Intelligent CALL: Using Pattern Matching to Learn English

Intelligent CALL: Using Pattern Matching to Learn English

John Blake (University of Aizu, Japan)
DOI: 10.4018/978-1-7998-2591-3.ch001

Abstract

This chapter shows readers the importance and application of pattern matching in learning languages; specifically, the application of natural language processing to address specific problems of Japanese learners of English at a public university. The chapter introduces the concepts of patterns, detection, and detection methods. The author turns to the pedagogic application of pattern matching, first discussing the relevant theory, then describing hacks developed by language teachers and learners. The final section describes and evaluates iCALL tools developed at the University of Aizu, including a mobile app and the Pronunciation Scaffolder, a real-time presentation script annotator.
Chapter Preview
Top

Introduction

Patterns permeate every aspect of life (Carbone, Gronov, & Prusinkiewicz, 2004) from predicting weather based on cloud formations to distinguishing friends from foe based on behaviour patterns. Some of these patterns are learnt simply through exposure, but others may involve various combinations of intuition and experience. Pattern recognition is also a key tenet in language learning and has been the focus of scholarly articles for many years. In fact, almost a century ago, Sapir (1925) published an article focusing on sound patterns. Actively assisting language learners to notice and use patterns is thought to help learners master a new language more effectively and more efficiently (Ellis, 1994). More recently, Hunston and Francis (2000) coined the term pattern grammar to describe the contextual usage patterns associated with particular word senses.

Most children begin learning their first language through extensive listening (Asher, 1972). During the receptive phase when children listen but are unable to speak, language patterns in the various inter-related language systems (e.g., phonemes, morphemes, lexis and grammar) are thought to be acquired. Second language learning may differ from first language learning depending on various factors, such as the type of tuition, purpose of learning, relationship between mother tongue and second language, and the age of the learner (Cook, 2016; Johnson, 2017). For example, adult learners already literate in their mother tongue may first focus on reading a second language and start identifying visual language patterns before audio language patterns. Learners of a second language with the same script as their first language are more able to draw on their existing knowledge of orthographic patterns. Lexis in languages that are linguistically close share more in common than lexis in language that are linguistically distant. The greater the linguistic distance, the greater the difficulty to learn the second or additional language (Piske, MacKay, & Flege, 2001).

This chapter aims to show readers the importance and application of pattern recognition in learning languages; specifically, it focuses on the application of natural language processing to assist Japanese learners of English at a public university. These learners studied English for at least six years before entering university and study for four more years during their university program. English language learning was a compulsory school subject from the first year of junior high school when students are approximately thirteen years old. The vast majority have extensive passive vocabularies, general knowledge of many grammatical forms, and detailed knowledge of some obscure grammatical oddities that are often tested on university entrance exams. Despite their extensive knowledge of rules and words, few Japanese university students can converse freely in English. King (2013), having observed thirty English language classes in multiple Japanese universities concluded that there was “a robust trend, with minimal variation, toward silence” (p. 337).

This chapter begins by reviewing the origins of computer assisted language learning (CALL) and the transition to intelligent CALL. This is followed by a discussion of patterns in language and pattern detection. The chapter then introduces the concept of discovery learning and describes the pedagogic application of pattern matching. Pattern matching and detection tools that were developed specifically for learning English are then introduced. Specifically, two iCALL tools, both of which were developed at the University of Aizu are discussed in depth. The first iCALL tool is a mobile app called WordBricks and the second, the Pronunciation Scaffolder, a real-time presentation script annotator. WordBricks (Purgina, Mozgovoy, & Blake, 2019) was designed to help language learners acquire knowledge of syntactical patterns through a gamified environment in which users attempt to piece together jigsaw-like texts. The second tool is the Pronunciation Scaffolder (Blake, 2019) that targets learners of English who can read and understand texts but have difficulty in reading aloud. It was designed to help Japanese learners more easily read prepared presentation scripts using colour, size and symbols to visualize various pronunciation features. The chapter concludes by discussing future research directions and noting the increasing importance of deep learning in iCALL.

Key Terms in this Chapter

Zipf's law: This law states that the frequency of a word is inversely proportional to the rank in frequency of the word.

Pattern: Pattern describes items that are organized regularly not randomly.

Regex: Regex and Regexp stand for regular expressions, which are powerful search expressions that can match characters, words and/or strings.

Hapax legomenon: Words that occur only once within a particular text or corpus of texts.

NLP: NLP stands for natural language processing, which uses computational methods to analyze natural language.

Algorithm: A set of rules to be followed.

Discovery Learning: A theory in which learners reflect on their experiences to discover new ideas.

Machine Learning: When computers use artificial intelligence to learn from data or experience.

Rule-based parsing: A process that use rules related to syntactic structure to divide written texts into components.

Complete Chapter List

Search this Book:
Reset