Skip to content

Phonetic symbols (IPA) in R graphics

8 messages · Giampiero Salvi, Paul Murrell, Hin-Tak Leung +1 more

Original
Giampiero Salvi
# 
Hi all,
I would like to add phonetic symbols in my figures. Usually I typeset
my documents in LaTeX and I use the tipa [1] package to get
International Phonetic Alphabet (IPA) fonts. So, my problem would be
solved if I could insert LaTeX commands in the text() function (I
guess at least).

I would like to avoid using psfrag (that is, inserting a string in
the eps figure and then substituting it with the right symbol with
dvips) because I would like to work with pdf pictures and pdflatex
as well as latex -> dvips -> ps2pdf.

If inserting special fonts is not possible at the moment, I would
appreciate some hints on how I could go about modifying text() to
include this feature. (This is why I post to r-devel instead of
r-help)

Thank you!
Giampiero

[1] http://www.essex.ac.uk/linguistics/clmt/latex4ling/tipa/
Paul Murrell
# 
Hi
Giampiero Salvi wrote:
Take a look at ...

Fonts, lines, and transparency in R graphics. R News, 4(2):5-9,
September 2004.

Non-standard fonts in PostScript and PDF graphics. R News, 6(2):41-47,
May 2006.

http://www.stat.auckland.ac.nz/~paul/Talks/fonts.pdf

Paul

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Giampiero Salvi
    

    

      
      #
    

  


  

      
Hi Paul,
I tried the following from your slides, perhaps you can see the
error:

1) I downloaded the fonts and the encoding file into the directories
   Type1 and encodings (in my current directory)
2) I ran R from the same directory (ver 2.4.0)
3) I added the font mapping for the pdf and postscript devices:

ipa <- Type1Font("InternationalPhoneticAlphabet",
                 c("Type1/tipa10.afm",
                   "Type1/tipabx10.afm",
                   "Type1/tipasi10.afm",
                   "Type1/tipabx10.afm"),
                 "encodings/silipa.enc")
pdfFonts(ipa=ipa)
postscriptFonts(ipa=ipa)

4) I created a simple figure:

plot(10,10,type="n")
text(10,10,"whatever",family="InternationalPhoneticAlphabet")

Here I get 13 warnings of the type:
"font family not found in X11 font database"
As far as I understand this is because I defined the fonts only for
the postscript and pdf devices.

5) I copy the device to postscript:

dev.copy2eps(file="whatever.eps")

Here I get the real problem:
Error in dev.copy(file = "whatever.eps", device = function (file =
ifelse(onefile,  :
        invalid character sent to 'PostScriptCIDMetricInfo' in a
single-byte locale

I googled for PostScriptCIDMetricInfo but only found a few pages in
Japanese, can you help me?

Thanks,
Giampiero

I can put the font and encoding files somewhere if necessary.
I am working on a RedHat linux computer with locale:
LANG=en_US.iso88591
LC_CTYPE="C"
LC_NUMERIC="C"
LC_TIME="C"
LC_COLLATE="C"
LC_MONETARY="C"
LC_MESSAGES="C"
LC_PAPER="C"
LC_NAME="C"
LC_ADDRESS="C"
LC_TELEPHONE="C"
LC_MEASUREMENT="C"
LC_IDENTIFICATION="C"
LC_ALL=C

        
On Thu, 14 Dec 2006, Paul Murrell wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Paul Murrell
    

    

      
      #
    

  


  

      
Hi

        
Giampiero Salvi wrote:

            

      
      
      
    

        
Exactly.  The process would be simpler if you drew directly to
PostScript or PDF via postscript() or pdf().

Also, the text() call should be ...

text(10,10,"whatever",family="ipa")

            

      
      
      
    

        
I'm not sure how you got there.  You've specified the family wrong so I
would expect more "family not found" warnings.  Let me know if this
persists with the correct family specified.

Paul

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Giampiero Salvi
    

    

      
      #
    

  


  

      
Hi Paul,
first thank you for helping out.

        
On Fri, 15 Dec 2006, Paul Murrell wrote:

        

            

      
      
      
    

        
[...]

