Biblical Hebrew Forum

["M & E Anstey" <anstey AT raketnet.nl>]

Hi Vince

> so you did get to holland after all: how is the situation there? how is
> the grad work progressing?

going well, but because we are missing family and friends we are returning
ot Australia after xmas. I'll continue with my phd, coming back for visits
every 6 months or so.

>
> (1) what alternative to linguistic theory is there for studying BH?

I think that from an empirical standpoint (in the light of contemporary
linguistic/scientific research), there is no alternative. But I also think
that the observations (however intuitive or "non-theoretical") of seasoned
readers of BH offer substantial insight into the text. I have conversed
about linguistic issues with many Hebraists whose knowledge of BH is vastly
more than mine and I invariably learn much from such conversations.
Likewise, many previous (ie from antiquity onwards) studies of BH employ no
"canonical" linguistic theory (to put it diplomatically!), but nevertheless
shed further light on the language. And I also think that the use of
linguistic theory is a necessary and important activity in studying BH.

>
> (2) FG = two representations? ockham's razor...?

Two responses: (1)Your example has one representation with two permutations,
one default I imagine and one transformational. I doubt whether one is truly
more economic than the other. (2) Who decides whether ockham's razor is an
appropriate criteria for evaluating a linguistic theory? This question is
really a can of worms. For example, in psycholinguistic research, efficiency
boils down to processing speed and accuracy, often resulting in a
proliferation of representations. By analogy with the PC, for humans memory
is cheap, manipulations are expensive. Thus Natural Phonology, however
distasteful to autosegmental or optimality people, is a strong candidate for
a linguistic theory that models what people do. (For example, Booij and
Baayen, both generativist phonologists, nevertheless argue very persuasively
that Dutch words undergoing final devoicing must be separately stored in the
mental lexicon, as in /honden/ (dogs), /hont/ (dog). So what counts as
economic?)  Naturally, a traditionaly generativist may appeal to the
performance/competence dichotomy, but consider Jackendoff's recent
critique's of GB's syntacticocentrism and the need to take into account
psycholinguistic research. And this dichotomy is even more controversial I
think than applying ockhams' razor, as argued by Nuyts, Berg, Harder, and
many others.

>
> (3) cognitive representations, &c?  in what way are these not notational
> variants?

I'm not sure what you mean. My point was that various theories disagree over
whatever it is that underlies language as it is expressed verbally,
orthographically, signed, etc. By cognitive representations I was thinking
of theories such as Cognitive Grammar, Mental Space theory, and so forth.
Other theories, such as Functional Lexematics derives syntactic structure
from the lexicon. They are not notational variants I would think, since they
are different theoretical constructs with different primitves, definitions,
operations, and so forth. They each claim to be modelling different things.
So in FG, the dominant view is that the underlying representations should
not be thought of as representing preverbal cognition (as it pertains to
language use).

>
> (4) interesting omission:  head-driven phrase structure grammar?  if i had
> a preference, i'd take the unification-based approach any day.
> unfortunately, most linguists are not up to computer science too. ;-)

My knowledge of HPSG is very minimalist ;-).

> re FG: the declarative and interrogative orders: these are arbitrarily
> stipulated? and unrelated?

In FG Decl and Int are operators. Operators are an abstract notation used to
represent any feature of a language that is expressed in a fully
grammaticalised form, such as inflectional morphophonology. This would be a
prototypical operator. Some operators, particularly illocutionary ones,
often have a complex expression in a language, an interaction of word order
and prosody, verbal morphology, and so forth. Operators are commonly used in
generative grammar (eg Seuren) and many other theories in just this way.

Now brief answers to your questions: Arbitrariness: Word order in
traditional FG is a combination of typological generalisations (such as the
preferred order of operator expression centripetal to the predicate) and
language specific parameters (such as prefield vs postfield languages). Most
importantly, however, word order is not a feature of the underlying
representation (UR)at all. The UR is hierarchically arranged into semantic
layers. Thus all word orders are the "product" of the application of
syntactic rules to URs.

Relatedness: Now, the relationship between decl and their int counterparts
has caused problems for all linguistic theories. Why do some languages
prefer intial Q words, others final, others both, and so forth? In FG,
briefly, languages prefer to put Q constituents in special focus positions.
All languages have the first position of the clause as a special focus
position (P1 in FG) and then additional focus positions in languages further
into the clause are typologically less common as you progress further
towards the end of the clause. The placement of a Q word in the first
position has different effects on different languages for the remaining
words.


>
> in other words: any order could be declarative, and any order could be
> interrogative, and the orders would not be related in any way?

No, as should be clear.
>
> on my approach, the declarative follows from syntactic structure. the
> interrogative is systematically derived in one way and not another.

The UR in FG is not a syntactic structure and thus, as mentioned above, has
no word order. They are both derived from two URs that are related by the
fact of being identical at the layers below the illocutionary layer.
  it
> seems to me that languages have a limited range of deviations for marked
> clause types, and it would be a failure to capture this generalization
> that i'm worried about.....

I think studies in word order in FG have more than amply demonstrated its
ability to handle word order generalisations across typologically diverse
languages (Bakker, Rijkhoff, Siewierska, Hengeveld, Cuvalaay-Haak, and so
forth).

The generalisations however are reached at through different means, using
different criteria, and comparing the theories and their results is by no
means straighforward. Each linguistic theory offers a window on the
complexity of language, and such a window brings into focus some things and
hides others. At least, that's my experience.


regards
Matthew

Matthew Anstey
------------------------------------------
Vrije Universiteit, Amsterdam
Faculteit der Godgeleerdheid
"Werkgroep Informatica"
anstey AT raketnet.nl

+31 (0)20 - 444-6626 (W)
+31 (0)255 - 52-6541 (H)