<!doctype html public "-//W3C//DTD W3 HTML//EN">
<html><head><style type="text/css"><!--
blockquote, dl, ul, ol, li { padding-top: 0 ; padding-bottom: 0 }
--></style><title>MLU counts</title></head><body>
<div>I am coming to this discussion of MLU late in the day (so,
apologies if someone has already covered the points below and I
somehow missed them), but I wanted to add a couple of points:</div>
<div><br></div>
<div>1. WITHIN-LANGUAGE CONCERNS FOR MLU:<x-tab>
</x-tab></div>
<div><br></div>
<div>In the discussion of MLU counts that are valid for the individual
language, I haven't seen anyone mention that it is extremely important
to know what is productive for the child. Even in Roger Brown's
original criteria for counting MLU for English-speaking children, he
makes it clear that one cannot just assume that if the adult language
makes use of two or more morphemes in a given word (e.g.,<i> fell</i>
=<i> fall</i> + past tense), this necessarily means that the child has
two morphemes there. One must always examine the other
utterances of the child to see what variations on the given words are
present (does the child also have "falling",
"falls", or other forms?). This question is perhaps
even more important in inflectionally rich languages than in more
analytic languages like English.</div>
<div><br></div>
<div>A second issue is what "counts" as a morpheme. If
MLU means "mean length of utterance<u> in morphemes</u>",
then one has to consider what a morpheme is. I don't know of any
linguistic treatment of morphemes that says that the number of
morphemes in a form is equal to the number of meanings contained in
that form. The definition of an inflectional language includes
the fact that individual morphemes encode a composite of meanings (in
fact, they are called "fused" for this reason). But it
doesn't suggest that each of those forms is then made up of a number
of morphemes equal to the number of meanings encoded. So, for
example, one could say that the -o in Spanish<i> com-o</i> 'I eat'
encodes (a) first person, (b) singular number, (c) present tense.
But I don't know of anyone who would count it as 3 morphemes, instead
of one. (Similarly with<i> die</i> in German--this encodes (a)
feminine gender, (b) nominative case, and (c) singular [or plural]
number, but that doesn't mean that<i> die</i> is made up of three
morphemes. It fuses three meanings into<u> one</u> morpheme.)
(This does not mean the morpheme count is necessarily non-problematic,
as such things as zero morphemes and 'process' morphemes
(reduplication, ablaut, lengthening, etc.--when these have
morphological content) are traditionally considered morphemes like any
other more overt forms.)</div>
<div><br></div>
<div>2. ACROSS-LANGUAGE CONCERNS:</div>
<div><br></div>
<div>If one accepts a construction grammar approach that says the
structural properties/categories of a language are always
language-specific (see, e.g., Croft's work), then any attempt to
equate MLU across languages is doomed to failure--at least if one is
trying to equate MLU in language A directly with MLU in language B.
The only way out of this that I can see, if one wants to try to retain
some comparability across languages in relation to MLU, is to
establish separate relations within each language between MLU and some
other criteria (e.g., vocabulary, number of verbs per utterance)
(along the lines of what Nina Hyams was suggesting), where more is
known about the comparability of those other criteria across
languages. Once those within-language relations are established,
one could then compare MLUs across languages on the basis of those
language-internal relations. E.g., one could know that in
language A a slightly higher MLU might correlate (for the purposes of
comparing levels of development across languages) with a slightly
lower MLU in language B, because the additional criteria (vocabulary,
verbs per utterance) lead to this type of matching.</div>
<div><br></div>
<div>Just some thoughts to add to the mix.</div>
<div><br></div>
<div>Ginny Gathercole</div>
<div><br></div>
<div><br></div>
<x-sigsep><pre>--
</pre></x-sigsep>
<div><font color="#000000"><br>
Virginia C. Mueller Gathercole,
Ph.D. <span
></span
> <span
></span> <font face="Times"><br>
</font>Reader<font face="Times"><br>
<br>
</font>Ysgol
Seicoleg <span
></span
> <span
></span> <x-tab>
</x-tab>School of Psychology<font face="Times"><br>
</font>Prifysgol Cymru, Bangor<x-tab>
</x-tab
> <span
></span> <x-tab
> </x-tab>University of
Wales, Bangor<font face="Times"><br>
</font>Adeilad Brigantia<x-tab>
</x-tab><x-tab>
</x-tab><x-tab>
</x-tab><x-tab> </x-tab>The
Brigantia Building<font face="Times"><br>
</font>Ffordd Penrallt<x-tab>
</x-tab><x-tab>
</x-tab><x-tab>
</x-tab><x-tab>
</x-tab>Penrallt Road<font face="Times"><br>
</font>Bangor LL57 2AS<x-tab>
</x-tab><x-tab>
</x-tab><x-tab>
</x-tab><x-tab>
</x-tab><x-tab>
</x-tab>Bangor LL57 2AS<font face="Times"><br>
</font
>Cymru <span
></span
> <span
></span
> <span
></span> <x-tab> </x-tab>Wales<font
face="Times"></font></font></div>
<div><font
color="#000000"
> <span
></span
>
| /\<font face="Times"><br>
</font
> <span
></span
>
| / \/\<font face="Times"><br>
</font>Tel: 44 (0)1248 382624 | /\/ \
\<font face="Times"><br>
</font>Fax: 44 (0)1248 382599 | / ======\=\<font
face="Times"><br>
</font
> <span
></span
>
| B A N G O R</font></div>
</body>
</html>