Example records provide the easiest and directed illustration of why it is beneficial to move towards XML-style representations of the linguistic information with which we are concerned. XML supports technology that is being developed extremely rapidly. One aspect of this is the ability to provide flexible views of the structured information represented as XML; this technology is being built into standard web-browsers. This means that once information is available in XML-form, we can use the standard technology for presenting views of this information as required for particular tasks. An example of this, using the first stage of XML-representation described on this page, is given here.What is important is that as long as some tool produces XML-conformant representations of the kind illustrated, then all of the developed presentation styles can be applied, regardless of the original tool used for creating the XML form.
| Description |
The example records are the most complex of the currently envisioned representations. They combine information about all kinds of instantiated potential: i.e., syntagmatic structures (at various ranks), paradigmatic paths through system networks (for each rank) and strata. No currently available tool can support all of these kinds of information in combination, although selections from the full set are frequently handled. The idea of having a single representation for instantiations is that this representation can be used by various tools, each of which would take the components that it can deal with, leaving the rest. Eventually tools may be provided which can deal with the entire representation, although it is unlikely that any single tool will simultaneously satisfy all requirements. |
|
| Representations |
Schema are being proposed in two stages. The first is a simple representation already produced by KPML and combines syntagmatic and paradigmatic instantial information organised around a single tree structure. Representations conforming to this structure can be read by KPML and used to produce structure graphs, highlighted paths within system networks, links from structural constituents to paradigmatic options taken up and so on. Collections of such representations can be used to link to text instances of selected paradigmatic options, to provide text profiles and so on. Examples of representations conforming to this first stage representation are given here. The second stage employs XML technology more fully in order to provide full representations of multilayered, multistratal instantial representations. This captures metafunctional layering, where the layers may break up units independently of one another, as well as multistratal representations. For multilayering, standard XML stand-off annotations are used taking the the lowest represented rank of the stratum as baseline. For multistratal representation, standard XML techniques for alignment between text representations are used. Alignment as developed for, for example, parallel text corpora (as with parallel translations) provides a convenient way of linking between, for example, intonational units and lexicogrammatical units, without falsely enforcing primacy of one of the forms of representation over the others. The proposals for this much more sophisticated form of representation will be discussed here and input from all who have considered these issues is very welcome! |
|