User talk:Peter238/Archive 3
 | This page is an archive of past discussions for the period April 2015. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Norwegian vowel charts
Hi, I got a message from you, but it rather looks as if it was intended for someone else. Not that I'm not interested in formant charts, but I have no expertise in Scandinavian languages. RoachPeter (talk) 09:42, 1 April 2015 (UTC)
- OK, I'll be glad to check the chart over. At first sight, some vowel positions look a bit surprising, but formant plots never come out looking just like auditory vowel charts, however much we may tweak them. In passing, I should say I feel it's a pity there isn't a fuller (illustrated) explanation about constructing formant plots in the Formant article. I'd been thinking of writing something, when time permits. RoachPeter (talk) 16:33, 1 April 2015 (UTC)
- Sorry to ask such an obvious question, but have you tried plotting the Norwegian data on a simple F1 x F2 plot? I just tried a few of your vowels in that layout, and got the feeling they were a slightly better fit to what the phonetic symbols suggest. I would need to plot them all to see if it really made a difference, but I thought I'd see if you had tried first. RoachPeter (talk) 17:13, 2 April 2015 (UTC)
- Hi, I didn't mean to suggest that you couldn't get a useful formant plot from the data you have, and I am sure that what you are doing is worthwhile so that there is a meaningful basis of comparison if some new data comes along later. I just meant that many different ways of organizing formant plots have been tried so as to bring a closer correspondence between the appearance of auditory and acoustic vowel quadrilaterals. Fischer-Jorgensen's formant plot looks almost too good to be true! As you probably know, phoneticians have tried using the Mel scale or the Bark scale, or having one axis or the other, or both, logarithmic. Some very ad hoc mathematical transformations have been applied to data for the sake of a good diagram. I think that your plot is probably the best you can get, and having seen the F1 x F2 plot you sent me I can see why you want to use a different scale. RoachPeter (talk) 18:49, 2 April 2015 (UTC)
- I think the best approach would be to leave in the existing vowel chart and explain that this is based on auditory judgments, then add your formant plot with an explanation that this represents acoustic measures. I think you need to leave in the figures on the frequency axes - that's the usual convention. The only problem I have now is that the horizontal scale looks to me to be non-linear, but I can't make out what it is - possibly some sort of log scale. Do you have the information on that scale? That is information that ought to go in the caption or explanation that goes with the plot. RoachPeter (talk) 10:08, 3 April 2015 (UTC)
- I'll be travelling for the next few days, but I'll look into the scale business when I'm back. Actually, I rather like the red lines, but I suppose WP must be monochrome. RoachPeter (talk) 19:35, 3 April 2015 (UTC)
- OK, I have had a quick look through my collection of books and papers on speech acoustics. I haven't found anything to change the view I put in my last message. I have not been able to identify what the horizontal scale is, but I think as long as you include the figures you should be all right. RoachPeter (talk) 09:22, 7 April 2015 (UTC)
- I think that what you have put looks fine, and I'm sure you have made the best decision. The data you have provided gives a useful set of reference values for anyone wanting to do some acoustic analysis of their own. I still think there is a debate to be had on the general usefulness of formant plots, and I think the WP article on Formants should have a section reflecting the issues and problems. I intend to have a go at that, but don't have enough time just at the moment. RoachPeter (talk) 15:31, 25 April 2015 (UTC)
- OK, I have had a quick look through my collection of books and papers on speech acoustics. I haven't found anything to change the view I put in my last message. I have not been able to identify what the horizontal scale is, but I think as long as you include the figures you should be all right. RoachPeter (talk) 09:22, 7 April 2015 (UTC)
- I'll be travelling for the next few days, but I'll look into the scale business when I'm back. Actually, I rather like the red lines, but I suppose WP must be monochrome. RoachPeter (talk) 19:35, 3 April 2015 (UTC)
- I think the best approach would be to leave in the existing vowel chart and explain that this is based on auditory judgments, then add your formant plot with an explanation that this represents acoustic measures. I think you need to leave in the figures on the frequency axes - that's the usual convention. The only problem I have now is that the horizontal scale looks to me to be non-linear, but I can't make out what it is - possibly some sort of log scale. Do you have the information on that scale? That is information that ought to go in the caption or explanation that goes with the plot. RoachPeter (talk) 10:08, 3 April 2015 (UTC)
- Hi, I didn't mean to suggest that you couldn't get a useful formant plot from the data you have, and I am sure that what you are doing is worthwhile so that there is a meaningful basis of comparison if some new data comes along later. I just meant that many different ways of organizing formant plots have been tried so as to bring a closer correspondence between the appearance of auditory and acoustic vowel quadrilaterals. Fischer-Jorgensen's formant plot looks almost too good to be true! As you probably know, phoneticians have tried using the Mel scale or the Bark scale, or having one axis or the other, or both, logarithmic. Some very ad hoc mathematical transformations have been applied to data for the sake of a good diagram. I think that your plot is probably the best you can get, and having seen the F1 x F2 plot you sent me I can see why you want to use a different scale. RoachPeter (talk) 18:49, 2 April 2015 (UTC)
- Sorry to ask such an obvious question, but have you tried plotting the Norwegian data on a simple F1 x F2 plot? I just tried a few of your vowels in that layout, and got the feeling they were a slightly better fit to what the phonetic symbols suggest. I would need to plot them all to see if it really made a difference, but I thought I'd see if you had tried first. RoachPeter (talk) 17:13, 2 April 2015 (UTC)
English /a:/
Sudan and South Sudan. — kwami (talk) 02:50, 14 April 2015 (UTC)