Testing the Efficacy of Concgram Measures in the Extraction of Igbo Multiword Expressions: Implications to Lexicography

Published: January 2020

Author(s): Ochu, Michael Chima

About Author(s): University of Botswana

Download PDF


This is a corpus-driven study on the extraction of Igbo multiword expressions. The paper argues that multiword expressions are found in all types of writings in Igbo. The objective of this paper is to extract multiword expressions which are possible candidates for Igbo dictionary compilation. To achieve this, concgram, a feature of Oxford Wordsmith Tools, was used in extracting multiword candidates from an Igbo corpus of a little more than 2-million tokens. Five (5) highly frequent Igbo words were selected to determine the efficacy of concgram measures in the extraction of Igbo multiword expressions. The words are: akwụkwọ (book/paper), ala (land), anu (meat/flesh), ikpe (judgment), and okwu (word/speech). Through the use of concgram measures, some of the extracted multiword expressions include: akwụkwọnri (lit: book food; fig: vegetable); akwụkwọndu (lit: book life; fig: fresh vegetation); ozialanna (lit: message land father; fig: national service); alanri (lit: land food; fig: fertile/productive land); anuahu (lit: meat body; fig: physical); anuulo (lit: meat house; fig: domestic animal); ikpeazu (lit: judgment back; fig: trial in absentia). Results show that the concgram tool is effective in the extraction of Igbo multiword expressions. The motivation for the extraction of multiword expressions in this study is for lexicon development, which can be in the form of recognition of multiword expressions and entering them into the dictionary of Igbo.

Keywords: congram measures, corpus analysis, extraction, multiword expressions,