A comparative quantitative analysis of Greek orthographic transparency

ELENI L. VLAHOU 0 0 Institute for Language and Speech Processing , Maroussi, Greece and University of Crete , Rethymno, Greece Orthographic transparency refers to the systematicity in the mapping between orthographic letter sequences and phonological phoneme sequences in both directions, for reading and spelling. Measures of transparency previously used in the analysis of orthographies of other languages include regularity, consistency, and entropy. However, previous reports have typically been hampered by severe restrictions, such as using only monosyllables or only word-initial phonemes. Greek is sufficiently transparent to allow complete sequential alignment between graphemes and phonemes, therefore permitting full analyses at both letter and grapheme levels, using every word in its entirety. Here, we report multiple alternative measures of transparency, using both type and token counts, and compare these with estimates for other languages. We discuss the problems stemming from restricted analysis sets and the implications for psycholinguistic experimentation and computational modeling of reading and spelling. - Alphabetic orthographies differ in their degree of transparencythat is, in the systematicity of the mapping between letter sequences and phoneme sequences. Inconsistencies in the soundspelling mappings arise when single orthographic units have multiple pronunciations or single phonological units have multiple spellings. Quantitative assessments of such ambiguities have been carried out in several languages with alphabetic orthographic systems, both in the feedforward directionthat is, from orthography to phonology, as needed for reading aloud printed wordsand in the feedback direction, from phonology to orthography, as needed for spelling (e.g., Borgwaldt, Hellwig, & De Groot, 2004, 2005; Treiman, Mullennix, Bijeljac-Babic, & Richmond-Welty, 1995; Ziegler, Jacobs, & Stone, 1996; Ziegler, Stone, & Jacobs, 1997). The study of orthographic transparency is important for theoretical and practical reasons, since ambiguous mappings have been found to affect reading and spelling performance (Spencer, 2007, in press). For example, feedforward-inconsistent orthographic units (i.e., letter sequences that can be pronounced in more than one way) slow down word naming (Burani, Barca, & Ellis, 2006; Jared, 2002; Treiman et al., 1995), whereas ambiguities in the feedback direction affect spelling performance (Burt & Blackwell, 2008; Lt, Peereman, & Fayol, 2008). A counterintuitive finding is that feedback inconsistency also affects reading. That is, words with predictable pronunciation but unpredictable spelling are read and recognized more slowly than words with predictable spelling (Grainger & Ziegler, 2008; McKague, Davis, Pratt, & Johnston, 2008). However, in a review of studies on feedback inconsistency effects, Kessler, Treiman, and Mullennix (2008) pointed out a number of methodological shortcomings that need to be addressed before a final conclusion can be reached. To pursue these issues in additional languages, detailed quantification of orthographic consistency in both directions is necessary. Ambiguity does not affect all sublexical units equally. In more opaque orthographies, smaller units tend to be less consistent than larger units (Ziegler & Goswami, 2005). For example, graphemes1 are less consistent than orthographic bodies (spellings of a syllabic rimei.e., of the nuclear vowel and any consonants that follow it) in English monosyllables (Treiman et al., 1995). There is thus a functional pressure for readers to develop both small-unit and large-unit recoding strategies. As the grain size grows, the number of distinct orthographic units rises. This granularity problem is more pervasive in opaque orthographies, whereas readers of more transparent orthographies can focus on finer grain sizes (Ziegler & Goswami, 2005). This assertion remains to be substantiated with specific estimates of the transparency and granularity of specific orthographic systems. In the present work, we address two issues. First, we provide a systematic quantitative exposition of transparency in the Greek orthography. The goals of this presentation are to support researchers working on Greek with information relevant to stimulus selection and experimental design and to provide researchers working in other orthographies with a comparison reference. Second, we use Greek as a test case to examine and evaluate a multitude of approaches to the quantification of orthographic transparency. By comparing and contrasting various alternative methods that have been proposed in the literature, we are able to test certain assumptions underlying them and indicate weaknesses that may limit their applicability. Thus, our analysis suggests, directly or indirectly, ways in which the quantification of orthographic transparency may be improved in future research. In the following sections, we first will present existing approaches to the quantification of orthographic transparency in different alphabetic writing systems, discussing their methodological strengths and weaknesses. We then will introduce the most important aspects of the Greek orthography. We will report analyses of the transparency of Greek orthography, comparing and contrasting calculations of regularity and consistency, on the basis of a word-form list from a representative corpus of contemporary written Greek texts. We will identify the grapheme phoneme level of analysis as the most appropriate for Greek, and we will present statistics relating individual phonemes and graphemes, as well as an ordered set of rules maximally capturing graphophonemic transcription. Finally, we will discuss the implications of our findings for cross-linguistic evaluation of orthographic transparency, for learning to read and write in Greek, and for modeling reading Greek. Quantitative Indices of Orthographic Transparency Regularity. The regularity approach assumes the theoretical position that there are regular mappings, governed by symbolic transcription rules, and irregular mappings, which violate the rules. Under this framework, the problem consists in the specification of a set of rules that relate individual graphemes to the corresponding phonemes (for the feedforward direction, or the reverse for the feedback direction). In cases in which the mapping deviates from one-to-onefor example, when a single grapheme can have multiple pronunciationsthe most frequent mapping is considered regular, and the others irregular. Regular words are words whose pronunciation or spelling is correctly produced by the graphemephoneme correspondence rules of a language, whereas irregular or exception words are words whose pronunciation or spelling cannot be predicted from these rules (Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001; Ziegler, Perry, & Coltheart, 2003). Regularity is thus conceptualized as a categorical distinction (Zevin & Seidenberg, 2006). Zi (...truncated)