<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML xmlns:o = "urn:schemas-microsoft-com:office:office" xmlns:st1 =
"urn:schemas-microsoft-com:office:smarttags"><HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=Windows-1252">
<META content="MSHTML 6.00.2800.1561" name=GENERATOR></HEAD>
<BODY>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>dear
everyone,</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006></SPAN></FONT> </DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>two
weeks ago I asked for ways to count syllables with CLAN. I tried the way Brian
MacWhinney suggests with postcodes, but as I need a distinction between "filler
syllables" and others this would leave me with too many different postcodes. I
decided to make a %syl: tier and have a distinction between filler syllable
<STRONG>$fill</STRONG> & counted syllable<STRONG> $syll</STRONG>. And just
tap for every syllable once.</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>this
goes quite quick, although it doesn't look very clear. </SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>A
typical line looks like:</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006></SPAN></FONT> </DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006>*ANA: Yes, I think uh that these uh
would not be relevant.</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006>%syl: $syll $syll
$syll $fill $syll $syll $fill $syll $syll $syll $syll $syll
$syll</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006></SPAN></FONT> </DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>and
then counting them with FREQ</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006></SPAN></FONT> </DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>freq
+t%syl file.cha</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006></SPAN></FONT> </DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>giving
an output like</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006>$syll 11
</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006>$fill 2</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006></SPAN></FONT> </DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>Thanks
for your answers below you find them.</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=314135214-23102006>bye,
bye,</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006>Marije</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006></SPAN></FONT> </DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN
class=314135214-23102006></SPAN></FONT> </DIV>
<DIV><FONT color=#0000ff><SPAN class=314135214-23102006>
<P><FONT color=#008000><FONT face=Arial size=2><SPAN
class=314135214-23102006>Answer of Brian MacWhinney:<BR></SPAN></FONT><FONT
face=Arial size=2>This is the type of analysis that will eventually be best
</FONT><FONT face=Arial size=2>supported by the Phon program that Yvan Rose,
Greg Hedlund, and I are </FONT></FONT><FONT face=Arial><FONT size=2><FONT
color=#008000>developing (see http://childes.psy.cmu.edu/phon/) But PHON can't
really do this yet. So, for the time being, hand coding in CLAN is the best
method and the method you propose seems excellent. Obviously, you can get total
counts by just multiplying in FREQ. Moreover, you can also check for sequences
of monosyllables and such with COMBO. So, I like your solution in the general
case. In fact, PHON will be implementing something like this too, I believe.
However, if all you cared about was the total number, you could save yourself a
bit of work by must appending a postcode such as [+ 6] and then running FREQ on
the postcodes.<SPAN class=314135214-23102006> </SPAN>One could imagine a special
purpose version of MOR that was used just to yield syllable counts. You could
use the "english translation" field for this. You would get<SPAN
class=314135214-23102006> </SPAN>%mor: pro|I=$1 v|walk=$1 adv|home=$1
adv|tomorrow=$3</FONT><SPAN class=314135214-23102006><BR></SPAN><FONT
color=#008000>Then you would FREQ to focus on the items after the =<SPAN
class=314135214-23102006> </SPAN>To make this work, you would need to have all
words recognized by MOR </FONT></FONT></FONT><FONT color=#008000><FONT
face=Arial size=2>and you would have to add syllable numbers to all words in the
</FONT><FONT face=Arial size=2>lexicon. Not an easy job, but eventually we will
have to do this.</FONT></FONT></P>
<P><FONT face=Arial><FONT size=2><FONT color=#008000>--Brian MacWhinney<SPAN
class=314135214-23102006> </SPAN></FONT></FONT></FONT></P><FONT
color=#000000><FONT face=Arial><FONT size=2><SPAN class=314135214-23102006>
<P><FONT face=Arial><FONT size=2><FONT color=#ff00ff><SPAN
class=314135214-23102006>answer of Eva Aguilar Mediavilla:<BR></SPAN>For my
codification I use a dependent tier %syl: where I put the number<SPAN
class=314135214-23102006> </SPAN>and type of syllables ex. (in Spanish) tomato
%syl: W-S-W and the in<SPAN class=314135214-23102006> </SPAN>other dependent
tier I codified the number of syllables in the word
ex.<BR></FONT></FONT></FONT><FONT color=#ff00ff><FONT face=Arial size=2>%wor: 3.
