Quantitative Research Methods for Linguists

a questions and answers approach for students

Tim Grant, Urszula Clark, Gertrud Reershemius, Dave Pollard, Sarah Hayes, Garry Plappert

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781351722971
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

Quantitative Research Methods for Linguistics provides an accessible introduction to research methods for undergraduates undertaking research for the first time. Employing a task-based approach, the authors demonstrate key methods through a series of worked examples, allowing students to take a learn-by-doing approach and making quantitative methods less daunting for the novice researcher.

Key features include:

• 
  
  

• Chapters framed around real research questions, walking the student step-by-step through the various methods;
• Guidance on how to design your own research project;
• Basic questions and answers that every new researcher needs to know;
• A comprehensive glossary that makes the most technical of terms clear to readers;
• Coverage of different statistical packages including R and SPSS.

Quantitative Research Methods for Linguistics is essential reading for all students undertaking degrees in linguistics and English language studies.

Part

I

Basic statistical ideas

1

Basic concepts of quantification and number

This chapter and the following one in the first part of this book introduce you to basic concepts in quantitative research. Undertaking quantitative research involves working with numbers and statistics. Statistics involves collecting and analysing numerical data in large quantities drawn from representative samples and from which conclusions or inferences can be made. As a student of applied linguistics, English language or linguistics, you may not have worked with either numbers or statistics very much, if at all, as part of your post sixteen study. Often, statistics can be perceived as too difficult. Why is it necessary, then, to work with numbers and statistics? It is because statistical analysis can be an important tool in linguistic investigation. Many of the research questions we pursue in linguistic research involve measuring and observing data. Examples of this kind of research are to be found in Part 2 of the book.

An important part of learning to use statistics is knowing how to select and how to use the appropriate or right statistical technique as part of a research investigation. Many books in statistical and quantitative methods concentrate on this aspect, explaining the various statistical tests that are available for use. However, in order to understand how statistical tests work, it is necessary first of all to understand the basic and underlying concepts upon which statistical tests depend, and an understanding of these basic concepts is often assumed in many books that deal with quantitative methods. This book is organised differently. The chapters in the first part of this book start by explaining the various concepts that underpin the use of statistics and statistical tests, before going on to explain the tests themselves in the second part of the book. A glossary at the end of the book explains the terms used throughout the book.

This chapter focuses upon quantification and number. After an initial discussion of quantification, the chapter explains different types of numbers and ways of classifying numbers, including explanations of:

• Nominal numbers (numbers used as names)
• Continuous numbers (numbers you can perform mathematics and statistics on)
• Ordinal numbers (numbers indicating a ranked order)

1.1 Why quantify?

Designing research that concerns the use of quantitative research methods and statistics involves collecting data that can be measured in a numerical way. In order for this to happen, the data set you collect needs to be sufficiently large so that it can be quantified and analysed by drawing upon a range of statistical methods. The word quantify derives from the Latin verb quantificare and the corresponding noun quantus, meaning ‘how much’ or ‘how many’. When we talk about quantifying something, we generally mean that we are going to calculate or measure something and express the calculation or measurement as a number. Quantifying also implies quantity: that is, that the data we measure will be of a size to generate meaningful results. Asking how much data is enough to undertake statistical analysis is a bit like asking how long a piece of string is. It all depends upon the research question you want to answer, or the hypothesis you want to test, as the research questions in Part 2 of the book demonstrate. So, too, is the question of which statistical test to choose to analyse your data. It all depends upon the nature and type of the data you collect and the questions you are asking.

The reason we might want to quantify something will generally be to do with a research question to which we want to find an answer, or a hypothesis we want to either support or disprove. The ‘something’ we wish to quantify can take many different forms and be of different kinds. The research undertaken in applied linguistics and linguistics research is very similar to that undertaken in other social science subjects such as economics, politics and sociology, in that it is concerned with aspects of human behaviour in areas such as language teaching, phonetics and phonology, sociolinguistics and psycholinguistics. For example, we might wish to evaluate the impact of a new pedagogic approach in language teaching upon students’ language learning and its assessment, or provide descriptions of sociolinguistic phonological variation between dialect groups or the rate at which children acquire particular aspects of speech. Linguistics researchers also, in contrast to other social science researchers, undertake textual analysis that relies upon a different set of measures and techniques than those normally used in the social sciences and areas of linguistic research mentioned above. In corpus linguistics and forensic linguistics in particular, textual analysis often concentrates upon the distributional structure of word use, frequency of word use or collocations, or the distribution of errors in language learners’ written texts.

