Corpus Linguistics 08

Corpus Linguistics (1)

• Multilingualism is a common phenomenon globally
• Diglossia involves using two languages in a speech community for different functions
• Code-switching signifies changing between languages based on the situation
• Code-mixing involves switching between languages within communication
• English usage varies globally, including in Anglosphere countries (UK, Ireland, USA, Canada, Australia, New Zealand) and former colonial territories in Africa, Asia, the Caribbean, and Oceania
• World Englishes are modeled using models like Kachru's Three Circles Model and Schneider's Dynamic Model
• Pidgins and creoles are typologically unrelated languages, evolving through language contact and exhibiting simplified structures
• These languages are prevalent in West Africa, the Caribbean, the US and Central America, and the Pacific.

Today's Lecture

• The lecture covers Introduction, Corpora, Concordancing, and Part-of-speech Tagging

Introduction to Corpus Linguistics

• Corpus linguistics is a scientific method of language analysis relying on data from language corpora to support statements about language.
• The BROWN Corpus (1 million words), released in 1967, was a significant early corpus
• Modern corpora encompass billions of words
• Corpus linguistics aids in language trend analysis and authentic language teaching materials

Why Corpus Linguistics is Useful

• Language trend analysis: Identifying trends in language use over time and social media
• Language teaching: Creating authentic teaching materials based on real-life language use and ESL textbooks
• Lexicography: Enhancing dictionary definitions with usage information
• AI & NLP: Improving computational models of language understanding and training AI models (Siri, Alexa)
• Discourse analysis: Uncovering patterns and structures in spoken/written text (e.g. political speeches)
• Linguistic hypotheses testing: Validating or refuting language usage theories

What is a Corpus?

• A corpus is a collection of electronic texts designed to represent a language or a language variety, and is built according to specific criteria for purpose of study.
• Corpora aim to represent language naturally
• Corpora can consist of written or spoken texts (including transcripts).
• Corpora are machine-readable and are sampled to be representative
• Corpora are large collections of texts which enable investigation of language phenomena

Difference between text and Corpus

• A text is a whole, coherent communicative event that is read horizontally and is an individual act of will.
• A corpus is fragmented and read vertically for formal patterning and repeated events; it represents language use as social practice.

Corpus Structure

• Corpus structure and level of annotation vary greatly
• A corpus frequently consists of annotated transcripts of written and spoken data (e.g. in file form).
• Corpora often include a manual, metadata (e.g., author, genre, title) and/or an online interface for exploration
• Metadata includes details like author, genre, and title

From Raw Data to Corpus File

• Raw data can be transformed into a corpus file using XML (Extensible Markup Language).
• Metadata includes aspects such as the date, location, author, and the nature of the written or spoken text

Corpora Overview

• Hundreds of corpora exist, covering various aspects of the English language
• Corpora can be categorized by mode (written, spoken, both), time depth (synchronic, diachronic, historical), and topic specificity
• Comprehensive online corpus lists exist

The BROWN Family

• A large collection of written English
• Comprises 8 major genres (press, religion, skills, hobbies, popular lore, belles-lettres, miscellaneous, humor)
• Provides data for diachronic and comparative analysis due to its development over time

The International Corpus of English (ICE)

• A 1 million-word collection
• Includes written and spoken data (student essays, news reports, conversations) for different varieties of English.

ICE Design: Spoken and Written

• Contains multiple varieties of English, with specific quantities for different genres for both written and spoken

COCA and COHA

• COCA (Corpus of Contemporary American English): 1 billion words; covers 1990-2019
• COHA (Corpus of Historical American English): 475 million words; covers 1820s-2010s

GloWbE and NOW

• GloWbE (Global Web-Based English) comprises 1.9 billion words from 20 countries across various linguistic domains; collected between 2012-2013
• NOW (News on the Web) is an up-to-the-minute corpus comprising 20.3 billion words, collected daily

BNC and ICLE

• BNC (British National Corpus): 100 million words; encompasses written and spoken British English from 1980-1993; 7 categories of textual data
• ICLE (International Corpus of Learner English): 3.7 million words, focusing on learner essays and diverse varieties

Concordancing Introduction

• Concordancing helps linguists investigate word occurrences and behavior in real-life contexts.
• Traditonal approaches rely on the native speaker's intuition
• Concordance offers a listing of word forms paired with their surrounding context; also known as co-text

AntConc

• AntConc is widely used for corpus analysis
• It facilitates diverse searches from individual letters to larger constructions, like sentences
• It's available across many platforms, is freely accessible, well-documented and user-friendly

Part-of-speech (POS) tagging Introduction

• POS tagging assigns grammatical categories (e.g., noun, verb, adjective) to words in a text
• This aids in understanding word classes and inflections.
• CLAWS is a commonly used automated tagging system

POS tagging Usefulness

Analysis of different tagged words and quantities shows how often words are used in different contexts.

• Provides insights into how frequently a word appears in different formats, like an adjective or adverb, helping to better understand usage

TagAnt

• TagAnt is a versatile free tagging tool for corpus analysis.
• Various tagsets(specific groupings of words) may be selected based on the project needs

Keywords

• A list of common terms used in corpus linguistics, including types of corpora, analysis techniques, and softwares/tools.

This quiz covers key concepts in Corpus Linguistics, focusing on multilingualism, diglossia, code-switching, and language variation. Learn about the global usage of English and the emergence of pidgins and creoles, along with various models explaining World Englishes. Test your understanding with questions about language dynamics and their implications.