etc.

[Tom Givon]

Dear Funk people,

I have been waiting for someone to say this, but apparently in vain. To
anyone who has followed (and sometime participated in) the last 30-odd
years of cumulative comparative-typological work and argumentation on
the subject of the grammatical status of DAT/BEN objects, the following
would surely appear familiar, indeed superfluous. But it seems that, as
always, we linguists are doomed to chase our tail in ever-narrowing
circles, recycling the same data and arguments ad infinutum, often under
the guise of new terminology, new filing schemes, new formats or--alas--
'theories'.

There is a vast comparative-typological literature on the obligatory or
the near-obligatory "promotion" of dative/benefactive arguments to
grammatical DIR OBJ. It begins (implicitly) with Keenan's 1975,1976
seminal papers on GRs (inpired by the RG mini-explosion of the
LSA-summer 1972); Judy Aissen's description of "ascention" in Mayan;
Hawkinson & Hyman's 1974 paper on Bantu; Alexandre Kimewnyi's 1976
dissertation on KinyaRwanda; Noel Rude's 1985 dissertation on  Nez
Perce; Matt Dryer's work on "primary object"; Matt Shibatani's work on
GRs; among many many others. Among the the well established facts
comking out of this literature, the following are relevant to the
current discussion:

(a) In many, perhaps most languages (and language families) known to us,
the dative/recipient object/argument (of verbs such as 'give', 'send',
'show', 'tell', 'bring') and the optional benefactive argument are
obligaqtorily coded as the DIR OBJ of the clause, given whatever
language-specific criteria for DIR OBJ-hood exist in the language
(word-order, nominal case-marking, verbal morphology, pronominal or
otherwise, behavior/control properties). Mayan (Tzotzil), Bantu,
Uto-Aztecan (Ute), So., Arawak (Machiguenga), Sahaptian (Nez Perce),
Athabaskan (Tolowa) easily come to mind, but the list can go on and on
seemingly forever.

(b) Relatively few languages allow the DAT/BEN argument to be coded at
all as an INDIR OBJ (English being a prime example here). But even in
sauch languages, the vast majority of DAT/BEN objects in discourse are
still coded as DIR OBJs ('She gave him a book', 'She cooked him a
stake') rather than INDIR OBJ ('She gave a/the book to Joe', 'She cooked
a/the stake for him', resp). And the vast majority of the DAT/BEN direct
objects in discourse are pronouns. In my text counts in written English
(found in several publications beginning with ca. 1984), these
generalization approach the 90% level, and it is a safe bet that counts
in informal spoken English will show an even higher correlation of
DAT/BEN > DIR OBJ.

(c) In the Generative tradition, the difference between a 100%
(obligatory) and 90%  (optional) grammatical process is highly
significant. But if the grammaticalization literature of the very same
30-odd years has shown anything, it is that at 80-90% usage frequency,
the difference between the two "types" of grammar becomes negligible,
and sooner or later the 90% is interpreted as 100% by naive speakers.
(This "glossing over the difference"  is not really about language, but
about cognition and the rise of automaticity. Both automatic processing
and grammaticalization are usage- frequency-driven. There's a vast
literature on that too).

(d) Lastly, and methodologically sobering: Most of the languages with
reported "optional promotion" of DAT/BEN to DIR OBJ (English, Hebrew,
Spanish etc.) are languages with a long literate tradition, with highly
literate linguists who tend to count--if they count usage frequencies at
all--written discourse. In contrast, all the languages reporting
"obligatrory promotion" are recently-described spoken languages. As most
of us know, literacy exerts a notorious slow-down effect on
grammaticalization. So in such languages it may well be that the
reinterpreetation of 90% > 100% is already long completed in the spoken
language; but the conservative written forms preserve older relics--the
recalcitrant 10% that makes the process seem still "optional". And
literate linguists tend to report such relics as highly significant.

With apologies & best regards,  TG