Towards a typology of phonemic scripts

ABSTRACT

This paper focuses on those scripts whose basic grain size is the phoneme or phonological segment, arguing for both the essential unity and the diversity of such scripts. On the one hand, these scripts encode individual phonemes, unlike syllabaries, and should therefore be recognised as constituting a class. On the other hand, they vary in which vowels they represent, if any, and in how the relationship between vowels and consonants is encoded. These dimensions of which segments are represented and how those segments are arranged to form larger phonological structures vary independently of each other. Furthermore, some phonemic scripts also encode featural or moraic information. Thus, a short list of simple one-word names, like alphabet, abjad and abugida, does not capture the full range of segmental scripts or the relationships between them. A typology using more descriptive terms is presented.

KEYWORDS:

• Script typology
• Alphabets
• Alphasyllabaries
• Abjads
• Abugidas
• Phonemic writing systems

Acknowledgements

Thanks to the Neubauer Collegium for Culture and Society at the University of Chicago for hosting me in the final stages of writing this paper. Thanks also to Peter Daniels, David Share and the audience at Stony Brook University’s linguistics department colloquium for comments on a talk based on an earlier version of this paper. My thanks to the colleague who initially urged me to consider using the term segmentary. The comments of two anonymous reviewers substantially improved the paper, but all remaining errors and infelicities are my own.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 Though not central to the present work, the choice is not arbitrary. Phonographic can easily be understood to refer to writing which is related to a language’s phonology. By contrast, cenemic is a remarkably opaque term, and phonetic wrongly relates writing to a language’s phonetics rather than to the more abstract, functional level of its phonology.

2 Syllabaries have not been without their controversy, however, centring around whether they represent syllables or moras (Gnanadesikan, 2011; Poser, 1992; Rogers, 2005).

3 Gelb’s classification of the Semitic consonantal scripts as syllabaries is a misstep for which he has been roundly condemned (see Daniels, 1990; Voegelin & Voegelin, 1961; and citations therein; Daniels, 2001; Faber, 1992; Rogers, 2005; Trigger, 2004; among others).

4 The letter which stood for the syllabic consonant /r̩/ in Sanskrit is read as /ri/ or /ru/ in the modern languages. The letter which stood for the long syllabic consonant /r̩ː/ in Sanskrit is not used in the modern languages.

5 This view is strongly articulated by Havelock (1982, 1999), who considers the invention of the Greek alphabet as a dramatic leap forward in intellectual history, and the Greek alphabet as qualitatively different from consonantal scripts such as its predecessor, Phoenician. He even ascribes the invention of literacy to the Greeks. It should in fairness be stated, however, that many grammatologists who accept the narrow definition of alphabet also reject Havelock’s Eurocentric view. For example, Daniels (1996b) roundly condemns it while simultaneously (Daniels, 1996a) arguing for a narrow definition of alphabet.

6 It is also odd that the term abugida is taken from the Ethiopic script, when the behaviours picked out by the term first arose in and are far more widely dispersed in South Asia.

7 Vocalisation refers to the optional practice of using diacritics to indicate short vowels (and consonant gemination) in Arabic. It is not usually used except in pedagogical or religious contexts.

8 Fully vocalised texts in Arabic, of course, take things to the logical extreme of representing every vowel of the language. However, in such uses it is uncontroversial that the script is not behaving as an abjad but rather as what Daniels considers an alphabet (since all vowels are written). Presumably, Bright would consider it an alphasyllabary, since the vowels are written as diacritics on the consonants. However, see note 11.

9 There are actually two vowel-support letters in Tibetan, but they represent a difference in the vowel’s tone rather than a specific difference in vowel quality.

10 DeFrancis (1989) and Taylor and Taylor (2014) also point out that the script is not learnt as a featural system. However, I agree with Sproat (2000) that this point is irrelevant to the question of how a script should be structurally classified—though very relevant indeed to the optimal pedagogical methods designed for it. Given that the structures of spoken language are there whether speakers attend to them consciously or not (and usually they don’t), the absence of explicit attention being paid to some aspect of orthographic structure also does not render the structure absent or irrelevant.

11 The fully vocalised version of the Arabic script presents an interesting wrinkle, in that it represents short vowels with diacritics, but long vowels with both the in-line characters used in ‘unvocalised’ texts and the short-vowel diacritics. Thus, a long vowel is a short vowel (a diacritic) plus something else (an in-line character). This is phonologically accurate, in that in CV phonology, the short vowel receives one V slot and a long vowel receives two. Thus, in fully vocalised Arabic it is the first V slot, which constitutes the head of the syllable peak, that is given a special representation as a diacritic rather than the full VV peak.

12 Ideally a term that captures the sense of ‘all vowels but one’ would be better than ‘mostly vowelled’, but none has suggested itself.

13 An interesting case is Pahawh Hmong, in which there is a default, unwritten onset, which consists of the consonant /k/ (Ratliff, 1996). However, that writing system is an onset-rhyme system, not a segmentary.

14 The modern Greek alphabet (Threatte, 1996), for example, is the only writing system that comes to mind that regularly represents word stress.