<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body>
<div class="moz-cite-prefix">Yes, Larry, often our colleagues make arguments to persuade us, but we can reject their proposal because it is not the only possibility. This is often frustrating, and I think it shows that part of our energies are wasted.<br>
<br>
But it's still true that we need abstract concepts of some sort, because there are some language-internal generalizations that MUST be captured in some way. Maybe not so much in phonology (where there is a limited number of segment, syllable and pattern types),
but definitely in morphosyntax.<br>
<br>
I think the key concept is *productivity*: If a pattern productively extends to new cases, we MUST posit an abstract category. Which category is the best one is rarely clear – e.g. it's unclear whether we need "wh-movement" or "modal verb" in English, because
different subtypes of these behave differently, and we need to have these subcategories anyway (and whether we need the broad categories is unclear).<br>
<br>
But we need some kind of rule of Verb-Subject Inversion in English, for example, because we cannot provide a list of all English Subject expressions. The unlimited productivity of constructions like "has my sister come", "will Kim leave", "must we posit a rule",
and so on, shows that there must be SOME abstract category.<br>
<br>
In phonology, productivity is sometimes illustrated with the ill-formedness of */bnik/ in English, and the well-formedness (but non-existence) of */blik/. To the extent that such judgements are consistent, I find this argument persuasive.<br>
<br>
(This is not like Franklin/Crick/Watson's DNA, but that's a different level of generality anyway. I think Zipf's discoveries about frequency effects on length are perhaps on a par with DNA, because they are extremely general.)<br>
<br>
Best,<br>
Martin<br>
<br>
<br>
On 24.03.20 17:40, Larry M. HYMAN wrote:<br>
</div>
<blockquote type="cite" cite="mid:CAMQd4M3njk8RH0Z0HdpSh6SqCZfvDXqfjbQz0f0HfffWXSrKLg@mail.gmail.com">
<div dir="ltr">Thanks for the response--and the challenge, Martin. I reread the blogposts to refresh my memory, all of which makes me think of Molière's, "C'est trop de grâce que vous me faîtes!" I doubt I'll fully take up your challenge, but yes, as I've indicated
before, I agree with the distinctions you make, although I'd like to make a small commentary on the following that you wrote in one of the blogposts,
<div><br>
</div>
<div><span style="color:rgb(55,55,55);font-family:"Helvetica
Neue",Helvetica,Arial,sans-serif;font-size:15px"><b>Language-particular theories (=analyses) often make use of abstract concepts (such as the syllable or the adjective) because language-internal
generalizations cannot be captured otherwise. (<a class="moz-txt-link-freetext" href="https://dlc.hypotheses.org/263">https://dlc.hypotheses.org/263</a>)</b></span><br>
</div>
<div><span style="color:rgb(55,55,55);font-family:"Helvetica
Neue",Helvetica,Arial,sans-serif;font-size:15px"><br>
</span></div>
<div>where I wonder what "cannot be captured otherwise" might mean. I often find myself saying that there is no knock-down argument in linguistics--nothing like Crick and Watson's discovery of the double helix, which others could then ignore only at their own
peril! Coming back to linguistics, the no knock-down argument is also partly due to another thing I find myself lamenting too often, that it's too bad science has to be practiced by human beings: If someone is determined to have a theory without syllables
and does not buy arguments of "insight", "simplicity" or "generality", they will continue to deny the reality not only of syllables, but potentially of any other useful innate or comparative linguistic concept (cf. Kiparsky's discussion of this re the syllable
on pp.77-78 in the Hyman & Plank <i>Phonological Typology</i> volume that you cite). Look at all of the deniers of consonant and vowel segments who blame our use of these on "alphabetism"! As I point out in the 2008 "Universals in Phonology" article you cite,
"claimants [for or against universals] should make clear what they would accept as a potential counterexample" (p.109). If it has already been pre-determined that there can be no counterexamples, then the claim is not interesting.</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Tue, Mar 24, 2020 at 4:06 AM Haspelmath, Martin <<a href="mailto:haspelmath@shh.mpg.de" moz-do-not-send="true">haspelmath@shh.mpg.de</a>> wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px
0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
That's great, Larry! Maybe you could write a paper about this for tone <br>
languages. (Or maybe you already have a paper about this?)<br>
<br>
We had some discussions about "normalized concepts" (= uniform <br>
yardsticks for measurement) earlier (see, e.g., this 2012 blogpost about <br>
a "normalized" syllable concept: <a href="https://dlc.hypotheses.org/263" rel="noreferrer" target="_blank" moz-do-not-send="true">
https://dlc.hypotheses.org/263</a>).<br>
<br>
And there was an interesting controversy between you and Kiparsky in the <br>
recent Hyman & Plank volume "Phonological typology", which I discussed <br>
here: <a href="https://dlc.hypotheses.org/1817" rel="noreferrer" target="_blank" moz-do-not-send="true">
https://dlc.hypotheses.org/1817</a><br>
<br>
Kiparsky works with the traditional generative idea of uniform innate <br>
building blocks, and it seems that he doesn't even entertain the idea <br>
that one could have two different descriptions: one that takes into <br>
account all the language-particular generalizations, and one that is <br>
"normalized" (= allows uniform measurement).<br>
<br>
Since your name is associated (at least by some older folks like me) <br>
with "generative phonology" (your 1975 textbook was very influential), I <br>
think you would be uniquely placed to explain the difference between <br>
innate building blocks, language-particular analyses, and comparison via <br>
normalized concepts.<br>
<br>
Best,<br>
Martin<br>
<br>
On 23.03.20 17:26, Larry M. HYMAN wrote:<br>
> Thanks, Martin. I am sensitive to what you wrote. In fact, in creating <br>
> my "catalogue", which I don't call a database since I prepared it more <br>
> as an "index" to the 665 tone systems so that I could find things, I <br>
> actually classified the tone systems both by the language particular <br>
> analysis AND by my attempted normalization.This is what allowed me to <br>
> find the examples so quickly (whichi of course would need to be <br>
> further scrutinized, as the descriptions also vary in quality). For <br>
> example, if a language was analyzed with H, L, LH, and HL tones, I <br>
> have a field that tells me there are two tone heights and another that <br>
> tells me that the author considered the system to have 4 <br>
> tones--whereas as an Africanist I would call it 2 tones, since LH and <br>
> HL are combinations. I also have a separate field for contour tones <br>
> where I can find which languages have how many rising or falling tones <br>
> (up to five each!), according to the author again. I did have to <br>
> "translate" the descriptions that use numbers to Hs, Ls and Ms. For <br>
> example a system such as Blang [BLR] reported as 55 31 51 13 would be <br>
> listed in one field as such, but in my general inventory field as H L <br>
> HL LH, with 2 heights in my tone height field, 4 tones in my # of <br>
> tones field, and 1F and 1R in the # of contour tones field. Best, Larry<br>
<br>
</blockquote>
</div>
</blockquote>
<br>
<pre class="moz-signature" cols="72">--
Martin Haspelmath (<a class="moz-txt-link-abbreviated" href="mailto:haspelmath@shh.mpg.de">haspelmath@shh.mpg.de</a>)
Max Planck Institute for the Science of Human History
Kahlaische Strasse 10
D-07745 Jena
&
Leipzig University
Institut fuer Anglistik
IPF 141199
D-04081 Leipzig </pre>
</body>
</html>