Alan R. King mccay at REDESTB.ES
Wed May 5 14:03:20 UTC 1999

>These are two serious questions. I would really like to know whether
>a) there are any Non- ProDrop languages outside Germanic/French, and
>b) there are any Verb-Second languages outside Continental Germanic?
>Help me.
>Leon Stassen.

As soon as I tried to answer these questions I realised that the answer
will depend on definitions: what is pro-drop? what is verb-second?  And I
don't mean in an abstruse theory-embedded sense.  The initial problem
arises from the fact that these two concepts are in practice familiar to us
from "very" European contexts in which we may think we know just what we
mean; but, in a broader typological perspective such as Leon asks for,
hence beyond those considerably language-specific (or
language-group-specific) contexts, do we, I ask?  So the questions need to
be addressed at two levels: (1) what are we looking for? and (2) can we
find it?  (And isn't this double question at the heart of the whole
typological endeavour?)


I have always felt that the concept, nay the very term, "pro-drop" betrays
where most modern linguistic theoreticians have been coming from: the
European world (linguistically speaking) in general and ENGLISH in
particular.  The original implication was surely that some languages show
the "surprising" property that ordinary clauses do not obligatorily contain
an overt subject NP; whereas if the initial vantage point had been almost
anywhere else in the world, the opposite observation would have been the
surprising one: that there exist languages in which the overt presence of a
subject NP is generally obligatory.  The term implies that, in most
languages (undoubtedly), when the subject isn't mentioned it is because
"something" has been "dropped", a notion which I think makes most sense if
the linguist happens to speak English.  So it is always with misgivings
that I go along with the current terminology and employ this annoying
expression, "pro-drop".

It is my understanding that, if we are going to talk about "pro-drop", this
term must be limited by definition in the way I just did in such a way as
to refer to overt NP arguments.  (The fact that it is often understood to
refer just to *subject* arguments is another giveaway revealing the
Anglocentric thinking that conceived it; we can just as easily talk about
object pro-drop, and other types too.  But in this message I am going to
assume that Leon intended the question to refer to subject pro-drop.)  That
is, the argument in question must be...

(1) OVERT: a non-pro-drop language is not merely one in which "the subject
need not be expressed but is always understood" (if it were, some linguists
would perhaps argue that all the world's languages are pro-drop in that
sense - or those languages in which "subjects" can be identified, at any
rate), but one in which the subject *always expressed*, i.e. represented by
morphosyntactic material present in the clause.

and (2) represented by a NOUN PHRASE, where NPs may of course be
constituted by pronouns, among other things: without this part of the
definition, Latin would be non-pro-drop, on the grounds that in, e.g., _Amo
te_ "I love you", the subject is identified by the verbal inflection
(_-o_).  But that is not the distinction we want to capture.  The
distinction we want to capture is the one betweeen Latin, with _(Ego) amo
te_ where the NP _ego_ is entirely optional, and French, with _Je t'aime_
where _je_ is entirely obligatory.  It is the subject NP that we're
interested in, not the indexing of the subject within the verb phrase.

I insist on saying NP, rather than pronoun; the only justification I see
for the "pro-" part of the term "pro-drop" is the fact that the least
explicit reference to the subject permitted in English is by means of a
pronoun; if one thinks only in terms of English, then the item that looks
like it has been "dropped" is therefore interpreted as a pronoun.  In fact,
the "dropped" subject might just as easily be seen to be any other kind of
NP: in Latin "Amat me" the supposedly dropped subject could be "Fred" just
as much as _is_ "he" or _ea_ "she".

The practical problem of identifying "(non)-pro-drop" languages therefore
presupposes our ability confidently to identify NPs and distinguish these
from what some of us might call agreement markers but I prefer to call
indices and talk about indexing.  (Note that it only makes sense to say
that _amo_ in _Amo te_ "agrees" with the subject pronoun _ego_ if the
pronoun is assumed to be underlyingly present and therefore somehow
"dropped".)  As far as Latin is concerned I assume everybody will agreed
that _ego_ is an NP and _-o_ isn't; but what about more exotic languages?