It's a similar solution as you.<BR></FONT><FONT face=Arial size=2>Eva Aguilar
Mediavilla</FONT></FONT></P></SPAN></FONT></FONT></FONT>
<P><FONT><FONT face=Arial><FONT size=2><SPAN class=314135214-23102006>answer of
Ann Peters:<BR></SPAN></FONT></FONT></FONT><FONT><FONT face=Arial><FONT
size=2><SPAN class=314135214-23102006><FONT size=2>I have also wanted to count
syllables and have invented my own method which I will describe below. First,
however, I need to point out that it is important to be clear what one wants
syllable counts for. I suspect that your goal and mine differ. But perhaps
seeing how I solved my problem will help you solve yours.<BR>What I really
wanted to know what how the child's *utterances* increased in syllable length
and what kinds of prosodic patterns were occurring. In particular I wanted to be
able to track the appearance of "filler syllables", including where in
utterances they were appearing. So I devised a %syl tier that I have hand coded
(for all the tapes from 18-1/2 through 22 months). The specific codes I have
used are:<BR>L = stressed lexical syllable<BR>l = unstressed lexical
syllable<BR>f = filler syllable<BR>g = unstressed grammatical
morpheme<BR></FONT></SPAN></FONT></FONT></FONT><FONT><FONT face=Arial><FONT
size=2><SPAN class=314135214-23102006><FONT size=2>X = stressed unglossable
syllable<BR>x = unstressed unglossable syllable<BR>S = animal sound or sound
effect (stressed)<BR>s = animal sound or sound effect
(unstressed)<BR></FONT></SPAN></FONT></FONT></FONT><FONT><FONT face=Arial><FONT
size=2><SPAN class=314135214-23102006><FONT size=2>Words within an utterance are
linked with _ [underline]. (On the main line @fs is the suffix for filler
syllables. On the %pho tier, ^ signifies main stress.)) <BR>Typical early tiers
look like this:<BR>*CHI: ready
?<BR></FONT></SPAN></FONT></FONT></FONT><FONT><FONT face=Arial><FONT
size=2><SPAN class=314135214-23102006><FONT size=2>%pho: rEdi<BR>%syl:
Ll<BR>*CHI: nono@c ?<BR>%pho: n6^n6<BR>%syl: lL<BR>*CHI: tape .<BR>%pho:
tIp<BR>%syl: L<BR>*CHI: xxx ?<BR>%pho: h6^h6<BR>%syl:
xX<BR></FONT></SPAN></FONT></FONT></FONT><FONT><FONT><SPAN
class=314135214-23102006><FONT face=Arial size=2>*CHI: n@fs Daddy ?<BR>%pho: n
^d@di<BR>%syl: f_Ll<BR>*CHI: eee@p !<BR>%pho: ^i:-^i:<BR>%syl: SS</FONT></P>
<P><FONT face=Arial size=2>Later tiers look like these:<BR>*CHI: a@fs get diaper
?<BR>%pho: 6 g6 ^dap6<BR>%syl: f_l_Ll<BR>*CHI: a@fs floppy@c ?<BR>%pho: 6~
fa^piy<BR>%syl: f_lL<BR>*CHI: read a@fs book .<BR>%pho: riy6 bUk .<BR>%syl:
L_f_L<BR>*CHI: close'it .<BR>%pho: ^kotsIt<BR>%syl: L'g<BR>*CHI: n@fs play pans
?<BR>%pho: 6m ^pey ^p@ns<BR>%syl: f_L_L<BR>*CHI: un@fs brush a@fs teeth
?<BR>%pho: 6m br6sh6 ^tiyf<BR>%syl: f_L_f_L<BR>*CHI: thankyou ,, Daddy
.<BR>%pho: ^thanky6 </FONT><A href="mailto:^d@diy"><FONT face=Arial
size=2>^d@diy</FONT></A><BR><FONT face=Arial size=2>%syl: Ll_Ll</FONT></P>
<P><FONT face=Arial size=2>Note that each %syl tier has only a single "word" in
it. (This was Brian's<SPAN class=314135214-23102006> </SPAN>suggestion.) Then I
use FREQ to sort the patterns, as follows:<BR></FONT><FONT face=Arial
size=2>GREQ file.cha +t*chi -t* +t%syl +k +f<BR>This sorts and counts the
patterns in my %syl tiers. I then hand-sort the<SPAN class=314135214-23102006>
</SPAN>output in order to extract information about e.g.<SPAN
class=314135214-23102006> </SPAN>How many utterances with N syllables did he
produce?<SPAN class=314135214-23102006> </SPAN>How many were iambic vs.
trochaic?<SPAN class=314135214-23102006> </SPAN>When did medial fillers come in
and then disappear?<SPAN class=314135214-23102006> </SPAN>I hope this gives you
some ideas.<SPAN class=314135214-23102006> </SPAN>good luck!<BR></FONT><FONT
face=Arial size=2>ann</FONT></P></SPAN></FONT></FONT>
<P><FONT color=#000000><FONT face=Arial><FONT size=2><SPAN
class=314135214-23102006></SPAN></FONT></FONT></FONT> </P>
<P><FONT color=#000000><FONT face=Arial><FONT size=2><SPAN
class=314135214-23102006></SPAN></FONT></FONT></FONT> </P><FONT><FONT><SPAN
class=314135214-23102006>
<P><FONT face=Arial color=#000000
size=2> </FONT></SPAN></FONT></FONT></SPAN></FONT><BR><BR></P></DIV></BODY></HTML>