[Lingtyp] comparative concepts

[Matthew Dryer]

The claim that there are no bounds on what is possible in language does 
not mean that anything is possible.  Consider the trivial example of 
whether there are bounds on the number of phonemes in a language. I 
think people would agree that in some sense a language with only two 
phonemes is not possible. But that does not mean that there is a bound 
on the number of phonemes in the sense that there is some number n such 
that a language with n phonemes is possible but a language with n-1 
phonemes is not.  Rather there exist factors that mean that below some 
n, the probability of a language having n phonemes gets increasingly 
unlikely.

And even if there is some n whereby fewer than n phonemes is impossible, 
it is impossible to determine what that value n might be since languages 
with small number of phonemes may be possible but unlikely to be found 
in a sample of less than 100,000 languages and since there are far fewer 
languages like that, we can never know from the absence of a language 
with a given number of phonemes whether such a language is possible.

For this reason, hypothesized absolute universals fall into two 
categories. They are either trivial (like every language must have at 
least three phonemes) or untestable. What this means is that even if 
there are absolute universals, trying to identify what they are is 
fruitless.

Matthew

On 1/24/16 9:02 AM, David Gil wrote:
> On 24/01/2016 22:38, Stef Spronck wrote:
>> Apologies for jumping into a discussion in which so many senior
>> colleagues have made much more qualified comments. But surely the
>> issue is not whether anybody believes that 'languages may differ
>> without bounds'?
>> Don't the differences of opinion rather lie in whether in order to
>> discover the nature of those 'bounds' it is most
>> sensible to presuppose categories based on our deep analysis of the
>> relatively limited range of individual languages we have at our
>> disposal, or that we would like to treat the extent of the variation
>> as an empirical question (and how to best do that)?
>>
>> Both positions would seem entirely defendable to me, but result in
>> very different conceptualisations of typology.
>>
>> Best,
>> Stef
> Good question Stef (and no need for apologies!)
>
> Since I was the one who introduced the issue of "differing without
> bounds" into the discussion, let me try and clarify.
>
> First, I do believe that some of the participants in this discussion do
> subscribe to the notion that languages may differ without bounds, but
> since I state this based on impressions from casual conversations, I
> will leave it to the person(s) in question to fess up to this view if
> they so wish.
>
> But secondly, if you do — like myself — believe that there are limits to
> how languages may vary from each other, or, to use an alternative
> formulation, that there exist exceptionless linguistic universals, then
> the existence of such limits impinges on the discussion of language
> specific categories vs. comparative concepts.  The question then arises
> why such bounds or limits exist, and, as has been amply discussed, there
> could be many different reasons, including but not limited to the
> following: • monogenesis (all languages inherited the putative universal
> feature from a shared protolanguage) • historical accident (languages
> violating the putative universal feature could exist, and indeed may
> have existed in the past, it's just that today's contemporary languages
> all accidentally happen to uphold the feature in question • diachrony
> (there is no plausible path for a language upholding the putative
> universal feature to develop into a language violating it) • cognitive
> constraints (our minds can't deal with languages that violate the
> putative universal feature – be it a general cognitive constraint or one
> specific to a hypothetical grammar module).  Now it seems to me that
> some of these reasons, in particular the latter one, involving cognitive
> constraints, might arguably entail that the constraint underlying the
> universal in question could be represented not only at the level of a
> cross-linguistic generalization but also within the grammars of
> individual languages.  Though I'm sure that such a conclusion can and
> will be contested ...
>