In my experience, it is usually fairly easy to decide when we are dealing
with a pronominal item that constitutes, in syntactic terms, an NP
argument, and when we are not, although there are also cases susceptible to
debate.  But a major source of confusion is the historical accident whereby
most exotic languages have been described, at least initially, by scholars
or amateurs who often labelled as "pronouns" items which I would definitely
include in the second category (as indices rather than nominal elements),
and their terminology has subsequently usually stuck.  Owing to
preconceptions again attributable to the historical accident of the use of
European or "classical" languages as models, subject indices are generally
identified as such when they happen to take the form of inflectional
morphemes, but usually not when perceived as words or particles.  We must
therefore read beneath the surface of the available grammatical
descriptions before being too hasty about classifying languages as
"(non)-pro-drop".  Getting misled in this respect will tend to inflate the
number of apparent non-pro-drop languages.

KIRIBATI (Micronesia; East Oceanic, Austronesian) is a good example.  There
are the following personal pronouns (three persons, two numbers), which
have NP properties:

ngai, ngkoe, ngaia, ngaira, ngkami, ngai'a

and a corresponding set of subject-indexing particles which precede the
finite verb, and have no NP-like properties whatsoever but are called
pronouns in grammars:

I/N, ko, e, ti, kam, a.

On this analysis, Kiribati is pro-drop.  If, however, the second set are
called "subject" pronouns, then Kiribati will be non-pro-drop.

There may be other reasons of a completely different kind for hesitating
about the classification of a given language.  Here are two such:

In WELSH the modern colloquial language is non-pro-drop pretty much across
the board, or close to it; the conservative literary language not as much
so.  The literary Welsh phrase that appears on some pound coins,

Pleidiol wyf i'm gwlad.
partial am to-my country
"I am loyal to my country."

displays a pro-drop usage in which the verb _wyf_ "am" stands without a
following subject pronoun that is impossible in the present-day spoken
language.  So whether Welsh is pro-drop or not depends on the variety

Classical HEBREW is pro-drop; finite verbs are indexed for the person,
gender and number of the subject, and a subject NP need not be overtly
present.  But post-classical Hebrew remodelled the tense system, developing
in the modern language a present tense from a periphrastic source.   Unlike
the modern past and future (inherited tenses which which still have full
subject indexing), the modern present tense lacks person indexing, and with
just this tense pro-drop is not allowed.  This of course responds to a very
widespread (though not universal) tendency for "pro-drop" to be permitted
just in cases where the argument in question is indexed.  So whether Modern
Hebrew is pro-drop or not depends on which tenses are considered.

The example of Hebrew reminds me of another kind of reservation we should
bear in mind.  As in many other languages, not all Hebrew clauses, or even
main clauses, contain a finite verb form; an example of a verbless clause is:

Ani talmid.
I student
"I am a student."

Also like many but not all languages, subject indexing in Hebrew is
restricted to the verb.  Given the principle referred to above for pro-drop
to be restricted to cases where the argument is indexed, it can be
correctly predicted that in Hebrew pro-drop is not possible in verbless
clauses.  So, in Hebrew and (even more notably) some other languages, the
presence or absence of pro-drop depends on the type of clause examined.  In
KIRIBATI, for example, the series of subject-indexing particles occurs with
finite verbal predicates but not with non-verbal (e.g. copular) predicates.
 In the latter the subject can only be expressed by the real pronouns of
the first series above.  (Whether Kiribati permits pro-drop, i.e. the
omission of the pronoun, in such cases, I don't know.)

One more "but".  Many languages have restrictions on the use of
third-person pronouns, e.g. to animate referents.  In such a language,
there is so-called "dropping" of the pronoun when this is "it" (say).  But
that is surely not the same thing as pro-drop, if in the remainder of cases
"dropping" does not occur.

So, back to Leon's first question: Are there any Non-ProDrop languages
outside Germanic/French?

I am not able to respond with a neat list of such languages, but here is a
quick-and-dirty first move: a few languages that MAY (or may not) be such,
according to the information (or lack thereof) that I have on hand.
Perhaps others more knowledgeable can comment on the rights of these
languages to be on the list, as well as adding others.


* IN EUROPE (the following are the only candidates I've noticed, apart from
Germanic and French):

1) (modern) Celtic languages: Welsh [see comments above], Gaelic...  Breton
allows pro-drop, which I think is a conservative trait in comparison to
modern insular Celtic, where English (or areal, if one prefers) influence
is probable.
2) Baltic languages: Lithuanian, Latvian.
3) Russian, Ukrainian (other Slavonic languages seem to be pro-drop).


