so-called "pro-drop" languages
W.Schulze at LRZ.UNI-MUENCHEN.DE
Tue Feb 10 15:52:20 UTC 2004
Gil's four 'types' may be helpful for diagnostic purposes, still I have some
doubts whether they help from a functional point of view. Sure, conceptual
clarity always is an explanatory advantage, but we should avoid to confuse
'conceptual clarity' meant as a metalinguistic condition with the conceptual
layers of natural language features (which rarely conform to the criterion of
'clarity'). In other words: Gil's types help to classify certain behavioral
patterns in order to make them accessible to generalizations, but they do not
necessarily reflect the functional layers of a given pattern. Let be briefly
refer to Gil's types more explicitly (here, I do not refer to types (b) and (d)
which raise different problems):
Gil's types suggest that we have safe knowledge about the opposition
'morphological vs. lexical'. For certain functional patterns, this opposition
may be clear (at least in some languages), for instance German 'ich geh-e'
(I:nom go:pres-1sg:pres). But what to do with strategies that are marked for the
suffixation (or clitization) of 'former' lexical elements? Consider Aghul, an
East Caucasian (Lezgian) language: Basically, Aghul verbs do not show agreement,
(1) zun laxnil xisu
I:abs work:sa:super go:fut
'I'll go to work'
But especially with negated verbs, clitic variants tend to occur:
(2) amma lixanas x'is-t:a-zas
but work:masd become:fut-neg-1sg:dat
'But I won't be able to work'
Now, is _zas_ (formally equal to the dative of the 1sg pronoun) a lexical or a
morphological feature? How to decide (here), whether the second phrase belongs
to Gil's type (a) or (c)? Sure, we may simply start from the formal expression
claiming that *if* the 'clitic' corresponds (in form) to a pronoun, it's just a
pronoun (in a different shape). However, we all know that clitization a) has its
proper functions and b) often affects the formal properties of the clitisized
element (English she is here ~ she's here etc.). If we subscribe to the claim
(which I do) that *nothing* in language (be it substance, structure or process),
we cannot say that a personal clitic is the *same* as it's non-clitisized form.
It's different, just because it *is* a clitic. But does this necessarily mean
that the clitic is 'morphological'? A standard assumption is that morphemes
differ from lexemes in that they cannot be uttered as such. Well, Aghul _zas_
*can* be uttered as such, although it will slightly differ phonetically from
what is produced in phrases like (1).
But *when* is the phonetic difference important enough to allow a segment to be
labeled 'morpheme' instead of 'lexeme'? Take an example from Krim Nogay (from
the poem Süygen z^ureg 'The Loving Heart'):
(3) ereg ne-ge ket-e-sin
distance which-loc come-pres-2sg
'(From) where do you come?'
mi.t-ti.-n mi. sen z^ol-un-du
forget-past-2sg:past q you:sg way-3sg:poss-acc
'Did you forget your way?'
The first phrase would correspond to type (a), the second phrase to type (c).
However, the alleged opposition between _sen_ (you:sg) = pronoun and _-sIn_
(2sg) = morpheme (?) is established mainly on the basis of features of vowel
harmony which again is coupled with the clitization criterion. I do not want to
go into the details of Turkic morphology here (I guess that Marcel will raise at
least one of his eyebrows), nevertheless the examples illustrate that the
opposition lexical-morphological may face considreable problems. Note that the
same problem can occasionally be found *within* a language. For instance, in Udi
(another Lezgian language), the first person _zu_ clearly has a clitic variant
(-zu) which differs from the lexical form by loosing it's stress properties. But
with respect to the second person, things are different: Here we have lexical
_un_ vs. clitic _-nu_, compare:
(4) zu kala-zu (Clitic, lexical?)
(5) un kala-nu (Clitic, morphological, *not* a metathesis!)
It is interesting to see, by the way, that in Udi the 'distortion' of clitics
happens with the 2nd person only (sg and pl), whereas we have true clitisized
formes with the 1st person (sg and pl). For Udi, we would have to assume that
the language subdivides its paradigm of personal clitics producing a lexical
*and* a morphological layer. Note that from a diachronic point of view, the Udi
2sg clitic, too, stems from the corresponding pronoun, which, however, had been
added to the old focus clitic *-ni (*-ni + *-(v)un > -nun (Old Udi) > -nu).
In sum, I think Gil's classification is useful for heuristic purposes. However,
I have slight doubts on whether it helps to explain what's actually going on in
at least some 'pro drop' or 'pro add' languages....
Prof. Dr. Wolfgang Schulze
Institut für Allgemeine und Typologische Sprachwissenschaft
Department 'Kommunikation und Sprachen' (Dep. II) - F 13/14
Tel.: ++49(0)89-2180-2486 (Sekr.) / -5343 (Büro)
Email: W.Schulze at lrz.uni-muenchen.de
More information about the Lingtyp