Creating natural language item descriptions
In most taxonomic monographs, in addition to the information provided in dichotomous or synoptic keys, each taxon is described in a coherent text, which may look like the following example:
Stromata small, superficial, orbicular to elongate, at first gray, later black, 1-3 x 1-2 mm and 0.5 mm high; perithecia semiglobose, 0.5-0.6 mm in diam. with papillate ostiola; asci cylindric, 80-90 x 7-11 µm; ascospores uniseriate, elliptical, with obtuse ends, dark brown, often with large oil drop in center, 15-18 x 5-8 µm; paraphyses present.
The creation of natural language descriptions from a descriptor project is a relatively straightforward process. For each character present in a given item, the natural language phrase for the character (CharPhrase, if missing CharName can be used instead) is output and the character state name is appended to it. In the case of numeric characters, the numeric values and the optional Unit is appended to the character phrase. If several character states are present, the states are appended with or, and, or to as defined in the MultistateType attribute. If the states present in the current item all end with the same words, these words can be omitted except for the last state. Similarly, if several characters start with the same words, and are defined as a LinkGroup, the start of the characters is abbreviated. Any text in larger/smaller signs (<>) is omitted in the natural language description (this is the only place where DeltaAccess respects this DELTA convention, otherwise it resolves all comments during import and stores them in appropriate attributes).
Thus, if all characters in the following example form a single link group:
#1. leaves <hairiness>/
1. not hairy/
2. sparsely hairy/
3. densely hairy/
#2. leaves <shape>/
#3. leaves <length>/
and an item includes the following DELTA codes:
1,2/3 2,1 3,12-20
it would be rendered as:
leaves sparsely or densely hairy; linear; 12-20 mm long
A simple version of such text can be automatically produced from a structured DELTA data set and is called a natural language item description.
The export format Natural language description creates a simple natural language item description, formatted as Rich-Text-Format (RTF). RTF can be read by Microsoft Word and most other word processors under either the Windows or the Macintosh operation system. The natural language export is currently not very sophisticated and rather a "technology preview". I am aware of some places where the output is less than satisfactory. If you are interested in using this feature, you can help improving it by pointing out what it does, and what you expect it to do.
The main reason for this feature is that I want to use it in the next version of DeltaAccess during interactive identification. At present, for generating better natural language descriptions or printed dichotomous keys, I recommend using M. Dallwitz’s Confor or R. Pankhurst's Pankey (see DELTA software sources on the internet).
The following attributes (compare the CharEdit form) are evaluated mainly during Natural language description. Follow the links to obtain more information on an attribute:
Return to Export Main table of contents Next