Appendix talk:Index/American Sign Language

Organization
The main purpose of Index:American Sign Language is arguably to help the reader who has seen an ASL sign produced but does not yet know the sign's meaning. Unfortunately, certain features of a sign vary from one speaker to the next, and it can be difficult to remember certain specifics of a sign's production upon initial exposure. To help with such challenges, User:Positivesigner has proposed a new organization scheme for the ASL index, based on ASL Sign Jotting (ASLSJ), a system he created. (A bonus of Tom's reorganization effort is the introduction of SignWriting symbols, e.g. .)

Considering our "minimize OR" guideline, I am seeking to reconcile ASLSJ with existing sign language phonetics research. Some aspects of it seem compatible with the "Hand Tier" models that are popular among sign language linguists (e.g. Sandler, van der Kooij, and van der Hulst). Such models organize sign features into a dependency model where the "selected fingers" feature is at the head of the hand configuration. Other features, including the aperture feature (a.k.a. finger position, e.g. open, closed, bent, curved) are modeled as dependent on the selected fingers feature.

The proposed index organization groups the following shapes into a single top-level group:

Viewed from the perspective of the Hand Tier model, that grouping makes sense because those handshapes have all of the fingers selected. The proposed index organization also groups together, , , and. The Hand Tier model agrees that 9 and SmallO are similar shapes, but an index organization based primarily on Hand Tier models would probably have grouped O and FlatO in with 4/5/B/C.

Anyway, I can do more research on the various Hand Tier models to see how compatible the proposed index organization is with existing published points of view, but I'm excited about the possibilities suggested by this reorganization effort. Anyone's feedback is welcome. —Rod (A. Smith) 21:21, 27 April 2009 (UTC)

Rolling this around for awhile, I think we should use ASLSJ as a set of guidelines for crosslinking between phonetically similar index pages. That is, ASLSJ creates a rough characterization of each sign by collapsing phonetically similar ASL phones into "super-phonemes". That's useful because it ignores subtleties that a reader may not notice upon encountering a sign. If a reader is looking for a sign made with a handshape similar to 1, but doesn't know whether the thumb was also extended, it could be helpful to let the reader start searching under the index page for signs with the 1 shape, help the reader browse to the section with signs made in the recalled location, and then show "see also" links to some alternate places to look for the sign based on other phones of the super-phoneme "DT". This approach would make alphabetization very straight-forward and still would help readers who are uncertain about certain subtleties of the sign's production. —Rod (A. Smith) 16:49, 21 September 2009 (UTC)

User:Positivesigner explained by e-mail how collation works in ASLSJ. It characterizes each sign according to the following seven super-phoneme groups:


 * 1) Dominant hand handshape
 * 2) A-I-S
 * 3) 1-D-T
 * 4) 3-H-V-N
 * 5) W-M
 * 6) F-O
 * 7) 4-5-B-C
 * 8) Hand orbit
 * 9) One-hand "only" for one-handed sign / fingerspelling
 * 10) Two-hands "side-by-side"
 * 11) Two-hands "swapped-sides" like LOVE and BUTTERFLY
 * 12) Two-hands "above-below" with dominant hand in either position
 * 13) Two-hands "inside-outside" with dominant hand in either position
 * 14) Dominant hand primary movement axis, regardless of repetition or circling
 * 15) Dominant hand "rotates" at wrist while possibly changing handshapes without any axis movement
 * 16) Dominant hand moves "in-out" along depthwise axis
 * 17) Dominant hand moves "left-right" along horizontal axis
 * 18) Dominant hand moves "up-down" along vertical axis
 * 19) Dominant hand repetition style
 * 20) Dominant hand moves "once" along axis or fingerspelling
 * 21) Dominant hand moves "repeatedly" along one axis
 * 22) Dominant hand moves "circling" repeatedly to another axis
 * 23) Hands' movement style
 * 24) Sign uses "one-handed" movement, regardless of hand orbit, or uses fingerspelling movement
 * 25) Both hands move "in-parallel" with each other
 * 26) Both hands move "split-toward" each other
 * 27) Dominant hand moves circling along an axis with the non-dominant hand "following-behind"
 * 28) Dominant hand body height
 * 29) Dominant hand starts and stays in "neutral" space, without touching the body or referring to a part of the body by elevation
 * 30) Dominant hand starts on or ends up touching the "chest"
 * 31) Dominant hand starts on or ends up touching the "chin" / cheek / jaw, or refers to this area by elevation (FEED YOU)
 * 32) Dominant hand starts on or ends up touching the "eyes" / nose / ears, or refers to this area by elevation (LOOK AROUND)
 * 33) Dominant hand starts on or ends up touching the "forehead" / temple, or refers to this area by elevation (SNOW)
 * 34) Dominant hand body distance
 * 35) Dominant hand starts and stays in neutral "space", without touching the body or referring to a part of the body by elevation
 * 36) Dominant hand refers to a part of the body by elevation "away" from the body
 * 37) Dominant hand touches "against" or very close to a part of the body

It groups together all signs with the same seven super-phonemes (e.g. it groups together all signs where the dominant hand makes a 1, D, or T shape, the nondominant hand is not involved, the hand rotates or does not move, and the other four super-phonemes match). Within any given combination of those seven super-phonemes, it sorts signs according to the actual phones. The justification for the complex collation is that a reader may be looking for a sign made without first knowing the sign's meaning or precise production details.

