It can be difficult to identify the language that a tweet is written in. In addition to being very short, they often include code-switching, where the user uses two or more languages together, or names borrowed from a different language. This dataset contains tweets from a variety of languages, tagged for whether they are in English or not, whether they contain code-switching, whether they includes names from a different language and whether they were generated automatically. This dataset contains 10,502 tweets, randomly sampled from all publicly available geotagged Twitter messages, annotated for being in English, non-English, or having code switching, language ambiguity or having been automatically generated. It includes messages sent from 130 different countries. The file all_annotated.tsv contains the dataset of 10,502 tweets used in the paper. Text is encoded as UTF-8. The column headings (also given in the .tsv file) are: tweet ID, ISO country code, tweet date, tweet text, definitely English, ambiguous, definitely not English, code-switched, ambiguous due to named entities, and automatically generated tweets.

Resource Fields

Project Open Data Required Fields (version 1.1)

Project Open Data Additional Fields (version 1.0)

Project Open Data Required-if-Applicable Fields (version 1.1)