[Lingtyp] What's the point of the phonological phrase? (and some questions about p-linguistics)

Tue Jan 7 17:42:49 UTC 2020

Hello all - a few responses...

Dear Joan,

The assumption underlying your comment seems to be that phonological facts can provide evidence for constituency in general. I can't think of any way this is methodologically tractable except by rejecting the a priori assumption that phonological and morphosyntactic constituency are really seperate things. I completely agree with this. But then, why do we even use terms like phonological word and phonological phrase at all? Why not just say constituent structure and point out that it can be motivated by different types of facts?

Dear Robert,

I would be up for the ideas that linguistic systems are best understood in terms of components if the theories didn't seem to just evolve into elaborate tautologies. I'm wondering how do we could validate componential approaches without stipulating they are true a priori. I'm really interested in knowing what the relationship is between good descriptive practice and the borrowing of terminological frameworks from such modular approaches.

Dear Martin,

It is often said that we should "describe languages in their own terms", which implies that it is possible not to - what does a description look like that does not describe a language in its own terms? I assume this refers to cases where unmotivated architectural assumptions and categories are being overlayed on the description of the language in such a way that it obscures the facts of the language or does not allow one to clearly articulate the facts. Does anyone have a better formulation of what it might mean to *not* describe a language in its own terms and how we know this is being done?

So with this in mind, when someone borrows some concept from generative linguistics in the description of a language (out of inspiration, fealty or whatever), to what extent are they imposing the architectural assumptions of that theory on the language and therefore deviating from the ideal that they should be describing the language "in its own terms"?

Sometimes, I would assume, its harmless - just a a convenient piece of terminology adapted and changed for the case at hand. But I think that with categories of the prosodic hierarchy, like the phonological phrase, this is not necessarily the case. The idea that there is a seperate phonological constituency relies on some important claims that are unstated or not taken up in descriptive grammars that adopt notions like phonological word, phonological phrase etc. There's at least two of them: homologies across CS representations projected from POSs and the distinction between lexical and post-lexical phonology:

Homologies across constituent structures projected from POSs:
Crucially, phonological phrasing rules are supposed to make reference to "arboreal" constituent structure representations abstracted from part of speech categories. If they can make reference to part of speech categories then they are not general enough to rise above the level of diacritics. So a phrasing rule can see something like root-affix-clitic or [[XP] XP] but not something like verb root-valence marker-aspect or NP[obj]-V. To make a phrasing rule general one has to show (or assume) structural homologies between different POS domains. But there are no consistent rules of inference for arriving at constituency structure representations (especially when we allow transformations) even when all of the relevant ingredients of the theory are stipulated (binary branching, a deep structure where arguments are faithful to UTAH, etc.). Is it "describing a language in its own terms" to just stipulate homologies so that phonological phrasing can look general across different constructions? Otherwise, why not just use junctures if its easier to understand what the phenomena are.

(To be more concrete, say we have a phonological process like Meeussen's rule HH->HL, applying at junctures between prosodic words in some domain. We observe it applying at and Adj-N juncture, but not a Adv-V juncture. The syntactic relation between these categories should be different according to the premises of prosodic phonology - but how do we know? which consituency tests do we rely on? - seems like there is some implicit methodological opportunism, just embedded in the assumptions of phonological phrasing. I'm concerned because there are no general or consistent methods of inference for arriving at constituency structure representations and so the whole theory risks being a tautology (e.g. assume different or identical CS representations when it suits the phrasing rule).

Lexical vs.post-lexical distinction:
In order to maintain that prosodic constituents cannot see POS categories, one has to bury certain morphophonological facts into a lexical component. I think a distinction between lexical and postlexical rules should be argued for independently of the fact that it let's the linguist ignore morphophonological facts that reference POS categories. If you can't motivate such a distinction, does it belong in a grammar? Again, the prosodic hierarchy stipulates this architectural assumption ?  But should languages be described this way if we can't empirically motivate a lexical vs. post-lexical phonology distinction?

Martin points out that in specific languages the concept of a phonological phrase is fine because they could useful ways of describing a specific language. I have two questions about this. First, is this really the case? Could the facts not be described equally well with the old-school juncture approach or a domain based approach that makes no specific reference to levels (e.g. stress domain) especially given that it is not burdenned with the architectural assumptions of prosodic phonology? If, its just expositionally better (I'm skeptical of this too), then that should be explicitly stated, no?

