>John and I talked this morning about strong and weak syllables within
>feet: in particular, the issue of whether you can have more than one
>strong syllable within a foot. We have no firm conclusions yet, and we
>think we need a bit more help from Jill first.
>
This is exactly the question I was trying to address when we met, when I
asked you how many feet there were in "YorkTalk"! To an extent it's an
empirical issue that I would like to get a handle on. It's important in
defining (a) pitch accent domains and (b) timing, and there may be
something of a mismatch between the optimal units for pitch and timing
purposes.
>So -- can you tell us how phrases like
> blackboard duster
> elevator operator
>fall into AGs and feet for you?
I'll try. But this is just the area where ambiguity comes in. If we
assume a rendering of your first example where it is given compound stress,
i.e. primary stress on "black", and falling intonation (i.e. H* on "black",
followed by L- L%), both the following analyses are possible:
(1) {AG[F blackboard][F duster]}
s s s w
(2) {AG[F black][F board][F duster]}
s s s w
The advantage of (1) is that it captures the temporal adjustments that you
expect inside an Abercrombian-style foot, and could be used to motivate
quasi-isochronous rhythm. A disadvantage must be that if you allow strong
syllables to be foot-internal, you lose simplicity and multiply the number
of possible foot structures (s s w, s s w w ??) -- consider
(3) {AG[F black board e][F raser]}
s s w s w
Rhythmically the two sentences are very similar and I'm not sure I'd want
to say they have different numbers of feet.
The advantage of (2) is that it consistently promotes strong syllables to
the status of foot, regardless of perceived rhythmic stress. The temporal
adjustments would have to be motivated differently -- some effect from
having three adjacent monosyllabic feet, or something. This is its main
disadvantage. A further advantage may arise if it turns out that the
boundary between "black" and "board" blocks the spread of pitch accent
alignment from the preceding syllable. I don't know the answer to this --
needs testing -- but I would hypothesise, based partly on intuition, partly
on the literature, that a simple H* (NOT L*+H, or Rise-Fall, but simple
Fall) pitch accent cannot be shifted rightwards into the following syllable
"board", whereas this could happen quite readily in a case like "blackberry
bush". The realisation of the fall in pitch itself can be spread over the
weak syllables in "blackberry" in a way it can't over the "board" in
"blackboard".
In your second example, pronounced the same way (as a compound), people
seem most readily to come up with an analysis like:
(4) {AG[F ele][F vator][F ope][F rator]}
s w s w s w s w
If a weak syllable intervenes between two strong ones, it pushes you to
analyse each strong syll as foot-initial. In temporal terms, you may be
occupying little more time than in (1). The feet themselves alternate
strong - weak, reflecting a metrical structure (in this case each word
could belong to a "superfoot", a left-headed unit branching into strong and
weak feet). This would make my pitch accent domain = the head of a
superfoot. I think. I have been reluctant to propose this as an
additional layer in the hierarchy because it complicates things quite a
lot, but if you think it is a valid piece of structure which could give you
interesting and useful results it may be worth considering. Another
possibility is to capture relative timing using some kind of metrical grid
but I haven't been thinking in those terms up to now.
(NB I'd wondered whether the heavy ~ light distinction could be used to
solve problems like "blackboard", but I don't think so -- "brickbat" would
behave in a similar way.)
Sorry if this message muddies the waters rather than clarifying anything.
I hope you can see what I'm getting at, even though I'm not proposing a
solution. I can work with either analysis, provided the relevant
information is stored somewhere. Lexical secondary stresses which precede
the primary stress have to be treated differently again, since they have
the potential to be accented (AG-initial), which post-primary ones don't.
(e.g. "indivisibility").
Oh joy. Can we agree on a working solution? Can't think further tonight.
Your further thoughts are welcome!
>our conception of the foot in YorkTalk was
>very word-based; above the word I don't think we really tested any fancy
>stuff like this. (Or do you feel this is not something we need to worry
>about at the moment?)
>
I think we have to decide what we'll do in these cases -- they're pretty
central.
Jill