I have had an interesting voyage to get to this book, not least because I kept assuming it had the word 'stress' somewhere in the title. It is the Book of the Database: the StressTyp database, to be precise. (You see why I thought the title contained the words "Typological survey of stress".)
I do not intend to cover all 10 chapters on different language families this week, but I expect I will keep coming back to it. This week, we look at Chapter 1, an overview of word accents.
Chapter 1: Word accent: Terms, typologies and theories
The authors make a distinction between accent and accent cues, since the cues vary massively between languages, but the function of accent does not.For example, Tokyo Japanese has pitch accent, in which the accented syllable, and those preceding it, have a higher tone. English, however, has a stress accent, where stress is a complicated bundle of cues that includes duration, loudness, pitch, and vowel type.
Cues for accented syllables include:
- Longer
- Balanced spectral tilt (i.e. ratio of loudness of high and low frequencies)
- Louder
- Higher pitched
- Greater precision or other phonetic traits (e.g. English vowels never reduce to ə, 'uh' in accented syllables; e.g. hippopotamus has aspiration ('h') after the stressed /p/, but not after the unstressed one.)
- Phonotactics (which sounds can occur in a syllable - more complex syllable types are often accented)
- Site of morphological process (combining meaningful parts of words - e.g. English infixing: in abso-***-lutely, 'lute' is accented, in fan-***-tastic, 'tas' is accented)
- Anchor for intonational tones (intonation is the rise and fall of the whole phrase - such as rising at the end of a question. How intonation is assigned can be entire lecture course in itself, so I'll stop there for now.)
Types of accent systems
There are 28 quantity sensitive systems (16 bounded and 12 unbounded), and 8 quantity insensitive systems:| Weight sensitive? | First or last heavy? | First or last other? | Bounded? | Right or left bounded? | Extrametricality? | Direction of extrametricality? |
|---|---|---|---|---|---|---|
| No | - | First | Yes | Left | No | - |
| No | - | First | Yes | Left | Yes | - |
| No | - | First | Yes | Right | No | - |
| No | - | First | Yes | Right | Yes | - |
| No | - | Last | Yes | Left | No | - |
| No | - | Last | Yes | Left | Yes | - |
| No | - | Last | Yes | Right | No | - |
| No | - | Last | Yes | Right | Yes | - |
| Yes | First | First | No | - | No | - |
| Yes | First | First | No | - | Yes | Left |
| Yes | First | First | No | - | Yes | Right |
| Yes | First | First | Yes | Left | No | - |
| Yes | First | First | Yes | Left | Yes | - |
| Yes | First | First | Yes | Right | No | - |
| Yes | First | First | Yes | Right | Yes | - |
| Yes | First | Last | No | - | No | - |
| Yes | First | Last | No | - | Yes | Left |
| Yes | First | Last | No | - | Yes | Right |
| Yes | First | Last | Yes | Left | No | - |
| Yes | First | Last | Yes | Left | Yes | - |
| Yes | First | Last | Yes | Right | No | - |
| Yes | First | Last | Yes | Right | Yes | - |
| Yes | Last | First | No | - | No | - |
| Yes | Last | First | No | - | Yes | Left |
| Yes | Last | First | No | - | Yes | Right |
| Yes | Last | First | Yes | Left | No | - |
| Yes | Last | First | Yes | Left | Yes | - |
| Yes | Last | First | Yes | Right | No | - |
| Yes | Last | First | Yes | Right | Yes | - |
| Yes | Last | Last | No | - | No | - |
| Yes | Last | Last | No | - | Yes | Left |
| Yes | Last | Last | No | - | Yes | Right |
| Yes | Last | Last | Yes | Left | No | - |
| Yes | Last | Last | Yes | Left | Yes | - |
| Yes | Last | Last | Yes | Right | No | - |
| Yes | Last | Last | Yes | Right | Yes | - |
In other words, for all languages, accent either defaults to the first or last syllable of some domain.
If the domain is bounded (no more than 2 syllables from one edge of the word), that boundary must either be at the left or right edge of the word. If the language is not weight sensitive, then the domain must be bounded.
If the language is weight sensitive, the accent goes to a heavy syllable if possible; either the first or last heavy syllable of the domain. In the absence of a heavy syllable, it is the defult, which may not be the same direction.
There can be an 'extrametrical' syllable that is outside the domain - in English, this is the last syllable in the word, so we have antepenultimate stress (3 syllables from the end). For bounded systems, the direction of this extrametrical syllable is predictable; for unbounded systems it's an extra parameter that can vary.
There are also count systems, where accent is given to e.g. the rightmost even syllable (counting from the left), or leftmost odd, etc. These seem to exist mostly to mess with linguists, preventing us from deciding whether rhythm is universally derived from accent or vice versa. I expect it will turn out that there is no universal rule.
There are also, just to keep life interesting, lexical accent languages - where it is completely unpredictable, and used to change the meaning of the word. English does have some lexical accent words (to record vs a record) but it's not a lexical accent language. Only a small percentage of words are lexically accented, and trying to categorise languages by whether they have a minority of words with a different accent pattern makes things too complicated.
References
Van Der Hulst, Harry, Rob Goedemans, and Ellen Van Zanten, eds. (2010) A survey of word accentual patterns in the languages of the world Walter de Gruyter
http://www.isca-speech.org/archive_open/archive_papers/prosody_2001/prsr_010.pdf