Words and other lexical (and non-lexical) ranks

An important PARTOF relation is that of rank, as discussed above, a notion conspicuous in the work of Jespersen and Halliday, and in theories such as Stratificational Grammar. It is practically absent in self-styled `mainstream' theoretical linguistics. Sentences do not simply consist of words, for example; they consist of other sentence-like things like subordinate clauses and phrases. Words consist of word-like things, like other words, stems, roots, affixes, which, more subtly, also have different combinatorial properties. At each rank, there are stratified constituent structures represented by the bar levels in X-bar theory, for instance, at the ranks of sentence and word; at each rank, there are both lexical signs and non-lexical (transparently compositional) signs -- lexical signs are not restricted to words, and, conversely, compositional signs are not restricted to phrases (contrary to much current wisdom). The following is an initial approximation to the rank hierarchy, from largest to smallest, with characterisations of typical rank-specific structures:

Discourse: involves dialogue and social conventions about the relations between its parts; category structure: typically iterative (i.e. `flat recursion' without self-embedding).

Text: involves monologue, spoken or written, and rather inexplicit relations between sequential parts; category structure: typically either iterative or with trees of finite depth (e.g. chapters and sections of books), with somewhat more complex but finite extensions (such as tables of contents, bibliographies, cross-references).

Sentence: has complex and highly systematic relations between parts; typically recursive (i.e. permitting self-embedded structures of arbitrary depth):

Clause: contains finite verb, subject etc.

Phrase: contains no finite verb (in the sense of traditional grammar; in generative terminology, a non-finite verbal syntagma or group is often called a `verb phrase').

Word: with complex, highly systematic but highly inexplicit relations between parts; with respect to semantics recursive, with respect to phonology iterative (leading to `bracketing paradoxes' like transformational grammarian bracketing as (transformational grammar) ian with respect to semantic interpretation, transformational (grammarian) with respect to surface interpretation):

Inductive definition of `stem':
1. Base condition: a root is a stem (namely a simple stem).
2. Induction condition: A stem with an affix is a stem (namely a derived stem).
3. Exclusion condition: Nothing else is a stem.
There are more complex cases, e.g. phrasal stems as in `a very bring-back-home-able young man', but it is beyond the scope of this introduction to discuss these.
Root: a simple stem with no affixes;
Affix: an item which, when attached to a stem yields a stem (if derivational) and a word (if inflectional);
Word: an inflected stem;
Derived stem: one root with derivational affixes;
Compound word: a word preceded by a sequence of words or stems.

In the lexicographic context, it is relevant to note that each rank has both lexicalisation and combinatorial properties; at the text rank, for instance, an anthology of poems is entirely analogous to a lexicon of idioms or words at lower ranks.