<<   >>   TOC

5. Interactions of H tone and depression

In the last section, I have provided evidence for assuming a segmental feature as the defining characteristic of depressor consonants. The initial independence of H tones and depression thus finds a simple explanation: phonological tones are abstract elements that are manipulated with respect to abstract host units (TBUs, most likely moras in Zulu). The depressor feature, an abstract element of different categorial status, is manipulated with respect to a different set of host units (namely the laryngeal nodes), as indicated in (21).

In categorial terms, H and [lax vf] are completely independent and may ultimately target host units that correspond to the same skeletal slot (like x2 in 21). A conflict may arise when H is translated down into segmental features for articulation, e.g. [tense vf] as suggested above. (21) would then yield (22) in its relevant aspects:

At x2, two antagonistic articulatory features are now linked to the same node. H tone shift, in the form of delinking and spreading [tense vf], could therefore be regarded as a conflict resolution mechanism:

I assume that the spreading behavior of [tense vf] is the same as the one of [lax vf], since both operate on the same tier. Thus, [tense vf] will establish a rather localized spreading domain, where it provides the necessary acoustic cues to be reinterpreted as a H tone.

If x3 carries relevant laryngeal specification, i.e. if x3 is a depressor consonant, spreading of [tense vf] would be blocked. A lowered H tone with an onglide results at x2, as the contour representation in (24) indicates.

If the proposed representation is correct, an otherwise puzzling characteristic of depressor induced tone patterns finds a straightforward explanation: while Bantu H tones typically are highly mobile and can surface far from their origin, depressor induced H shift is a strictly local phenomenon. The reason for this change in behavior lies in the fact that H shift occurs not within the tonal module but within the segmental feature geometry. The Zulu constraint against contour representations on the tonal level does not hold in the segmental geometry, which explains why depression may lead to monomoraic surface contours even if underlying tonal contours typically require bimoraic manifestation.

<<   >>   TOC