Psycholinguistic Methodology in Phonological Research

by

Shigeto Kawahara

• LAST REVIEWED: 21 January 2016
• LAST MODIFIED: 21 January 2016
• DOI: 10.1093/obo/9780199772810-0021

Introduction

The primary data collection strategy deployed by modern theoretical linguistics is to use intuition of native speakers. This methodology has been criticized throughout the history of generative syntax, although this criticism itself has been much debated (see the Oxford Bibliographies article “Acceptability Judgment”). In phonology, the situation is slightly different, as the target of the phonological studies seems less vague or less unreliable, although upon closer inspection, solely relying on the intuition-based phonological data could be demonstrably problematic. Most phonological data come in two kinds. One is alternation. A sound can change its shape depending on its phonological or morphological environment; for example, the English plural suffix is pronounced as [s] after cat, but as [z] after dog. The other is phonotactics: languages have restrictions on how the sounds can be arranged; for example, English allows a [tr] cluster but not a [tl] cluster word initially. At first sight, neither phonological alternations nor phonotactic patterns seem unreliable for theory construction. Indeed, much of phonological research, at least until the 1990s, had developed primarily based on nonexperimental data. The data are often gathered based on fieldwork research, dictionary searches, or intuitions provided by native speakers. Questions have been raised, however, regarding whether particular alternations or phonotactic patterns are indeed internalized in the minds of native speakers. One obvious, yet important, alternative hypothesis is that speakers of a particular language remember all the words in their lexicon, so that these patterns are also remembered on an item-by-item basis. In this view, phonological alternations do not need to be modeled as phonological processes, because all the words are stored in the mental lexicon, i.e., English speakers remember how to pronounce both cats and dogs, without an abstract phonological principle that governs the realizations of the plural allomorph. English speakers also know no words that begin with [tl], without necessarily referring to abstract phonological restrictions. One constructive response to this alternative is to test whether the sound patterns under question can be replicated with nonce words, thereby addressing whether the existing patterns generalize to new words, i.e., whether knowledge under question is generative. Much of the psycholinguistic research in phonology has thus been focused on how native speakers produce or respond to nonce words. To the extent that observed patterns are replicable with nonce words, another question that arises is whether the patterns observed in nonce words reflect grammatical knowledge or can be modeled via lexical analogy.

Resources

Cohn, et al. 2011 is a usable handbook on experimental phonology, now also known as laboratory phonology. Statistical methodology is essential in conducting and understanding experimental work. Introductory textbooks that are designed specifically for linguists are Baayen 2008 and Johnson 2008, both of which can be read without prior knowledge of statistics. The statistical program that is used in these textbooks is R (R Core Development Team 1993–), which is in fact used by many practicing linguists today. It is a free, open source software program for which many additional analytical packages are available. Macmillan and Creelman 2005 is an accessible introduction to Signal Detection Theory, which is essential to understanding and conducting speech perception experiments or any experiments that have to do with psychophysics. The book also contains discussions on various experimental paradigms. Praat (Boersma and Weenink 1999–) is free software that is extremely useful for phonetic analyses. Praat, like R, is scriptable. Experigen (Becker and Levine 2013) is a resource that allows researchers to perform phonological experiments online. A website maintained by John Krantz Psychological Research on the Net hosts many online psychological experiments, including linguistic experiments. This website is an efficient way to gather participants online.

• Baayen, Herald R. 2008. Analyzing linguistic data: A practical introduction to statistics using R. Cambridge, UK: Cambridge Univ. Press.

  DOI: 10.1017/CBO9780511801686

  This is an introductory textbook to statistical analyses written for linguists, using R. The book contains a whole chapter on linear-mixed modeling, which is becoming the common practice in experimental linguistics instead of more traditional ANOVA.

• Becker, Michael, and Jonathan Levine. 2013. Experigen: An online experiment platform.

  Experigen is a platform to perform various types of phonological experiments online.

• Boersma, Paul, and David Weenink. 1999–. Praat: Doing phonetics by computer.

  See also UCLA Praat script resources website online. Praat is free software that executes various phonetic analyses, including acoustic analyses and perception experiments. The software comes with a scripting function, which automates many repetitive processes. Many scripts are made available online by a number of people. It can also implement some statistical analyses and learning algorithms in Optimality Theory.

• Cohn, C. Abigail, Cécile Fougeron, and Marie K. Huffman, eds. 2011. The Oxford handbook of laboratory phonology. Oxford: Oxford Univ. Press.

  DOI: 10.1093/oxfordhb/9780199575039.001.0001

  This book is the most updated and comprehensive handbook on laboratory phonology, as of 2015. It covers a wide range of topics, including overviews of the issues addressed in the current field of laboratory phonology, current models of phonetics and phonology, various experimental methodologies, and statistical standards in the field.

• Johnson, Keith. 2008. Quantitative methods in linguistics. Malden, MA: Wiley-Blackwell.

  This is an excellent introduction to quantitative data analyses for linguists in general. It covers common statistical analysis techniques used in phonetics, psycholinguistics, sociolinguistics, historical linguistics, and even syntax.

• Krantz, H. John. Psychological research on the net.

  This website lists many online experiments. This website is an efficient platform to gather participants for online experimentation.

• Macmillan, Neil, and Douglas Creelman. 2005. Detection theory: A user’s guide. 2d ed. Mahwah, NJ: Lawrence Erlbaum.

  This book offers an accessible introduction to Signal Detection Theory, a theory of psychophysics that allows us to calculate a measure of sensitivity (d-prime) as well as a measure of bias (c). It also discusses several experimental paradigms for speech perception experiments. Some basic understanding of statistics is assumed.

• R Core Development Team. 1993–. R. Software.

  This free software allows us to perform various statistical analyses. Many packages are available for many different statistical, computational, phonetic, and psychophysical analyses. It also comes with a scripting function that automates repetitive processes, such as resampling and data processing.