[HPSG-L] Formalism/theory distinction: citations

[Gerald Penn]

Dear Olga Zamaraeva,

  It's probably a mistake to be looking for a genesis of this distinction
within
HPSG, because HPSG itself developed within a context of multiple interacting
theories that partly shared their formalisms, and this distinction was
already
known within that context.
  The oldest discussion that I am aware of was in Mark Johnson's
dissertation in
the late 1980s, but it may very well go back earlier than that.  The
justification
at the time was a very pragmatic one:  separating formalism from theory
allowed
us the freedom to choose a very standard formalism from logic and computer
science
in which the proof theory and search algorithms were already well understood.

Regards,
Gerald

On Tue, January 5, 2021 3:49 pm, Olga Zamaraeva wrote:
> Dear colleagues,
>
> In a number of papers, it is pointed out that there is an important clear
> distinction between formalism and theory in HPSG: the formalism can be
> used
> to encode a number of theories (e.g. with linearization and without,
> etc.).
> See e.g. Bender 2008
> <https://web.stanford.edu/group/cslipublications/cslipublications/TLS/TLS10-2006/TLS10_Bender.pdf>
> or Bender and Emerson 2020
> <https://hpsg.hu-berlin.de/Projects/HPSG-handbook/PDFs/cl.pdf>, or Daniels
> and Meurers 2004 <https://www.aclweb.org/anthology/C04-1025.pdf>.
> This distinction becomes crucial in implementations so it tends to come up
> in computational linguistics literature, though the principle seems to be
> a
> philosophical one.
>
> I would like to know (i) who was the first to describe this distinction in
> these terms; and (ii) where does the philosophical principle originate.
> As
> for (i), it appears it may have been Bender 2008 but it also seems like it
> could be much earlier than that. Everything earlier than that that I found
> does not talk about the distinction directly but rather seems to assume it
> as a fact. As for Pollard and Sag, they do say that they "eschew extreme
> formalization" in order to be able to talk about the theory but I am not
> sure I could infer the distinction from that?
>
> I'd be grateful for any thoughts or references!
>
> Yours,
> --
> Olga Zamaraeva
> _______________________________________________
> HPSG-L mailing list
> HPSG-L at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/hpsg-l
>