LINGUIST List 4.461

Tue 15 Jun 1993

Sum: V-AGR Merger & Methodology

Editor for this issue: <>


  1. Steven Schaufele, Sum: V-AGR Merger & Methodology

Message 1: Sum: V-AGR Merger & Methodology

Date: Mon, 14 Jun 1993 21:13:55 Sum: V-AGR Merger & Methodology
From: Steven Schaufele <>
Subject: Sum: V-AGR Merger & Methodology

I posted a query in LINGUIST 4-405 concerning the falsifiability of
Pollock's hypothesis that the direction of V-AGR Merger (whether V ascends
to AGR or AGR descends to V) is typologically related to 'richness' of
subject-agreement marking, 'richness' being (by assumption) defined in
morphophonological terms, i.e., number of different inflectional forms
corresponding to distinct person-number categories.

First of all, to all the people who asked for copies of unpublished papers
that i cited in my query, your copies will be in the mail by the time you
read this.

A few people referred me to a handful of works offering arguments against
the counterevidence i had mentioned vis-a-vis Italian and Scandinavian. To
them i say Thank you, and i have tracked down these items and will be
reading them very critically over the next few weeks or so. No-one seems
to be proposing an alternative analysis of the Kru languages, however,
according to which these languages, which have no subject-agreement
marking, actually have AGR-lowering contrary to Koopman's analysis. Michal
Starke <> says,

 >'the problems with the correlation between verbal morphology,
 >verb-movement, and null subjects are discussed at length in the
 >literature ... Many different examples, counter-examples,
 >counter-counter-examples exist and are discussed,'

but actual references for this discussion are so far not forthcoming.

Bernhard Rohrbacher <> posted a
response to the net (LINGUIST 4-407); i summarize here his comments and the
response i sent to him. He suggested that subject-agreement marking might
be richer in some French conjugation classes than in others; i agreed that
there was some variation, but not enough to make subject-agreement marking
*sound* impressively richer in French than in English. He noted that it
has been suggested that the obligatory use of nominative-case pronominal
proclitics in some colloquial French usages is developing into a new,
richer subject-agreement marking system. I granted that the hypothesis has
been suggested (as long as as Tesniere, i believe), but that

 (1) 'it has been pooh-poohed by some Gallicists of my acquaintance' and

 (2) if it is true 'it is true only of certain colloquial registers,
 >while Pollock's arguments are based on formal standard French, for
 >which the near-obligatory use of subject pronominal clitics is not
 >characteristic. I question to what extent a fact about one dialect or
 >register can be related as a diagnostic to a fact about another, for
 >which the proposed diagnostic is not actually a fact.'

Rohrbacher also implied in his posting that loss of the V-AGR raising
option in English followed hard and fast upon the loss of most of the
subject-agreement marking in that language. In my response, i said,

 >'my evidence indicates that, while most of English subject-agreement
 >marking had been lost by early in the 15th century, V-AGR raising by
 >Pollock's diagnostics was still a viable option 200 years later. Hence
 >my remark that, if Pollock is right, then several generations of
 >English people were violating the ECP. More to the point, if PPA had
 >evolved in the reign of Elizabeth I rather than Elizabeth II, would
 >anyone have been able to conceive of Pollock's hypothesis? Is the
 >coincidence of AGR-lowering with poverty of subject-agreement marking
 >merely a coincidence, a peculiar artifact of late-second millenium
 >English? Or was English somehow *compelled*, after losing most of its
 >subject-agreement marking, to lose the V-AGR raising option as well?

--even if it took several generations to do it?--

 >Are the Kru languages doomed to lose all V2 phenomena because they've
 >lost all subject-agreement marking? (Did they ever have any?) If not,
 >then what predictive value does Pollock's hypothesis have?

