JiscMail Logo
Email discussion lists for the UK Education and Research communities

Help for VAR-L Archives


VAR-L Archives

VAR-L Archives


VAR-L@JISCMAIL.AC.UK


View:

Message:

[

First

|

Previous

|

Next

|

Last

]

By Topic:

[

First

|

Previous

|

Next

|

Last

]

By Author:

[

First

|

Previous

|

Next

|

Last

]

Font:

Proportional Font

LISTSERV Archives

LISTSERV Archives

VAR-L Home

VAR-L Home

VAR-L  May 2012

VAR-L May 2012

Options

Subscribe or Unsubscribe

Subscribe or Unsubscribe

Log In

Log In

Get Password

Get Password

Subject:

Re: Normalisation of the PRICE vowel

From:

Dom Watt <[log in to unmask]>

Date:

Tue, 8 May 2012 15:58:23 +0100

Content-Type:

text/plain

Parts/Attachments:

Parts/Attachments

text/plain (31 lines)

Hi Sarah
Looking over your message, and at your vowel plots, I have a couple of questions/observations re: your PRICE data.

~ Are your formant data for the monophthongal PRICE measured at a single time slice, or are they averages over a selection?

~ Did you consider using formant tracking rather than coming up with a single point to represent each vowel? Even if you're satisfied that a vowel is phonetically truly monophthongal there's nothing to stop you tracking what the formants do over the vowel's duration, though you might need to do some hand correction if the formant tracks jump about and give you spurious-looking results (unfortunately all too common when using Praat's LPC formant tracker, particularly with less than optimal quality recordings, as I'm sure many list members have found).

~ Are you happy your formant measurements are accurate and representative of the sounds in question, whichever way you made them? Is there any perceptually obvious interspeaker variation in height among your PRICE tokens? How high and low, in Hz, is F1 getting for the speaker you represent using the green points? It might be that that speaker just makes greater use of the available vowel space than the others do. Cf. some of the male RP speakers in Hawkins & Midgley's JIPA data - there are some speakers whose F1 and F2 ranges are very much larger than others'.

~ Why do you call them 'outliers' if you think your measurements _are_ accurate and representative of the sounds in question? Appreciable, often major, levels of scatter for a single vowel category on the F1/F2 plane are to be expected, after all. For what it's worth, I've been finding a lot of similar-looking F1 variation in monophthongised PRICE vowels in some casework recordings I've been working on recently. What sort of pattern were you expecting?

~ I'd urge you not to chuck these measurements out if, by all the tests of reliability you'd normally apply, the measurements seem perfectly good. If they're in line with plausible values for formant frequencies for a speaker of that sex and that age, they're probably OK.
 
~ You ask 'how can I trust my results if they can be changed like this?'. The results are only changing because you're changing the input; what you put in is clearly going to affect what you get out. The key thing is to have your results in raw Hz as good as you can get them before you do any normalisation, and to compare the outputs of different routines to see which do the best job according to whatever criteria you're applying.

~ Applying another normalisation procedure won't necessarily give you a markedly different result, though, and you should also consider what a 'better' result might mean in this context. Even if the outliers were 'brought to heel' a bit, so to speak, would that necessarily mean that the plot is then 'better' or 'more accurate'? No normalisation routine will tidy _everything_ up (whatever that would mean - when do we know it's worked properly? How much stretching and warping of the vowel space is too much?). Reaching the point at which you're happy with the outcome is always going to involve compromises of some sort. There's no magic wand, unfortunately.

It might be helpful to look at the chapter Anne, Tyler and I wrote for Di Paolo and Yaeger-Dror's 'Sociophonetics', Flynn & Foulkes' 2011 ICPhS proceedings paper (available via the conf website), or Kohn & Farrington's 2012 JASA paper, if you haven't done so already.

I hope these remarks are useful - let me know off-list if you want any more detailed feedback on any of them.
Dom

########################################################################

The Variationist List - discussion of everything related to variationist sociolinguistics.

To send messages to the VAR-L list (subscribers only), write to:
[log in to unmask]

To unsubscribe from the VAR-L list, click the following link:
http://jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=VAR-L&A=1

Top of Message | Previous Page | Permalink

JISCMail Tools


RSS Feeds and Sharing


Advanced Options


Archives

April 2014
March 2014
February 2014
January 2014
December 2013
November 2013
October 2013
September 2013
August 2013
July 2013
June 2013
May 2013
April 2013
March 2013
February 2013
January 2013
December 2012
November 2012
October 2012
September 2012
August 2012
July 2012
June 2012
May 2012
April 2012
March 2012
February 2012
January 2012
December 2011
November 2011
October 2011
September 2011
August 2011
July 2011
June 2011
May 2011
April 2011
March 2011
February 2011
January 2011
December 2010
November 2010
October 2010
September 2010
August 2010
July 2010


WWW.JISCMAIL.AC.UK

Secured by F-Secure Anti-Virus CataList Email List Search Powered by the LISTSERV Email List Manager