Consonant/vowel asymmetries in letter position coding during normal reading: Evidence from parafoveal previews in Thai

Abstract

Studies have revealed that consonants and vowels serve different roles during linguistic processing. Masked transposed-letter priming effects (i.e., faster word-identification times for words preceded by a transposed-letter than substitution-letter prime) occur for consonants but not for vowels in lexical decision (Perea & Lupker, 2004). Potential differences in letter position coding for consonants and vowels during silent normal reading were investigated in Thai using the boundary paradigm (Rayner, 1975). Thai has a distinctive alphabetic script with vowels taking a relatively subsidiary role in relation to consonants. Parafoveal processing of nonadjacent transposed-letter effects involving consonants and vowels was examined. Results for gaze durations revealed a transposition effect involving consonants but not vowels—thus extending previous findings with the masked priming technique but in a more ecological setting. Similar differential effects for consonants and vowels for first and single fixations were not found. An explanation is that consonants and vowels are not differentiated at this initial low level stage of processing (Johnson, 2007; Perea & Acha, 2009); it is only later in processing (as measured by gaze durations) that consonant/vowel status comes into play. Results support the claim that there are some fundamental processing asymmetries between vowels and consonants in normal reading.

Keywords:

• Boundary paradigm
• Letter coding
• Thai
• Transposed letters
• Visual-word recognition

Acknowledgments

We would like to thank Dr Sudaporn Luksaneeyanawin, Dr Wirote Aroonmanakun, and Dr Theeraporn Ratitamkul, Center for Research in Speech and Language Processing (CRSLP) and the Linguistics Department, Chulalongkorn University, Bangkok, for advice and assistance as well as use of the laboratory facilities. Also thanks to Mr Chalong Saengsirivijam for assistance with participant recruitment.

Notes

¹As a reviewer pointed out, the presence of a transposed-letter effect in gaze durations but not in first-fixation durations may be due to a differential refixation probability or to differences in the second fixation duration. Given that refixation probability is treated as a Boolean variable [0 vs. 1 for each observation], a linear mixed model with was fit by the Laplace approximation with word type (vowel, consonant) and preview (transposed-letter, substituted-letter) as fixed factors and participants and items as random slopes. (Other models with different complexity of random effects yielded essentially the same findings.) Results showed an interaction between preview word type, z=−2.18, p=.02. For consonants, the refixation probability was lower in the transposed-letter condition than in the replacement-letter condition (.31 vs. .36), z=3.05, p=.002, and the parallel difference was not close to significance for vowels (.32 vs. .29), z<1. There were not enough data points for a reliable analysis of the second fixation duration analysis, however.

²One important issue for further research is to examine how letter identity (e.g., via a delayed letter paradigm; Lee et al., 2001; see also Carreiras, Gillon-Dowens, et al., 2009) and letter position (via a transposed-letter manipulation) interact during parafoveal processing in reading (see Vergara-Martínez, Perea, Marín, & Carreiras, 2011, for evidence using masked priming lexical decision while recording ERPs).