nasality and negation, again

[Frans Plank]

Now, if you, Matthew, or whoever, went ahead and did this project 
properly, according to your specifications, and if it turned out that 
there is, in some sense, a statistically significant connection 
between negation (in one or the other or all of its possible senses) 
and nasality (in one guise or another), but not an absolute 
implication "If negation, then always a nasal":  What would that mean?

I believe this is a serious question.  My single little, ostensibly 
non-serious anecdotal example from Konstanz Alemannic -- where a 
negative marker has lost its nasal segment, itself of IE origin -- 
was intended to suggest that finding such a statistical correlation 
perhaps wouldn't mean much, as far as the human brain, human 
articulation, and human perception are concerned.  On my single piece 
of evidence I'd be prepared to conclude -- as the only *linguistic* 
conclusion -- that humans are free to either express negation through 
nasal segments or not to express negation through nasal segments.  If 
they, as learners, are confronted with a language where negation 
happens to be expressed though a nasal, they can either reproduce 
this state of affairs or change it, by way of just dropping the nasal 
(without even needing a productive phonological rule obliterating all 
alveolar nasals from onsets of grammatical, perhaps weak forms).

Well, not to be too negative, perhaps it IS of interest that in a 
language family where,some 8000 years ago, negative markers were 
innovated which happened to have a nasal in them, it occurred 
relatively rarely that that nasal was sacrificed (say, to phonology) 
by subsequent generations of learners (some 300 of them).

As I see it, a meaningful typological project then would have to be a 
diachronic one.  First, we'd have to look at the source expressions 
for the grammaticalisation of negative markers, across languages. 
The question here would be whether there is any skewing in favour or 
disfavour of nasals at the stage of grammaticalisation.  (Are, say, 
verbs of denying, or articulatory gestures of disgust, likelier to 
grammaticalise as negative markers when they contain nasals than when 
they don't?)  Second, we'd have to look at what happens to negative 
markers once they have been grammaticalised, across languages.  The 
questions would be (i) whether those that have a nasal in them are 
less likely than other, non-negative (grammatical) forms to lose 
their nasal over cycles of acquisition;  and (ii) whether or not 
those that do not have a nasal are likelier than other, non-negative 
(grammatical) forms to acquire a nasal (by whatever means -- 
spontaneous nasalisation, metanalysis of nasals from adjacent items, 
...).

Within this project, I don't think, incidentally, that Larry's and 
Derek's observations are "quite beside the point".  Their 
observations would seem to me to mean that negative markers have been 
successfully grammaticalised I don't precisely know how often (Lary 
and Derek, more info please!), and that they have been re-learnt over 
many cycles of acquisition, without nasal segments as favourable 
factors at any stage.  It's this sort of diachronic demonstration 
that I believe matters, linguistically speaking.  If we want to add 
crosslinguistic statistics, they better be statistics about 
diachronic events.

I wonder whether we are in agreement or disagreement over this 
general conclusion.

Frans



>I think we need to be careful in a number of ways about what 
>constitutes relevant evidence regarding the hypothesis of an 
>association between negation and nasality.  First, anecdotal 
>citations of examples of languages without nasals in negative 
>morphemes are largely irrelevant.  Even lists of negative morphemes 
>in hundreds of languages would tell us little (unless the 
>association were so strong that most languages had nasals in 
>negative morphemes, but that does not appear to be the case).  Such 
>lists would tell us little for two reasons.  First, it could be the 
>case that only a minority of languages have nasals in negative 
>morphemes but that there is still a statistically significant 
>association between nasality and negation, namely if nasals still 
>occur significantly more often in negative morphemes than in other 
>types of morphemes.  Second, unless one controls for genealogical 
>and areal relationships, the appearance of an association or lack of 
>an association might be an artifact of one's sample.  Solving those 
>problems is not a trivial matter, as I have argued in various 
>publications.  Larry Hyman's observation that there are a very large 
>number of Niger-Congo languages without nasals in negative morphemes 
>is quite beside the point.  To the contrary, what we need to do is 
>to avoid examining too many languages from the same family precisely 
>in order to factor out the distorting effects of large families. 
>Finally note that if we were to examine the hypothesis seriously, we 
>would also have to control for length of morphemes.  It is quite 
>possible, for example, that nasals occur in words meaning 'dog' 
>significantly more often than in negative morphemes only because 
>morphemes meaning 'dog' tend to be longer than negative morphemes. 
>The hypothesis deserves to be tested, but doing so would not be 
>trivial, and would require collecting data on other sorts of 
>morphemes as well in order to test whether negative morphemes have 
>nasals more often than other morphemes.
>
>Matthew Dryer