"The Theory Proposed Here Is Rather Different"

Fawcett (2010: 286-7):

The theory proposed here is rather different. The key categories are class of unit, element of structure and item. But a 'class of unit' is defined by its internal structure, the major classes (of English) being the clause and the nominal, prepositional, quality and quantity groups.  

Moving down the layers of a tree diagram representation of a text-sentence, we find that 'unit' and 'element' occur alternately (these being related to each other by the similarly alternate relationships of componence and filling), until the lowest element in the tree is reached and the relationship of exponence relates that element to an item.  

There is no place in the formal representations for the concept of 'word class' (although terms such as "noun" and "adjective' are used as convenient short forms for referring to classes of item that are ultimately defined by the part of the system network from which they are generated). 

To this core framework must be added the general concept of probability. More specifically, the theory provides that the likelihood that a given unit will fill a given element should be expressed in probabilistic terms (as well as absolute terms where it has a zero probability). The claim is that probabilistic statements about the potential of each class of unit to fill an element are more accurately predictive than the 'rank scale' predictions — and so more useful when the theory is being employed for the analysis of text-sentences (whether by a human or by a computer).

Blogger Comments:

The regular reader would long ago have noticed the extent to which Fawcett just keeps on repeating the same claims over and over and over. This is a deployment of the logical fallacy known as the argument from repetition, also known as argumentum ad nauseam.

[1] As previously explained, this is taking the (formal) view 'from below', and is contrary to the (functional) view 'from above' that is taken in SFL Theory.

[2] As previously explained, a clause is a unit, not a class of unit. A class of this unit is the traditional notion of an adverbial clause.

[3] To be clear, Fawcett's tree diagram combines his (vigorously denied) rank scale of sentence–clause–group & cluster–item with his proposed relations between these formal units and elements of function structure. 

The notion of componence misconstrues a formal unit as composed of functional elements; in SFL Theory, a formal unit is composed of lower ranked formal units, each of which realises an element of function structure in the higher ranked unit. 

The notion of filling corresponds, in SFL theory, to the relation between elements of clause structure and the syntagm of groups ± phrases that realise them, and the notion of exponence corresponds to the relation between elements of group and cluster structure and the words ("items") that realise them. 

And, as previously explained, Fawcett's notion of item confuses the grammatical and lexical notions of word, and misconstrues the meronymic relation between words and morphemes as co-hyponymy (words and morphemes as subtypes of item).

[4] To be clear, Fawcett does not supply any system networks that demonstrate how classes of item are generated.

[5] To be clear, this is still a bare assertion, still unsupported by evidence. Moreover, what Fawcett refers to as "rank scale predictions" is merely the rank scale itself: the modelling of formal constituency as clauses consisting of groups ± phrases, consisting of words, consisting of morphemes.