Aksharas, alphasyllabaries, abugidas, alphabets and orthographic depth: Reflections on Rimzhim, Katz and Fowler (2014)

Abstract

We contend that, contrary to Rimzhim, Katz and Fowler (2014), consonants and vowels in the Brahmi-derived scripts are not “on a par”, and, therefore, that it is inaccurate to depict these scripts as alphabetic. Furthermore, we consider the popular terminology “alphasyllabic” to be misleading because these scripts are neither alphabetic nor syllabic. We argue on historical grounds that Brahmi-derived scripts (the script family known as Indic) are in a category of their own and merit a unique descriptor such as “abugida”. We also consider the authors’ concept of orthographic depth to be problematic outside the context of European alphabets because orthographic depth across the full spectrum of the world's writing systems is multi-dimensional rather than uni-dimensional. We suggest that at least 10 dimensions of orthographic depth (or complexity) are needed to capture writing system diversity. Finally, we briefly discuss some educational implications of classification and mis-classification of writing systems.

Keywords:

• Akshara
• Alphasyllabary
• Abugida
• Alphabet
• Orthographic depth

View correction statement:

Corrigendum

Notes

¹ Use of the term ‘alphabet’ varies from author to author and has been used to refer more narrowly or more broadly to writing systems that (1) represent all consonants and all vowels on an equal footing, as do RKF; (2) represent all consonants and all vowels but with vowels having only subordinate status; or (3) represent all consonants but with vowels either incompletely represented or entirely unrepresented. Since we share RKF's strict or ‘narrow’ definition (for historical reasons elaborated in Daniels, in preparation), we confine our arguments to this definition alone.

² For present purposes, we retain RKF's cover term ‘letter’, but note that although this term (or the term ‘grapheme’) is widely used (but not entirely unproblematic) in European alphabets, its applicability to many writing systems such as Brahmi-derived Indic aksharas, Chinese characters, Japanese kanji, or Mayan glyphs is questionable.

³ We prefer to retain the standard English technical terminology borrowed from Sanskrit long ago rather than the authors’ transliterations of Hindi words (akshar, Devnāgrī, etc.).

⁴ A simple explanation for the equivalent sizes of European consonants and vowels falls out of the hypothesis (Jeffery 1961, p. 22; Daniels, 1992b, 2007) that the Greek vowel letters derived from a misperception of several Phoenician letter names that began with initial laryngeal/pharyngeal phonemes that did not exist in Greek and were (fortuitously) misinterpreted to represent only the following vowel (e.g., /he/ => /e/ (later called epsilon) (see also Gnanadesikan's elaboration of this idea, 2009, p. 208–214).

⁵ For ‘denote’ used in this sense, see Swiggers 1984.

⁶ A possible explanation for this anomaly comes from the fact that /r/ tended to disappear in all Prakrits except Gandhārī Prakrit (Salomon, 2007, p. 93; Oberlies, 2007, p. 165). It seems likely that when Brahmi was devised, or adapted from Kharoṣṭhi, r was overlooked and only added later as a ‘diacritic’. It may not be a coincidence that ⟨r⟩ is the narrowest C akshara in Brahmi and, following it, Devanagari and other modern Indic scripts.

⁷ We thank one of our anonymous reviewers who called our attention to the existence in Hindi of the presumably onomatopoeic कौआ ⟨kau.ā⟩ ‘crow’, pl. कौए ⟨kau.e⟩, with non-initial vowel aksharas.

⁸ Whitney (1877) counted 10,000 phonemes of Sanskrit from a wide variety of texts and found that nearly half of all the vowels (1,978 out of 4,352) were /a/ (with a further 819 /ā/).

⁹ Referring to the fact that within a Chinese character, the ‘semantic’ and ‘phonetic’ components can be left and right, right and left, above and below, inside and outside, etc.

¹⁰ The exception here is Mishra and Stainthorp (2007), who found that only syllable but not phoneme awareness correlated with Oriya reading in Oriya-medium schools.

