LINGUIST List 2.738

Fri 01 Nov 1991

Misc: R-linking, Invariance, Gemination

Editor for this issue: <>


Directory

  • Geoffrey Russom, Re: 2.720 R-Linking
  • , R-Linking and Natural Phonology
  • Jim Scobbie, Phonetic invariance
  • Jim Scobbie, phonetic invariance (extra comment)
  • , Languages with gemination phenomena

    Message 1: Re: 2.720 R-Linking

    Date: Thu, 31 Oct 91 10:37:22 EST
    From: Geoffrey Russom <EL403015brownvm.brown.edu>
    Subject: Re: 2.720 R-Linking
    Question for Nancy Dray: did the Senators who had r-linking after "issue" speak the dialect that has schwa for the final syllable of the citation form? If so, r-linking would be unremarkable. -- Rick

    Message 2: R-Linking and Natural Phonology

    Date: Thu, 31 Oct 91 11:14:30 EST
    From: <Alexis_Manaster_Ramermts.cc.wayne.edu>
    Subject: R-Linking and Natural Phonology
    A recent defense of David Stampe by Richard Goerwitz misses the point of my critique of Stampe's contribution. W/o dwelling on the emotional points of Goerwitz's contribution (whose feelings I apologize for out- raging), I would like to restate the substantive point in a way which is perhaps clearer and also to point out that this criticism applies to many other people's work in linguistics and has no special relation to natural phonology (which, as perhaps one or two people know, I have considerable sympathy for). Stampe's argument that the linking /r/ must in fact be underlying because (a) r-insertion is not phonetically natural and (b) it is automatic (a process rarther than a rule in NP terms), yet (c) according to NP only phonetically natural developments can be processes (i.e., automatic). My point was that NP, as originally developed, claimed it as a matter of FACT not of DEFINITION that automatic processes are precisely the phonetically natural ones. By making what I still consider a cardinal error in reasoning in his latest posting, Stampe runs the danger of turning this into a matter of definition and hence to make his theory no longer subject to factual challenge. That is the central point, and one which applies to many other cases in the linguistic literature. For example, to cite some once celebrated examples, take Relational Grammar (in its early days) and spontaneous demotion or the way in which one typically reasons about purported universals: A: I just found a language that violates the complex NP constraint. B: No, you have not, your examples are not complex NPs. A: How do you know they are not complex NPs. B: If they were, they would obey the complex NP constraint. My point was simply that NP seems to make many almost true predictions, and it would be better to revise the theory to deal with the counterexamples rather than take the route of definining the counterexamples away. For, in that case, you lose all predictive power. Goerwitz also asks how we know that some process is natural. I would point out that this not my problem as much as David's, but in fact I think the answer is reasonably clear. If you could show me that there are many languages in which children (or adults) spontaneously insert rhotics after central vowels before a vowel EVEN though /r/ was not previously lost in this environment in the history of the language, then I would agree that this is natural. This is something that I have pointed out since 1981 at least, arguing that it is precisely those dialects which lost /r/ in this position that then insert it (or, if you accept Stampe's analysis, generalize it to all underlying post-central vowel environments). Likewise, it is precisely those English dialects that lost final /l/ that then exhibit a linking-L phenomenon. Likewise, as I have pointed out since 1981, Korean lost initial /n/ before /i/ and /y/ (y means yod not a front rounded vowel here). Subsequently, this /n/ gets reinserted even in cases where it does not belong etymologically in sandhi environments. As a result of which, one can hear Korean speakers rendering English 'not yet' as /nannyet/. The /nn/ arises, apparently, because of the reinserted /n/ and then the assimilation (by a regular and well-known rule) of the final /t/ of 'not' to that /n/. It has also been pointed out that the form of the English indefinite article ('a' vs. 'an') is determined after rather than before speech errors (see Fromkin's work) and also that Spanish, Catalan, etc., speakers of English say things like an e-street (where, presumably. the /e/ is not underlying). And again, I would say that we know that the 'a'/'an' alternation is not natural because it is not something that we find children (or adults) doing spontaneously in languages of the world. If this was natural, then we would expect Dutch or German children to go through a stage in which they develop a similar alternation before "suppressing" it. Yet they do not seem to. On the other hand, the insertion of /e/ before /sC/ clusters IS natural in this sense. And, of course, NP predicts that all the rules (which are supposed to be unnatural) apply before all the processes (which are supposed to be natural). Now, as I said, it would be better to revise the theory to deal with these examples (which Stampe has known about for at least 10 years, not least because I keep reminding him of them several times a year) rather than attempt to immunize the theory--by defining natural to mean automatic--from factual challenge. Since all these examples involve external sandhi (ifEnglish examples like draw-r-ing are external), perhaps that is the relevant factor. And perhaps it is something quite different. Indeed, maybe NP is fundamentally wrong (though I rather doubt that). But we cannot find out if the factual issue is defined out of existence.

