A NOUN models a person, place or thing, or so we learn in school. For modeling, we need to get more detailed and precise, so effectively, this tree splits the world of NOUNS into four new parts of speech, each of which requires a unique set of additional features (TBDL) to properly describe each contained sense.

Other parts of speech may be subclassified here too, by using special-case rules. For a VERB, one first finds its gerund sense, then classifies that. The result will always be some kind of Situation, depending on whether or not the verb was copulative, continuing, recurring, etc. For a PREPOSITION, one should first mentally translate it to an equivalent VERB sense, then reuse the rules above on the result.

Modifiers fall into a separate division all their own - the Aspect - under more complex rules that will apply to any NOUN sense whose meaning is a part or attribute of another type of thing - its domain. We categorize these sub-types based on what domain they modify or partition. Modifiers and parts clearly differ, but both get used similarly in language because they have a domain. Most other types do not need one. (The collective is a special-case exception.)

The Aspect category of highest interest involves parts of a Situation. They appear naturally as roles in XTM-style associations, and also in case frames. In English, they (or synonymes) often have regular spellings built as a root VERB inflected by -er or -ed. Their roleplayer topics are also explicitly marked in every English sentence by special positions or PARTICLE/PREPOSITION spellings, which makes parsing them relatively easy if the case frame for the specific Aspect at issue is known. (Unfortunately, these rules are not regular, so they can only be defined sense-by-sense, as a data base covering all the verbs in English.)