All the specialists are agreed that the possibility of adding to multimedia objects some sort of "conceptual" annotation describing their information content under a form suitable for computer processing would contribute greatly to solve the problem of their "intelligent" indexing and retrieval. The existing projects in this domain, like Information Manifold, UNTANGLE or MIHMA, propose to build up the conceptual annotation (or "generalised bookmark") making use of very complex conceptual languages, like some sort of description logics. The major drawback of this approach concerns the difficulty of associating "by hand" a (syntactically very complex) generalised bookmark in the description logics style to WWW objects pertaining to a multimedia universe sufficiently large. To alleviate this problem, we propose -- in the context of a newly started project which concerns the establishment of detailed blueprints (and of a prototype) of a full object-oriented, WWW Distance Learning Environment -- to associate with the WWW objects not the final conceptual annotation, but a simple natural language (NL) caption in the form of short texts representing a general, neutral description of the informational content of the object. To accelerate this sort of operation, it could also be possible to make use of dictation systems like Dragon Dictate or IBM VoiceType. The NL caption should then be converted into a conceptual annotation in NKRL (Narrative Knowledge Representation Language), making use of an automatic "translation" system like those we have recently implemented in the context of projects partially financed by the European Union.