    Message 3: Phonetic invariance

    Date: Tue, 29 Oct 91 10:52:46 PST
    From: Jim Scobbie <scobbieCsli.Stanford.EDU>
    Subject: Phonetic invariance
    Richard Ogden and Nick Cambell bring up the issues: 1) what constrains the relationship between a feature and a sound? 2) what does it mean to compare features across languages/speakers? I don't know the answer to these questions, :-), and I'd like to see more discussion of such difficult issues here. Does no-one have an opinion? Is the lack of a satisfactory answer something to be contemplated in embarrassed silence? We seem to spend quite a lot of time in this group bandying relatively trivial stuff around. Yet we are all supposed to be language professionals. Am I missing something? I'd have thought this relatively casual forum would be a good one. We can indulge in after-dinner chats happily. People don't often scream for data, references... This net is the napkin of the future. Let's have some half-baked ideas, please. Let me punt... (2) and (1) seem to be the same question. The constraints on the realisation of some feature are going to vary depending on the system of features and the system of realisations. There is also going to be arbitrary variance. There is also going to be absolute constraints. So, [+low] will on average be in a certain place universally, but this will vary both with regard to the other features in the system, the realisations of the other features and other factors (socially selected?) Note also that the featural and prosodic context of [+low] is important. I know nothing about this stuff, so can some people in the know please argue with each other to further my education? ;-) Now the question about comparing [+low] across languages/speakers. Speakers I have no problems with. You and me have the same system but sound differnt. Fine. Languages... I have no problems with a universal phonological feature set, but am not convinced comparisons of some feature across langauges is always immediately useful. Richard seems to be pointing out the arbitrary component of feature-exponent mapping then saying consequently all features are uniquely defined in each language making cross linguistic comparison impossible. Is that right? 'If [+low] for me and you is different, what the hell is [+low]' ??? If this is the question (straw-man alert) then I don't buy it. It is precisely *because* the feature-exponent relation varies that we want to talk about there being a [+low]. Otherwise we wouldn't need to. -- James M. Scobbie: Dept of Linguistics, Stanford University, CA 94305-2150

    Message 4: phonetic invariance (extra comment)

    Date: Tue, 29 Oct 91 12:41:27 PST
    From: Jim Scobbie <scobbieCsli.Stanford.EDU>
    Subject: phonetic invariance (extra comment)
    Richard Ogden asked *two* questions here: > ` What then does it mean to say that certain phonological > features are 'the same' when their interpretation in > different languages is different? or when the features they > stand in relation to in different languages is different?' In a posting here I tried to debunk the 1st half of this. It is the difference in phonetic interpretation that partly creates the phonological level: to say two features are the same when their interpretations differ in different langauges is the basic stuff of phonology. However the 2nd part, where Richard askes how can people say the [+low] in a 1-vowel langauge is the same [+low] as appears in an n-vowel language --- that is the basic stuff of bad phonology. It isn't an impossible comparison, but we need to be careful. We also need to be careful when we claim that such and such a language has so many vowels. How many vowels has turkish got in non-initial syllables? Are we talking contrastive vowels or identifyable phonological configurations? I suspect the comparisons between languages in which this parameter is manipulated is a greater problem than the one Richard brings up. For example Navajo has /ieao/ vowels. What does this mean? What is the /o/? Is it really /u/? If it is a 4-vowel system what are the 4 vowels? /o/ has [u] and [o] allophones, [o] being the primary one. So Navajo has [ieaou]. What does this mean? Is it a 5 vowel system after all? -- James M. Scobbie: Dept of Linguistics, Stanford University, CA 94305-2150 -- James M. Scobbie: Dept of Linguistics, Stanford University, CA 94305-2150

    Message 5: Languages with gemination phenomena

    Date: Fri, 1 Nov 91 11:42:25 +0100
    From: <Michael.Cheneyteol.lu.se>
    Subject: Languages with gemination phenomena
    I saw the posting of Vieri Samek-Lodovici on geminiation in the linguist. Although I am not a professional linguist, I have written a thesis on the theories of the ancient Hebrew D-stem (i.e. the one with the doubled radical) and its relationship to the geminated D-stems in the other Semitic languages. Though the specific argument of the thesis itself might not be of such great interest, the bibliography and the general discussion might be of some interest. Among the literature on the subject that might be of interest are the following: Leemhuis, F. The D and H Stems in Koranic Arabic Ryder, S. The D-stem in Western Semitic Mettinger, T "The Hebrew Verbal System" in Svensk Exegetisk Aarsbok Jenni, E Das hebraeische Pi'el: syntakticsh- semasiologische Untersuchung einer Verbalform im Alten Testament I suppose that I should also (humbly) mention that I discuss the above contributions, along with a number of articles and shorter contributions to the subject in my thesis (available on microfilm or, with some work, electronically via the network). Michael Cheney Teologiska Institutionen Lunds Universitet Lund, Sverige + 46 46 10 47 52 Internet: cheneyteol.lu.se EARN cheneyseldc52 Fax: +46 46 10 44 26