Sarah Hawkins (sh110@cam.ac.uk)
Tue, 16 Nov 1999 10:20:32 +0000 (GMT)
I seem to have sent the following message ONLY to the prosynth archive,
which probably no one has checked.
Many apologies.
As an update, Richard is going to look into the accents of the 3 Ss who
are NOT Scottish in the Map Task corpus. (Ref is L&S 1991. 34. 351-366,
paper with 13 or so authors, first being Anderson).
there will be one more email coming that I made the same mistake with.
Sarah
---------- Forwarded message ----------
Date: Fri, 12 Nov 1999 12:24:04 +0000 (GMT)
From: Sarah Hawkins <sh110@cam.ac.uk>
To: prosynth archive only <prosynth_archives@phon.ucl.ac.uk>
Subject: LDC and other dbases
Dear All
I think I might have missed something, in that (re my enquiry re the LDC,
sent only to some of you) Richard has directed me to look at the map
task, available from the LDC. But it’s nearly all in Scottish English.
Does anyone know if there’s any SSBE stuff at all in it?
I know John is doing database stuff, but in trying to get my budget
organised, I wonder if we could establish in the next few days what we’re
trying to do re databases? Here’s what I think I know. I have looked at
the LDC web page, admittedly rather quickly. My feeling is confirmed that
LDC is not much different from the way it was when we rejected it as a
potential source of info last time: i.e. all items in "English" are in
American English, or else likely to be foreign-accented English (e.g.
airport communications). An (the?) exception is CELEX, but as far as I can
tell there’s no speech in it.
However, for what it’s worth, this is what I’ve found on LDC:
LDC93S12: HCRC Map Task Corpus:
The 64 speakers were all students at the University of Glasgow, 61 of them
being native Scots.
Therefore, even for function words, I’d have thought it was of extremely
limited application to ProSynth.
LDC99S78: SUSAS
American and probably a lot of southerners, (so not even GA) but does do
stress, emotion, with some simulated and actual stress-inducing tasks done
concurrently with speaking: 35 words only? Not continuous speech? But lots
of speakers.
OTHER THOUGHTS
I. OTHER DBASES
A. Other than the obvious things like MARSEC and the Oxford
acoustic-phonetic database (does anyone know what it’s like?),
B. it occured to me that we might consider Francis and Esther’s corpus of
intonation (in progress). For the most part, not useful, because not SSBE,
but they do have 16 Cambridge teenagers, who are sort of SSBE, doing 5
tasks, including reading, spontaneous speech, and a map task. Esther will
give me details soon. I will pass on the electronic ones immediately, and
give the paper ones on Friday 19. It’s free.
C. I have some sort-of spontaneous conversational speech that I collected
with Paul Warren, for given-new research. (Work published in JPhon, 1994.)
3 people (all SSBE) are describing and generally expressing opinions about
pictures, in the course of "conversations" with Paul and me. Presumably
contains lots of function words, and every sentence has intonation.
Includes repeated words, clarifications, etc etc but I doubt it’s any good
for seriously controlled stuff like more resonance work. On the other
hand, the words that are repeated conform to certain requirements like
controlled frequency, mainly trochees, and with obstruent in medial
position; no minimal pairs (see JPhon paper). Duration per person, about 1
hour or somewhat more. Quality of recording is moderate (done with
excellent equipment, but in a quiet office, not a sound proof room, and
little attempt to control distance from mike etc.) You can look at the
transcripts, listen to sample tapes, etc, if you’d like to, if you ask me
to get them for Friday. They are stored away, due to lab move, so I will
need several days’ notice.
II. COPY OF EARLIER EMAIL TO JOHN RE CAMBRIDGE’S DBASE NEEDS
Here is what I think Cam will need (but feel free to suggest amendments):
1. The same dbase(s) as everyone else for general work.
2. If that does NOT include the following, then we would also need:
a. speech in which the same words appear in clear and fast styles, as
well as given and new. Preferably in similar rhythmic and segmental
contexts.
b. I would sugggest that at least some of 2a includes something that
is easy to remember, *possibly* proper names, but more likely something
more predictable (adages?) and something that definitely is not easy to
remember, like flight numbers, arrival/departure times, telephone numbers,
and addresses (including postal codes).
3. For controlled study of the sort we have had to do for the resonance
work, we will need to have resources to collect data during the grant, as
new findings dictate. This will require budgetting for labour of
collection, labelling, checking the labels (at least).
Sarah
_____________________________________________________________________
Dr. Sarah Hawkins Email: sh110@cam.ac.uk
Dept. of Linguistics Phone: +44 1223 33 50 52
University of Cambridge Fax: +44 1223 33 50 53
Sidgwick Avenue or +44 1223 33 50 62
Cambridge CB3 9DA
United Kingdom
This archive was generated by hypermail 2.0b3 on Tue Nov 16 1999 - 10:22:01 GMT