¹¹ Often referred to in the literature as ‘syllabic awareness’.

¹² We use the term ‘alphabetic’ to denote phoneme-level processing, although it seems a reasonable hypothesis that, in Indic scripts, consonants enjoy privileged status.

¹³ While many writing-systems scholars have abandoned the classification of the Phoenician and similar abjads as ‘consonantal syllabaries’ that was required by Gelb's ‘Principle of unidirectional development’, some have not—e.g., Venezky, 1999, p. 4 (‘not totally defendable’); Salomon, 2012, p. 126; Rayner, Pollatsek, Ashby, & Clifton, 2012. The ‘Principle’ provided the theoretical underpinning for the influential Gleitman and Rozin 1973; and its refutation has generally not trickled down to elementary linguistics textbooks (the latest edition of the most popular one, Fromkin, Rodman, & Hyams, 2014, p. 537, finds it necessary to mention that “[s]ometimes [the Semitic systems] are considered syllabaries because once the vowel is perceived, the consonantal letter seems to stand for a syllable”), or to popular treatments. Among reading researchers, lamentably, Gelb's views still predominate (see Share, 2014).

¹⁴ The Greek alphabet is the ancestor of all past and current evolved alphabets: predominantly the Roman and Cyrillic, but also national alphabets including the Armenian and Georgian, so it is legitimate to speak of the (one) alphabet.

¹⁵ This is brought out especially clearly in an expression by an author who fully accepts the typology: “The difference between the Indian alphasyllabaries and the true syllabaries is clear enough …” (Sproat, 2010, p. 61).

¹⁶ As is clarified by Bright (2000), ‘alphasyllabary’ is meant to refer to the formal property of vowels other than /a/ being denoted by matras, and ‘abugida’ is meant to refer to the functional property of /a/ being inherent in the basic symbol. The test case proved to be the ’Phags pa script, which is an abugida but not an alphasyllabary.

¹⁷ It is probably not a coincidence that the first inscription using vocalised Ethiopic script is also the first inscription in which King Ezana (mid fourth century ce) professes Christianity. It is quite likely that his conversion was effectuated by missionaries who had sailed from the Christian community of the western India coast (Daniels, 1992a).

¹⁸ To be sure, by dint of population a name based in some aspect of Indian culture might have been more suitable, but ‘kakhāgighī’ or some such is less than inspiring. Now that the order of both consonants and vowels used by Kharoṣṭhi scribes has become known (Salomon, 2006), a name like ‘arepiconu’ might have suggested itself.

¹⁹ The terms ‘alphabet’ and ‘phoneme’ cannot be taken as simple pre-theoretic givens. In the scheme adopted in Daniels (1990), an alphabet ‘assigns a segment (consonant or vowel) to each symbol’. Thus Greek (and its descendants), Korean, and fully pointed Hebrew and Arabic are all written with alphabets. A phoneme is “the smallest unit of sound in a language capable of causing a difference of meaning” (Barry 2006, p. 345). ‘Grapheme’, by contrast, has had so many different interpretations that in writing systems theory it is meaningless (Daniels, 1991).

²⁰ Anecdotal evidence of this sort of depth in Spanish—a language conventionally placed at the ‘very shallow’ end of the single dimension of ‘orthographic depth’—is provided by Morrison (1945), who observes the word ‘I’ spelled both ⟨yo⟩ and ⟨llo⟩ and offers a number of other instances of uncertainty as to the correct spelling of various phonemes. Valle-Arroyo (1990) confirms these observations in a systematic investigation of spelling errors in Spanish.

²¹ It is clear that a syllabary would not serve English well, with its possible CCCVCCC syllables (e.g., ⟨strengths⟩ /streŋθs/ = [streŋkθs]!), whereas Hawai‘ian, with CV syllables comprising just 7 or 8 consonants and 5 vowels, is an ideal language for a syllabography. The abugida was devised for a language with CV syllables and a great preponderance of /a/ over its other four vowels.