The second issue regards Martin's distinction between p-lingusitics and g-linguistics. Its not just that some linguists adopt terminology from prosodic phonology - its that if you do not, you are criticized for it and it is strongly implied that you don't know what you are talking about. The distinction between p-linguistics and g-linguistics must be rejected by a lot of linguists even if they profess to find value in description in and of itself. I've been told that the prosodic hierarchy is motivated and its categories and architectural assumptions should be adopted - the only evidence that is brought to bear on this is that the metalanguage of the prosodic hierarchy is used in a lot of descriptions. If we take seriously the distinction between p-linguistics and g-linguistics,then this argument is moot to begin with. But even if we don't: given that the theory (and there are few versions) have had no empirical successes at all, I think we should be having a conversation about why descriptive linguists need to be browbeaten into adopting the terminology. Scheer is not the only source in this regard (see the papers by Schiering, Bickel and Hildebrandt).

The distinction between p-lingusitics and g-linguistics is useful because it allows us to avoid potentially pointless debates about whether some purported "universal" category is really there in some language and just get on with description and comparison. But I don't think this means "anything goes" in p-linguistics. I think we should favor more surfacy descriptions if the abstract categories require analytic shortcuts to get off the ground. We should wonder whether our analyses in p-linguistics are just expositionally useful or whether they can be thought of as motivated generalizations across the grammar. To take an extreme example, I think it would be wrong to insist that a descriptions isn't "typologically informed" because it does not refer to phases, little vP structures, movement operations etc. But what distinguishes the "phonological phrase" and other categories from the prosodic hierarchy from these other ingredients from generative linguistics?

best,

Adam

Adam James Ross Tallman (PhD, UT Austin)
ELDP-SOAS -- Postdoctorant
CNRS -- Dynamique Du Langage (UMR 5596)
Bureau 207, 14 av. Berthelot, Lyon (07)
Numero celular en bolivia: +59163116867
________________________________
De : Lingtyp [lingtyp-bounces at listserv.linguistlist.org] de la part de Daniel Ross [djross3 at gmail.com]
Envoyé : mardi 31 décembre 2019 17:02
À : lingtyp at listserv.linguistlist.org
Objet : Re: [Lingtyp] What's the point of the phonological phrase?

Dear Adam,

My first reaction to your message was to agree with what you wrote at the end, that “the phonological phrase” is simply “some morphophonological ... domain ... higher than ... phonological word”].

Phonology is often usefully described with different tiers of structure, from the phone to the syllable to the word to the phrase to the utterance. (I wouldn't rule out having multiple levels within some of those, e.g. successively larger "phonological phrases".) The only problem seems to be operating from the assumption that syntactic and phonological phrases should line up. If we do not assume that, both are useful in their own ways, and then a secondary but very interesting question is whether and how we can relate those two structural types.

Daniel

On Mon, Dec 30, 2019 at 4:27 AM Haspelmath, Martin <haspelmath at shh.mpg.de<mailto:haspelmath at shh.mpg.de>> wrote:
On 30.12.19 01:41, TALLMAN Adam wrote:
So, what’s the point of the mapping rules if they can do anything? A conclusion I would draw from this is that positing phonological words and phrases are basically theoretically vacuous terminological conventions.

I think one needs to distinguish between general theories and language-particular theories. G-theories are theories of Human Language, and p-theories are theories of particular languages.

A grammatical description has often been seen as a theory of a particular language (since the 1960s), and in this perspective, language-particular categories are not "vacuous". In fact, they are often crucial (and in other cases convenient) for formulating language-particular rules.

But it's true that the distinction between p-linguistics and g-linguistics is not always made in practice:

I think descriptivists have the impression that using terms like phonological word and phonological phrase makes their descriptions “typologically informed” in some sense. I think the opposite is true if there are no accepted mapping rules and no accepted understanding of what the morphosyntactic structure from which the mapping rules are defined is supposed to be. In the end a typology of morphosyntactic/phonological domains etc. that tries to capture the relevant phenomena with “mapping” will have to relate phonological domain back to how it maps from/onto morphosyntactic structure based on typologically comparable wordhood / constituent domains. Positing mapping rules apriori in descriptions does not achieve this goal and makes such a typology more difficult.

Languages often have similar rules, similar categories, and similar domains, so it's important to be aware of what other languages do – often one does not need to invent everything from scratch but can be inspired by other researchers (both by other p-linguists and other g-linguists).

But in order to know that two things are similar, one needs to have uniform yardsticks for comparison – and one cannot assume that the category of one language can serve as a general yardstick for comparison for all. I don't know enough about phonology to judge what might be a good yardstick for phonological phrasing phenomena...

It seems that Scheer (2008) does not make the distinction between p-linguistics and  g-linguistics, because his impression of "anarchy" seems to be a statement about languages in general (i.e. he has not found any universals). Particular language systems are law-governed by definition.

Best,
Martin

--
Martin Haspelmath (haspelmath at shh.mpg.de<mailto:haspelmath at shh.mpg.de>)
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

_______________________________________________
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
http://listserv.linguistlist.org/mailman/listinfo/lingtyp
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20200107/519593db/attachment.htm>