Quantifying often involves comparison. Quantifying allows us to make comparisons between, for example, individuals or groups according to some aspect of human behaviour, including the language used. So we can compare our age, weight and height against that of other people, or the number of times men interrupt a conversation as opposed to women. In turn, this allows us to formulate research questions in relation to comparisons between the different categories of data we collect or to test hypotheses about what we might expect to find. To do this the data that is collected can be analysed using an appropriate statistical test. Deciding upon the kind of statistical test you are going to apply to your data depends very much upon the nature of your data, as Chapter 2 of Part 1 explains in more detail.

Aliaga and Gunderson (2002: 14) describe what we mean by quantitative research as ‘explaining phenomena by collecting numerical data that are analysed using mathematically based methods (in particular statistics)’. Explaining phenomena is a key element of all research. When we set out to do research, we are always looking to explain something. In linguistics, this could be questions like those given in Part 2 of this book: exploring who actually uses a regional dialect (Low German) and in what way or whether Birmingham English really is more nasal than the varieties that surround it. In quantitative research, we collect numerical data, which is then analysed using mathematically based methods. Quantitative research is essentially about collecting numerical data to explain a particular phenomenon, and particular questions seem immediately suited to being answered using quantitative methods, such as those given above.

The last part of Aliaga and Gunderson’s definition refers to the use of mathematically based methods, in particular statistics, to analyse the data. Using statistics to analyse data is the element that puts a lot of people off doing quantitative research in linguistics, as the mathematics underlying the methods appears complicated and frightening. However, as researchers, we do not have to be particularly expert in the mathematics underlying the methods. Much of the statistical analysis we need to undertake in linguistics can be done through the use of computer software, such as a spreadsheet programme like Microsoft Excel, that allows us to do the analyses quickly and (relatively) easily. We do need, however, to understand which statistical tests to apply to our respective data sets in order to answer our research questions, and to do this we need to understand what they are doing.

Statistical analysis normally involves measuring or calculating something that involves the use of numbers. Numbers, though, can be categorised in different ways. For example, we know the date we were born and how old we are, expressed in numbers of years and months. We are all also generally aware of how much we weigh, expressed in pounds, stones or kilograms, and of how tall we are, expressed in feet and inches or metres and centimetres. At its most basic, quantifying is what you do when you express observations as numbers. What, though, is a number? Understanding the concept of number and the different ways numbers can be classified is fundamental to undertaking quantitative analysis. Arguably, it is more important than understanding the formulae employed by different statistical tests, since computer programs do the analysis for you.

1.2 What is a number?

The notion of numbers and counting dates back to prehistory, and most communities or societies, however simple, have some system of counting. As societies and humankind evolved and tribes and groups formed, it became important to be able to know how many members were in any one group, and also how many there were in any other, including possibly the enemy’s camp. It was also important to know how many animals there were in a flock or herd and whether or not it was increasing or decreasing in size. ‘Just how many do we have’ is a question we can imagine members of a community asking themselves or each other, in relation to the tribe itself and any animals they possessed.

It is thought that one of the earliest methods of counting items such as animals or people was with counting or tally sticks. These are objects used to track the numbers of items to be counted. With this method, each ‘stick’ (or pebble, or whatever counting device was being used) represents one animal or object. This method uses the idea of one to one correspondence. In a one to one correspondence, items that are being counted are uniquely linked with some counting tool.

In the picture below, you see each stick corresponding to one horse. If you wanted to ‘count off’ your animals to make sure they were all present, you could do this by mentally (or methodically) assigning each stick to one animal and continuing to do so until you were satisfied that all were accounted for.

The earliest counting device was the human hand and its fingers, capable of counting up to ten things (although toes were also used to count in certain cultures). Then, as even larger quantities (greater than ten fingers and ten toes could represent) were counted, various natural items like pebbles, sea-shells and twigs were used to help keep count. Another possible way of employing the tally stick counting method was by making marks or cutting notches into pieces of wood, as shown below. Such marks and sticks acted as a counting board.

With the invention of writing, symbols were found to represent the numbers just as they were found to represent letters. Over many centuries, the sticks were replaced with abstract objects and...

Table of contents

• Cover
• Title
• Copyright
• Contents
• Preface
• Acknowledgements
• Part I: Basic statistical ideas
• Part II: Asking and answering quantitative questions
• Glossary
• Index