            

      
      
      
    

        
Strangely enough I get the
"family 'ipa' not included in PostScript device"
if I run
text(10,10,"whatever",family="ipa")
while the "PostScriptCIDMetricInfo" error if I run
text(10,10,"whatever",family="InternationalPhoneticAlphabet"")

I get the same behaviour when I write directly to the postscript
device:

postscript(file="whatever.eps")
plot(10,10,type="n")
text(10,10,"whatever",family="InternationalPhoneticAlphabet")
or
text(10,10,"whatever",family="ipa")

I suspect the "PostScriptCIDMetricInfo" error depends on the fact the
the encodings/silipa.enc file contains all values from 33 to 255
(with a few undefines values). Can that be possible?

Thanks,
Giampiero

  
  


  


      

    

    
3 days later

    

      
      

        


  

        

      Hin-Tak Leung
    

    

      
      #
    

  


  

      
Hi,
I am not familiar with how R deals with postscript fonts, but
I am fairly familiar with some font technology, so here are a few
random thoughts:
(1) You are not supposed to specify just any random font family string
- postscript type 1 fonts contain their own family name inside.
(read the top of the font file - the header is is plain ascii). If it
works with a random font family name, it is probably because R does not 
check (?).
(2) if you are using symbol fonts, some of X11 funtionality may depend
on your locale - the warning message on PostscriptCIDMetricIndfo
did say something about you using a one-byte locale - LANG=en_US.iso88591 .
(3) There are various ways to add fonts to the X server - for new
linux systems you are just copy them into ${HOME}/.fonts (don't know if 
R can use fonts via xft), for older linux boxes, you can do
"xset fp+ <fontdir>", after running mkfontdir in that directory
beforehand.

        
Giampiero Salvi wrote:

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Giampiero Salvi
    

    

      
      #
    

  


  

      
Hi Hin-Tak Leung,
thank you for your help. Paul and I had solved the problem off-line.
Here is the solution: the encoding file contained in the tipa package
is not in adobe format and therefore causes the problem. The default
encoding, however, seems to work well. In case of fonts corresponding
to extended ascii, the octal code you find in appendix A in the tipa
manual [1] can be entered in the string in the form \xyz.

This code demonstrates the use of IPA fonts (provided you have
downloaded the fonts in the Type1 directory in your current path):
Note in the end that you have to explicitely embed the fonts into the
pdf file.

ipa <- Type1Font("InternationalPhoneticAlphabet",
                  c("Type1/tipa10.afm",
                    "Type1/tipabx10.afm",
                    "Type1/tipasi10.afm",
                    "Type1/tipabx10.afm"))
pdfFonts(ipa=ipa)

pdf("ipademo.pdf")
grid.text("whatever", y=2/3, gp=gpar(fontfamily="ipa"))
grid.text("WHATEVER", y=1/3, gp=gpar(fontfamily="ipa"))
grid.text("\312", y=1/4, gp=gpar(fontfamily="ipa"))
dev.off()
embedFonts("ipademo.pdf", out="ipademoembed.pdf", fontpath="Type1")

The only missing part is including symbols that are created in
tipa/LaTeX by macros, but this is another story...

Thanks to anyone who helped.
Giampiero

[1] http://tug.ctan.org/tex-archive/fonts/tipa/tipaman.pdf

        
On Mon, 18 Dec 2006, Hin-Tak Leung wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Jeffrey Horner
    

    

      
      #
    

  


  

      
Giampiero Salvi wrote:

            

      
      
      
    

        
I've recently added freetype font support (http://www.freetype.org/) and 
pdf output to Simon Urbanek's Cairo package:

http://www.rforge.net/Cairo/

which supports pdf and png output. Freetype supports many font formats, 
including truetype and the newer opentype format. When R plots are 
created with truetype (or opentype) fonts, Cairo will by default embed 
the fonts into the pdf file.

I've also noticed a couple of debian unicode truetype font packages with 
IPA extensions:

$ apt-cache search ipa | grep ttf
ttf-thryomanes - A Unicode font covering Latin, Greek, Cyrillic and IPA
ttf-junicode - a Unicode font for medievalists (Latin, IPA and Runic)

So, I could envision creating an R package very similar to the LaTeX 
tipa package that could spit out the IPA unicode code points for these 
truetype fonts, thus supporting plots with IPA symbols but exclusively 
for the Cairo graphics device.