Gripes about links in a paper dictionary

December 2000, English revision 28.08.04

Thinking about the link structure of encyclopaedias and dictionaries (#06), I took a closer look to the German etymological dictionary, Duden Volume 7 in its 2nd edition. Particularly, I examined the word family of 1Ball (comprises the German equivalents of ball; bantling, soon, deal, balloon, bull, to blossom, blood, bale, billow; Leopold, bar, phalanx, bulwark, flower, blossom, leaf, bloom, nap, foil, chlorophyll, blow, pock; budget, cushion, drunkard, balcony, plank, boulevard, flowery phrase, flora, flourish, floret, folio, feature pages).

The link structures were completely unsystematic/ inconsistent (see foil 1)

  • Back links (from child node to parent node of the article structure) are often missing (e. g. Phalanx > Balken)
  • Forward links from parent node to child node are often missing (e. g., blasen > Inflation).
  • transitive references (from grandparent node to grandchild node, e. g., from 1Ball to almost every children of blühen except Blust), and vice-versa (from Polster directly to Ball) are sometimes present, sometimes missing.
  • some word clans do not have a chief/head, and are always simultaneously referenced: blasen / blähen / Blatter, or Blaken and Bohle.
Alternative: foil 2.

Marking of links is completely inconsistent / unsystematic, or the systematics is not intelligible: with or without reference arrow, with or without italicizing, with or without words like "vgl." or similar (foil 3).

In any case, the various link markings do not seem to mark the differences between references from/to parent nodes, and cross references like

  • similar development of meaning (bald like schnell, Flora like Fauna, Folio like Oktav)
  • similar word-formation (blühen to Blume like säen to Same, blühen to Blüte like säen to Saat)
  • Vocabulary similarities (Folio like Format and Exemplar, Florett like Finte and 1parieren)
  • unrelatedness (1Flor "flowers abundance" vs. 2Flor "texture")
  • parent nodes outside the present word clan (e. g., with composites like Chlorophyll to Chlor, wechselbalg to Wechsel).
  • cross references within the present word clan (the meaning "leaf in book" was influenced by folium, see Folio).
The following link type marking suggests itself (foil 4):

up arrow, symbol font character #173, only for references to parent nodes,
down arrow (#175) just for references to child node, and
right arrow (#174) for genuine cross references.

(PS 2004: in the meantime, http://www.hrz.uni-dortmund.de/~hytex/texte/hyper2/pfeile.html shows something similar)

The selection of links could be cleaned-up, if the second-purpose of constructing an overview were achieved by dedicated visualization efforts (like this sample) rather than via attempts with compromises in text design.

Already the mere text presentation could contribute to visualization, e. g., bullet lists for meaning groups (blühen, blasen).

Also, the text itself could be cleaned from countless duplications if link handling were more consistent (e. g. with blähen and Blatter, the identical passage shows up "eng verwandt mit den unter blasen und ... behandelten Wörtern und gehört mit diesen zu der unter 1Ball dargestellten Wortgruppe".) If the (alledgedly always meagre) space were thus saved it could them be used for bullet structuring.

(At some point one arrives at the argumentation that many peculiarities of a dictionary are indispensable for the alphabetical accessability, and at some point one must in fact question the very alphabetical order. If a machine readable copy is present, why not arrange the paper text or also of the printer-friendly e-text sections by subject, instead ?)

Additional desirable visualisations include:

  • A more detailed "map" that does not restrict to the article web with the most-needed meaning hints and, even though it is no longer displayable on the computer screen, it supports navigation from a printed copy.
  • A real (geographic) map that facilitates the layman's overview of the words' derivation paths that are otherwise only gatherable in a cumbersome and error-prone fashion, including: idg. > griech. > klass. lat. > vlat. > gall.-lat. > mlat.; plus afränk. > afrz. > frz. > engl. or > langobard. > it. Nice to have would be sample words and arrows, colored according to age with rainbow-colors from purple (very old) to red (modern), positioned on the continent.
So that's my wish list for a more overview-focussed organization of dictionary information.

Back