Lars, My greater issue with your type classification has to do with my disagreement about how you have defined some of the types. I'm not disagreeing that there are functionally definable types for scripts, and that it is useful to divide up the scripts into those categories. But I disagree with some of the details of your definitions and about what got classified as what type.
Issue 1. Abugida and Alphasyllabary You've defined abugida as "a type of writing system whose basic characters denote consonants followed by a particular vowel, and in which diacritics denote the other vowels." You then point to Ethiopic, Cree (= UCAS in Unicode), and Tengwar as having that characteristic. But cf. Peter Daniels' definition of abugida: "In an abugida, each character denotes a consonant accompanied by a specific vowel, and the other vowels are denoted by a consistant modification of the consonant symbols, as in Indic scripts." In other words, Daniels is defining an abugida in such a way as to include all the Brahmi-derived scripts, which you have *excluded* from your definition of an abugida. I think Daniels' definition makes more sense, and we are planning to make use of it in the next edition of the Unicode Standard. Ethiopic is an interesting case. It is the original "abugida" in grammatology, but it started out as a Semitic-derived abjad, and in many ways it is more convenient now to analyze it as a featural syllabary. That is how we ended up encoding it in Unicode, how it is generally presented, and how many people think and interact with it. While the various flags and loops associated with the vowel ranks in Ethiopic obviously have a featural consistency, they don't have the kind of independent existence that the Indic matras typically have; instead, partly because of the complex placement rules for the flags and loops, depending on the shapes of the consonant bases, in Ethiopic it makes more sense to just see each combination as a syllabic unit as a whole, with "hints" about the vowel. The Canadian aboriginal syllabics, derived from a shorthand system, also should be analyzed as a featural syllabary, in my opinion. So I think the right thing to do here is to move your instances of "abugida" into featural syllabaries, and redefine your "alphasyllabary" as "abugida", more in line with Daniels' definition. Also, regarding your current class of "alphasyllabary", I don't think the critical issue here is linear order "that is congruent with their temporal order in speech." While it is true that many of the Brahmi-derived scripts now have reordrant or surroundrant vowel matras, for various reasons related to the writing history of the scripts (vowel marks on top that "migrated" around to the left of their consonant bases, mostly), there are also simplified descendants that would be hard-put to fit your definition. Take Tagalog, for example. It has two vowel marks, a dot above for /i/ and a dot below for /u/. On this basis alone, it would be hard to distinguish it from an abjad (consonant writing). The crucial difference, which tips it over to an abugida, along with the other Brahmi-derived scripts, is that the consonant letters all have the inherent vowel /a/. So /ka/ is just a single letter, /ki/ is /ka/ + /i/, /ku/ is /ka/ + /u/. And there is a killer (a different dot below), that is used to suppress the inherent /a/. *That* is the pattern which distinguishes all the Brahmi-derived abugidas, simple or complex. Issue 2. Featural script You define this as "a type of writing system whose characters denote phonetic features," and give one example: Hangul. Since Hangul characters (either considering the syllables as units or the jamo as units) don't "denote phonetic features" per se, even Hangul wouldn't fit that definition. The way Daniels defines it is "In a featural system, like Korean or 'phonotypic' shorthand, the shapes of the characters correlate with distinctive features of the segments of the language." Taken as a syllabary, Korean fits that bill, since the parts of the syllable (the jamo) do *correlate* with distinctive segmental features of the language. Furthermore, the single versus double jamo spelling for initial consonants correlates with the systematic manner distinction for consonants in Korean. Looked at this way, UCAS obviously fits as well, since the rotations and dot additions are systematically correlated with segmental distinctions in the various language phonologies. And if Ethiopic is treated as a syllabary, it, too, can be seen as a featural system. The distinction from non-featural syllabaries is that in the non-featural systems (Kana, Yi, ...) you cannot point to any graphic part of any of the symbols, and parse it off as having any systematic correlation with segments of the sound system. (Of course nothing is black and white here -- the voicing marks of Kana obviously are featural, and the tone marks for Yi are also featural.) --Ken
