25.10.2012 Views

Laurie Bauer - WordPress.com — Get a Free Blog Here

Laurie Bauer - WordPress.com — Get a Free Blog Here

Laurie Bauer - WordPress.com — Get a Free Blog Here

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

65 BINARITY<br />

Distinctive features<br />

Within the Prague School of linguistics in the 1930s, binarity in phonological<br />

distinctive features such as [� voiced] or [� back] is related to markedness.<br />

There is a marked value for every feature (the � value) and an unmarked value.<br />

In early markedness theory, the marked value is given to something which is<br />

physically present in some environment (e.g. nasalisation in [m] or [a~]), but<br />

absent elsewhere. However, as markedness theories became more sophisticated,<br />

it has be<strong>com</strong>e harder to maintain the principle that for any feature the same<br />

value is marked or unmarked in all environments. For example, ‘voiced’ seems<br />

to be the normal state for vowels (that is, we would want to say that [� voice]<br />

is the unmarked value for vowels), but the opposite is true for obstruents (the<br />

norm is for them to be voiceless).<br />

In a deservedly widely ignored paper, Halle (1957) appears to argue that<br />

because binarity works in many places it should be used everywhere. This<br />

seems to disregard the received wisdom of the period, inherited from the<br />

Prague School, that there are several types of distinctive feature only some of<br />

which are binary in nature. Thus we find scholars like Ladefoged (1971) reverting<br />

to features with many values (multinary features) for things like vowel<br />

height, since binary features (a) do not allow easily for five distinctive vowel<br />

heights and (b) do not allow for simple statements of rules which raise or lower<br />

vowels by a single step. We also find alternative approaches, such as the use of<br />

unary features (features with just one value; that is, the feature is either present<br />

or not) within Dependency Phonology. Even within Generative Phonology,<br />

once feature values are filled in by general rules, problems can start to arise with<br />

ostensibly binary features actually having three contrastive values (Stanley<br />

1967). For instance, if we suppose that a feature whose value has not yet been<br />

assigned is marked as u,we could imagine the three rules in (2) which would in<br />

effect be using u as a third contrastive value of the feature.<br />

(2) [u voice] → [� sonorant]<br />

[– voice] → [– sonorant]<br />

[� voice] → [– sonorant]<br />

The question of binarity has never been definitively settled within phonology,<br />

although modern versions of feature geometry seem to ignore it. It seems<br />

safe to conclude that it is at least not universally accepted.<br />

Binarity has not been questioned to the same extent in semantic features,<br />

perhaps because semantic features themselves have rather fallen out of fashion.<br />

Morphosyntactic features, such as those used to mark tense and case in abstract<br />

structures, lost an absolute requirement for binarity at least as early as Gazdar<br />

et al. (1985), where it is argued (p. 22) that ‘a feature value is either an atomic

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!