4) Persian?
5) Tagalog?
6) Pidgin and Creole languages? (some based on European languages...)
7) Artificial languages: Esperanto (actually European, of course!)

* SOME "DEBATABLE" EXAMPLES (cf. the discussion above; I believe these
languages should be analysed as having "pro-drop"; however in most cases,
at least some grammars refer to subject indices as "pronouns", according to
which they would be seen as non-pro-drop languages because the said indices
are generally obligatory)

8) Europe: Friulian
9) Africa: Hausa, Somali, Yoruba
10) Oceanic Austronesian: Kiribati (see above), Marshallese, Chamorro (all
in Micronesia), Fijian.


Again, questions of definition must be faced first.  Germanic languages
(except English) display the "verb-second" phenomenon with certain specific
(and well-known) characteristics.  We should not expect to find just that
set of specific features occurring together in other verb-second systems,
if such exist at all.  Rather, the verb-second principle might
(hypothetically) occur in the context of some other syntactic system which
will then probably give rise to somewhat distinct final results.  I assume
that the "verb-second" feature means, roughly, that as a general rule the
verb is preceded by exactly one "element" within the sentence, where one
problem will be to define precisely what the syntactic system in question
regards for this purpose as an "element" (and another, perhaps, what it
regards as a "verb": cf. below!).  My understanding of Germanic syntax is
that the pre-verbal "element" may be any of the following: a) unmarked
subject; b) marked focus; c) marked topic.  At least that's the way it
seems to me to work in Yiddish, for example.  However, different languages
with a "verb-second" feature (if such exist behyond Germanic) may differ
widely in the latter respect.

I either don't know or can't think of any languages outside Europe that
obviously fit this description; can anyone else?  By pushing matters just a
little, some languages in Europe that just might admit such an analysis,
and where such an analysis might even be helpful, do occur to me.  Trying
to be helpful, yet without wishing to put all my money on any of these
purported "cases" of "subject-second" outside Germanic, here are two attempts:

1) Breton?
2) Basque?

1) Breton

As a Celtic language, Breton is often regarded as VSO, but modern Breton
clause syntax shows a rather peculiar version of VSO.  First of all, it is
not merely possible but commonplace in Breton for a clause constituent to
precede V; the pre-V position is perhaps basically a focus position, but
there are some instances of elements obligatorily occupying this position
without being markedly in focus, as understood from a pragmatic
perspective.  The clincher, if there is one, is the last-mentioned point:
with certain syntactically-defined exceptions, the Breton verb as a rule
not only can but MUST be preceded by something.

The choice of the element to precede the verb depends on various factors
(and I certainly don't pretend to understand Breton syntax well enough to
state all the rules).  There may only be one constituent available in the
clause for the position.  When there are more than one, information
structure may determine the choice (treating the position as a focus slot).
 In other cases, certain "tricks" (in a manner of speaking) are used to get
something into the pre-verbal position.  Thus, if the verb is periphrastic,
consisting of an auxiliary and a participle, the auxiliary alone may be
considered V for the purpose, and the participle made to stand first in the
clause, thus reversing the usual AUX + MAIN VERB order of Breton.  If
necessary, a periphrastic verb form may be chosen (obligatorily) rather
than a simple one to "conspire" to make the aforesaid construction
possible.  Finally, certain particles are treated as candidates for the
initial slot, so that in clauses containing one of these particles the verb
is only preceded by the particle.  On one analysis, the introduction of
such a (sometimes semantically empty) particle may be forced by the need
for the verb to stand second.  Once we eliminate the clause types covered
by the above situations, the residue of actual verb-initial clauses in
Breton cover a narrow range indeed.  While perhaps some Breton grammarians
will prefer alternative analyses of the Breton sentence to that I have just
suggested, I believe it is not too far off the mark as a POSSIBLE way of
looking at Breton syntax, at the very least.

