more
Pairing computational methods and social theory offers a new perspective on how gender emerges as individuals position themselves relative to audiences, topics, and mainstream gender norms.
arXiv (Cornell University)Mapping the geographical diffusion of new words
36 Citations2012Jacob Eisenstein, Brendan O’Connor +2 more
Short message communications
23 Citations2012Robert Munro, Christopher D. Manning
It is concluded that there is only limited utility in treating SMS and Twitter as equivalent information sources -- perhaps much less than the relatively large number of recent Twitter-focused papers would indicate.
Multiple Narrative Disentanglement: Unraveling Infinite Jest
17 Citations2012Byron Wallace
This work proposes and evaluates a novel unsupervised approach to MND that is motivated by the theory of narratology, and achieves strong empirical results, successfully disentangling the threads in Infinite Jest and significantly outperforming baseline strategies in doing so.
Unsupervised Part-of-Speech Tagging in Noisy and Esoteric Domains With a Syntactic-Semantic Bayesian HMM
17 Citations2012William Darling, Michael J. Paul +1 more
This work presents Part-of-Speech LDA (POSLDA), a syntactically and semantically consistent generative probabilistic model that consistently achieves improvements in unsupervised POS tagging and language modeling over the Bayesian HMM approach with varying amounts of side information in the noisy and esoteric domain of Twitter.
Discourse Context & MediaThe discourse deictics ∧and <-- in a World of Warcraft community
8 Citations2012Lauren Collister