Leslie Barrett <> acknowledged that, as Pollock's
hypothesis makes some empirical claims, it should be subject to testing,
but argued that 'the burden of proof may not rest with him'. In response,
i agreed

 >that a scientist has a right to come up with a hypothesis in any
 >feasible fashion, at any point in hanns investigation, and if the
 >hypothesis promises rich theoretical dividends it may even be hanns
 >duty to publicize it without loss of time. ... However, once a
 >hypothesis, however promising, has been conceived, i believe in general
 >the very next step in scientific inquiry should be the definition of
 >possible counterexamples and the quest for same. I do not blame
 >Pollock for having formulated his hypothesis and broadcasting it at
 >various conferences. But i am inclined to hold Linguistic Inquiry and
 >its reviewers liable for having published his paper without first
 >confronting him with the (at least apparent) counterexamples in
 >Koopman's analyses of verbal syntax in the Kru languages and Hyams' of
 >Italian. These were readily available in the literature at the time.
 >And i am very concerned at the behaviour of a large number of
 >syntacticians since Pollock published his paper. Instead of defining
 >possible counterexamples to his hypothesis and searching for them, they
 >have apparently been so impressed with the theoretical dividends
 >promised by it that they have simply incorporated it into their
 >theoretical apparatus and proceeded to work on the assumption that,
 >e.g., if a language has rich subject-agreement marking it *must* ipso
 >facto have V-AGR raising, and any evidence to the contrary must be
 >somehow explained away ... even if contortionist analyses are
 >necessary, rather than allowing Pollock's hypothesis to be judged on
 >the basis of hard cross-linguistic data. I fear a tendency to treat
 >'exotic' languages (in this case, any languages outside the
 >Northwest-European Sprachbund) the way late mediaeval professors of
 >medicine treated corpses in their dissection theatres, rejecting out of
 >hand those that 'presumed' to contradict Galen.

 >A case in point. Rivero 1990 ('The Location of Nonactive Voice in
 >Albanian and Modern Greek', Linguistic Inquiry 21:135-146) presents an
 >analysis of verbal inflections in Modern Greek and Albanian consistent
 >with Pollock's hypothesis, which looks plausible if one doesn't have
 >much background in classical or comparative Indo-European studies. As
 >Joseph & Smirniotopoulos 1993 ('The Morphosyntax of the Modern Greek
 >Verb as Morphology and Not Syntax', Linguistic Inquiry 24:388-398)
 >demonstrate, however, a good grounding in the details of Greek grammar
 >easily demolishes Rivero's analysis. Would Rivero's analysis have been
 >conceived without the publication of Pollock's paper and the resulting
 >tendency to make of his hypothesis a Procrustean bed? I am inclined to
 >regard work such as Rivero's as more obscurantist than enlightening.

(because in their publication they obscure the true nature of the languages
in question, and in their need to be refuted they may take time away from
research which might itself be more enlightening. Joseph and
Smirniotopoulous were not bringing to bear any new facts, but facts that
have been available for a long time in commonly-available literature but
whose relevance was apparently overlooked by a number of syntactic
theorists.) I have Andrew Carstairs-McCarthy
<> to thank for these examples.
 Carstairs-McCarthy looks at this issue from a morphologist's point of
view, from which the whole approach implicit in Pollock's hypothesis seems
'morphologically naive'. More generally, he remarks,

 >I think part of the trouble is that many linguists assume that, if one
 >has a theoretical approach which severely restricts the range of
 >analyses which can be offered for a given set of data, one has ipso
 >facto made progress and said something with empirical relevance

--even if maintenance of said theoretical approach requires contortionist
reanalyses of other data than create more problems than they solve. It is
to some extent the problem of the 'beautiful theory threatened by an ugly

In conclusion, i would note that one of the big questions i tried to raise
in my query is still going begging: What would constitute a hard
counterexample to Pollock's hypothesis, one which the most dedicated
supporter of the putative typological link between richness of
subject-agreement marking and direction of V-AGR merger would acknowledge
seriously compromised it? Is Pollock's hypothesis falsifiable? I want to
pursue the issue further, but i want a general agreement on what sort of
data i (and anyone else who has in mind to challenge Pollock's hypothesis)
should be looking for. I am not keen on spending several months tracking
down data that seems to me to undermine it, only to be told when i try to
present it that it is quite irrelevant.
Dr. Steven Schaufele c/o Department of Linguistics
712 West Washington Ave. University of Illinois
Urbana, IL 61801 4088 Foreign Languages Building
 707 South Mathews Street
217-344-8240 Urbana, IL 61801
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue