9.255 layered hypertext, cont.

Humanist (mccarty@phoenix.Princeton.EDU)
Fri, 27 Oct 1995 08:35:27 -0400 (EDT)

Humanist Discussion Group, Vol. 9, No. 255.
Center for Electronic Texts in the Humanities (Princeton/Rutgers)

[1] From: John Merritt Unsworth (39)
Subject: Re: 9.255 humanities & software

[2] From: Language Technology <langtech@DGS.dgsys.com> (37)
Subject: Re: 9.255 humanities & software

[3] From: Brian Whittaker <BRIANW@VM2.YorkU.CA> (104)
Subject: Re: 9.251 more on layered hypertext

Date: Thu, 26 Oct 1995 09:23:45 -0400 (EDT)
From: John Merritt Unsworth <jmu2m@jefferson.village.virginia.edu>
Subject: Re: 9.255 humanities & software

In response to some comments on IATH's Inote software:

> From: flannagan@ouvaxa.cats.ohiou.edu
> >=20
> ....The project at UVa=20
> sounds promising, but the graphic model of drawing circles around=20
> patches of text bothers me a little, as do "note-pad" interfaces: it'=
> messy enough on paper when one is grading student themes, and it vari=
> from copy-editor to copy-editor. I like hot-keys linked to=20
> definitions, still images, video-clips, sound-bites. =20

Inote isn't intended to replace regular hypertext tools and interfaces,=20
(such as HTML and the Web) which provide the kind of links that Roy=20
favors: it's intended to supplement those interfaces with a way to talk
about and analyze images. In the case of an image of the printed page,=20
we assume that an SGML transcription of the page would also exist, and=20
might well be linked to other materials in the way that Roy describes:=20
the page image, then, would be used as the basis for hypertextual=20
discussion of the physical object.

> From: Willard McCarty <mccarty@epas.utoronto.ca>
> Subject: humanities computing software
> I'm glad to hear about IATH's super software, but on checking out the URL
> that John Merritt Unsworth provided, to
> http://jefferson.village.virginia.edu/software.html,=20
> I find that it is for the X-windows environment, which effectively stops=
> me cold. Sure, I can get to an X-environment, but not on my home machine,=
> where all my important work (such as it is) is done. So I for one would=
> encourage IATH to develop its tools on more commonplace equipment.=20

We understand that most humanists don't work at Xwindows stations, and we=
know that, at this stage, Inote will have a limited usefulness because it=
is a Unix-based program. However, Unix is an excellent environment for=20
prototyping software, and that's what Inote is at this point--a=20
prototype. As I think I mentioned in an earlier post, we are beginning=20
now to reprogram Inote in Java, a cross-platform programming language=20
that should make it possible for us to distribute Inote through the Web,=20
as needed, in conjunction with annotated images. If anyone is interested=
in more information about Java, see http://java.sun.com/

Meanwhile, and as we go through that re-programming process, we'd=20
appreciate comments or suggestions from Humanist readers who can get=20
access to Xstations or X-emulation software. =20

John Unsworth

Date: Thu, 26 Oct 1995 11:09:19 -0400 (EDT)
From: Language Technology <langtech@DGS.dgsys.com>
Subject: Re: 9.255 humanities & software

In reading responses on Williard's notion of 'layered text=20
representation', it occurred to me that the analogy needn't be too=20
exact. I thought of multi-layered transparencies such as those used on=20
an overhead projector with the primary text on the bottom and various=20
illustrative notations on flaps folded over as the discussion proceeds. =20
However useful that may be for making a presentation to a live audience,=20
using it on the computer might be a different matter. =20

I realized though [with Williard's complaint about the inaccessability of=
X-windows s/w] that there is a more immediate possibility, which is MS=20
Windows [let's not worry about which numbered version] copied from the=20
original Mac version. Recently I've been developing some NLP software in=
Windows, using Allegro Common Lisp programming environment [not for the=20
faint of heart, spirit, or endurance, I assure you]. This environment=20
uses multiple windows, a variety of pulldown menus, hotkey combinations,=20
etc. to allow a variety of editing, debugging, and tracing activities to=20
be accomplished at the same time [more or less]. While I don't recommend=
that you all rush out and learn Allegro Common Lisp or even its=20
environment, I see this s/w as an example of the multiple means for=20
manipulating text. One capability that might be compared is that for=20
dealing with various Lisp structures. If you know anything about Lisp,=20
it is that everything is structured by parentheses so the capability of=20
editing chunks of Lisp code [with the 'chunk' defined by a particular pair=
of parentheses] is comparable to similar manipulation of 'chunks' of=20
printed text by means of it textual (or linguistic) structure -- word,=20
phrase, sentence, paragraph, etc. =20

The point I'm trying to make by in the previous paragraph is=20
that some of these techniques for handling and annotating text already=20
exist and should be considered as sources for potential models for our=20
own research environment.