On my talk page Di gama suggests a different sorting approach:
 * 1) First posture dominant handshape group (optional)
 * 2) First posture dominant handshape
 * 3) First posture dominant height
 * 4) First posture other dominant location parameters
 * 5) First posture dominant palm facing
 * 6) First posture nondominant handshape group (optional)
 * 7) First posture nondominant handshape
 * 8) First posture nondominant height
 * 9) First posture other nondominant location parameters
 * 10) First posture nondominant palm facing
 * 11) Subsequent postures (same order as all above)
 * 12) Transitions

It's good to facilitate searching for a sign using fuzzy search criteria, but the ASLSJ collation system is quite complex. It's hard to imagine ever getting it right, and easy to imagine a bewildered reader. So, I suggest we sort signs by something like Di gama's straight-forward approach, organizing the ASL index by the actual phones in the sign:


 * 1) First posture dominant handshape
 * 2) First posture dominant location
 * 3) First posture dominant orientation
 * 4) First posture nondominant handshape
 * 5) First posture nondominant location
 * 6) First posture nondominant orientation
 * 7) Subsequent postures and movements

To assist with fuzzy searching, we can define the phone ordering such that phones for a given ASLSJ super-phoneme sort near each other. That is, put all of the signs made with the A shape together, followed by all of the signs with the OpenA shape, followed by all of the signs with the S shape, followed by all of the signs with the I shape. Similarly, we can define the sort order of locations such that those with similar height sort near each other.

Then, we can add "see also" links throughout the index achieve the searchability goal of the ASLSJ collation approach. For example, a reader who knows that the index finger is extended in a particular sign but doesn't know whether the thumb is also extended could start looking at the index page for the shape 1, in the section for the relevant location and orientation. "See also" links from there would show some alternate places to look for the sign based on the other phones in the ASLSJ super-phoneme. —Rod (A. Smith) 19:59, 25 September 2009 (UTC)


 * The idea behind my suggestion was to make a system which can be figured out (by a bot) by analyzing the entry name. Although the ASLSJ approach is very phonemic and complete, not all the parameters can be inferred from the entry name, which makes it difficult for a bot to upkeep. A solution to this is to put a template on every page to describe their position in the index, which would solve the problem which I have yet to bring up, namely the templatization of the ASLSJ lists (which are nigh-on unreadable in wiki-text right now, and make the ASLSJ indices very difficult to use). I was planning on doing it myself, but since this is all very clearly in flux, I don't want to make any permanent changes.


 * As for the "see also" lists, I'm confused: don't we already have this? I didn't think they were strictly based on these super-groups, but then, I haven't been looking very closely. Are you suggesting more integrated see-also lists (like top-of-head A to top-of-head S section links) or are we just talking about the ones at the page-bottom? in which case I don't see the difference. —Di gama (t • c • w) 02:28, 28 September 2009 (UTC)


 * Yes, bot-driven maintenance is important. Adding sort keys manually to individual entries is too prone to error for my comfort, though.  So, I'd rather we just define a sort order for the phones in the entry names themselves and collate the index based entirely on those phones.  We have limited use of "see also" links now.  They currently link only between similar handshapes and are only at the main level of each index.  For a more robust fuzzy-browsing solution, we'd also need to allow "see also" links to be added to each second- and third-level section of the index.  For example, a "see also" may link from the A@InsideChesthigh section to the A@InsideSternumhigh section.  —Rod (A. Smith) 15:49, 28 September 2009 (UTC)


 * I see your point, but there appear to be logistic issues. For example, if we had a "see also" list at the third level, following your proposed order above, we would have:
 * 3 to 11 other-handshape links
 * 5 or so to very many other-location links (depending on how we group these; the worst case "very many" is in the hundreds at least, but I doubt we'll use that)
 * 6 to 18 other-orientation links (depending, likewise, on the grouping)
 * Even the conservative choices in this list would be a considerable number, and since these show up at every third-level section, that is quite a few see-also's. Of course, not all inter-section links would exist at early stages (like now), and the entries themselves would dwarf the see-also links at later stages, but it would still seem excessive. That is, unless I've misunderstood your suggestion. —Di gama (t • c • w) 00:41, 30 September 2009 (UTC)
 * Offhand, your estimates seem a bit high to me, but not terribly so. You may be right.  I should create an example of what I mean by editing one of the existing index pages.  If that example appears to be inundated with "see also" links relative to actual entries, I'll adjust or abandon this idea.  —Rod (A. Smith) 16:00, 2 October 2009 (UTC)

Photo recommendation
I would like some feedback about the dual-view ASL phoneme picture I uploaded. It really helps me to understand the depth aspect of the hand placement. -Positivesigner 06:31, 17 June 2011 (UTC)

Open8@Chest-PalmBack-Open8@Chest-PalmBack_Open8@SideNeckhigh-PalmBack-Open8@SideNeckhigh-PalmBack