2) Basque

Basque is sometimes stated to be SOV, though certainly not rigidly so.  In
fact word order is, subject to information structure constraints, quite
flexible; SOV is only an unmarked order, not an obligatory one.  Basque
word order is actually better represented as TFV, i.e. TOPIC + FOCUS +
VERB.  The frequency or unmarkedness of SOV undoubtedly derives from the
strong cross-linguistic tendency, in general, for SUBJECT to correlate with
TOPIC, and OBJECT with FOCUS.  But these correlations are easily reversible
in Basque.  An unlimited number of constituents may also follow V in the
Basque clause (except for certain subordinate clauses); these are in
principle simply constituents marked neither for focus nor topichood; a
more complete statement of word order is thus TFVX.

So far none of this looks like it's going to get us to verb-second syntax.
However, one fact is obscured in the TFV formula: far from showing the same
degree of syntactic incorporation into the clause, T is actually the least
fully incorporated constituent, and F the most.  First, T is absolutely
optional, while F is often described as obligatory (whether the F position
may be truly empty in Basque sentences is an issue to complicated to enter
into here).  The verb is also in principle obligatory, so we really have
(T)FV (or something more complicated but bearing a relation to this).
Secondly, when T is present it may be followed by a pause or pronounced as
a separate intonation unit, and followed by a comma in writing.  F cannot
be followed by a pause and typically forms a single intontation unit with
the following V.  It has even been argued (e.g. in a recent unpublished
paper by Phyllis Bellver and Laura Michaelis) that T is an "extraclausal
position... reserved for detached topics".  A further argument that should
be mentioned is the fact that the same syntactic constraints to be
mentioned in the next paragraph that apply to absolute clause-initial
position apply to post-TOPIC position.  To the extent that we accept the
peripherality of T in the Basque clause, then, the latter can be seen as
essentially FVX, and therefore verb-second.

Perhaps this looks like a rather forced argument, but a few other details
about Basque syntax may change our minds about that.  Even though,
following a standard rule of Basque grammar, I called F obligatory above,
there are clauses in which, depending on how you look at it, either there
is no F, or the predicate (roughly, V) is itself F.  Here F cannot precede
V, and we would expect to find V in initial position (or preceded only by
T).  Yet for some reason, Basque syntax appears to conspire against V
becoming absolutely and unreservedly clause-initial, especially in main
clauses.  And despite the enormous distances separating Breton and Basque
clause structure, the "tactics" used in each language to carry out the
conspiracy are uncannily reminiscent of those used in the other.  In Basque
the verb may be clause initial but only if it is periphrastic.  Now in
periphrastic verb forms in Basque the usual order in MAIN VERB + AUXILIARY;
only the auxiliary is a finite verb form, so that in these cases it is
plausible to regard the auxiliary as representing V (and the main verb as
F), under which analysis initial-V is effectively avoided here.  Simple
verb forms, lacking an auxiliary, cannot normally stand clause-initially,
unless preceded by a particle; one such particle is semantically empty.  So
in both Breton and Basque clauses, you can have the following:

FOCUS + VERB + ...

but not:


In the case of Basque, we may add (TOPIC +) in parentheses at the beginning
of each formula; note that the starred ones will remain starred.


I have deliberately glossed over an interesting complication - an
inconsistency, we could say - in the Basque system.  On the one hand, where
the rule for focus placement is concerned, "verb" (i.e. what must follow F)
may represent either a synthetic or a periphrastic verb form, rarely the
auxiliary of a periphrastic form alone.  In other words, the usual orders
+ AUXILIARY + MAIN VERB is highly marked, and in some varieties of
questionable grammaticality or ungrammatical.  Yet on the other hand, where
FOCUS doesn't appear as a separate item, the fact that MAIN VERB +
AUXILIARY + ... is possible clause-initially but SIMPLE VERB + ... is out,
implies that the "verb" here comprises either a synthetic verb or an
auxiliary.  One interpretation of these facts is that Basque syntax bends
over backwards to maintain a verb-second principle of sorts (or at least a
non-verb-initial one).  Why?


Alan R. King, Ph.D.
alanking at
- - - - - - - - - - - - - - - - - - - - - - - -
SNAIL: Orkolaga plaza 3 1A, 20800 Zarautz, Basque Country, Spain.
PHONE: +34-943-134125   /   FAX: +34-943-130396
Alternative email addresses:   mccay at, a at,
70244.1674 at
Internet:   <>

More information about the Lingtyp mailing list