Also, the most recent versions of WordPerfect include editing=20
capabilities and embedded notes that should be investigated. =20

Now I must get back to the aforementioned CL environment for the moment!=20

Mary Dee

Mary Dee Harris, Ph.D. =09=09202-387-0626
Language Technology, Inc. =09=09langtech@dgs.dgsys.com
2153 California St. NW =09=09=09mdharris@aol.com
Washington, DC 20008

Date: Thu, 26 Oct 95 13:59:33 EDT
From: Brian Whittaker <BRIANW@VM2.YorkU.CA>
Subject: Re: 9.251 more on layered hypertext


We may share identical fantasies of a multi-layered hypertext
environment. Indeed, it was the idea of the possibility of a
*transparent* (in every sense) analytic environment that drew
me to the Macintosh for its amphibious text/graphic abilities
ten years ago and has left me frustrated ever since.

For some years there has been a programme that tantalizinly
perpetuates the fantasy: NISUS, the Macintosh word processor
which, from its inception, has had a graphics layer that lets
you doodle right on your text, circling, drawing arrows, and
so on. This is perfect for demonstrating stylistic patterns,
either on screen or for textbook illustration. To suggest
some trivial examples, you could circle the personal pronouns
and draw arrows back to their antecedents, you could circle
lexemes from the three leading notional sets in a text
using three distinct colours, or you could draw your
favourite syntactic diagram above or below the sample (chinese
boxes, binary trees, or those loopy diagrams that look like
circuitry exercises set by Tolkien's elves). The three problems
are that there are only two layers, one for text and one for
graphics; that the graphics layer cannot be searched; and
that even if one could search the graphics layer, the search
would not terminate in the original phenomenon in the text

Let's explore the fantasy. Imagine a hypertextual environment
with multiple transparent layers stacked on top of one another,
like the three boards in a three-dimensional chess set. On the
bottom layer you have the text; on the second layer you have
a phonetic transcription of the text; on the third layer you
have a grammatical description using your favourite diagrams;
on the fourth layer you have a formal lexical description that
marks patterns of collocation; on a fifth layer, you have arrows
to sources of quotations and references; on the sixth layer you
have a contextual description relating the other layers to the
social situation. (This fantasy uses categories from a particular
school of stylistics; those who belong to different denominations
are free to make substitutions and alterations.)

Why bother? First of all you could combine several layers on screen
on in printout to examine by eye or teach in class the interaction
of the levels of linguistic and stylistic analysis. In addition,
you could ask the computer, for example, (and here we move deeper
into the realm of fantasy) to search for occurrences of the
grammatical pattern:

Prepositional Phrase
/ =9B
Preposition Noun
/ | =9B =9B
/ | =9B =9B
/ | =9B =9B
/ | =9B =9B
Quantifier Attributive Countable Prepositional
Noun Noun Phrase
/ =9B
Preposition Noun Phrase

(for example "five dog collars behind him")

and list all such phrases where the attributive noun is from the
field of discourse for THEOLOGY and collocates within a distance of
five lexemes with a lexeme from the field of discourse for
HORSES, in order to test in what percentage of such occurrences
the attributive noun also alliterates with at least two other
lexemes (not function words) in the same line and, further, whether
this pattern is more common or less common in lines that draw on
Ovid's _Metamorphoses_.

It is probably possible to do such a search with a text heavily
marked up with SGML, and equally possible for me to read a
report set up in terms of the SGML markup.

HOWEVER, I want to use the computer to make my work easier, and
I find it easier to mark some patterns graphically rather than
in SGML; MOREOVER, I want to use the computer to make my
teaching clearer, and most of my students find a graphic
markup elucidates patterns and structures in a text,
whereas a cryptic linear gloss between words adds to their
confusion. I suppose the crux of the matter is that I want
my hypertextual environment to fulfil two functions: research
and communication, for my students as well as myself. There
are enough insidious political and budgetary wedges between
research and teaching, and I would like my computer to bridge
the gap rather than drive another wedge in.

What is needed is first to be able to search graphic layers
as well as text layers with a search engine that is capable of
jumping, at the appropriate juncture, to another layer and
contnuing the search, and then to be able to produce a
multi-layered report.

I have done enough elementary programming to think that such
a hypertextual/hypergraphical environment should be possible
and to know that I shall not create it in this lifetime. After
being informed last summer by my university's Tenure and
Promotion Committee that developing the first usable Old
English font for the Macintosh could not be considered
as much of a professional contribution as any one of my
first half dozen conference papers, I would hesitate to
embark on a much more ambitious computer project with much
less likelihood of success.

Well, Willard, how does my hypertextual fantasy compare
with yours?

Brian Whittaker BRIANW@VM2.YorkU.CA
Atkinson College, York University (Please do not omit the W)