Some Problems With Martin's Notion Of A 'Subjacency Duplex Structure'

Construing entities: types of structure

J. R. Martin

In this talk I address some issues arising from recent SFL work on language description, focusing on types of structure. In particular I will look at SFG's traditional distinction between multivariate and univariate structures and their association with non-recursive and recursive systems respectively (Halliday 1981, 1979). Drawing on descriptions of nominal group structure (in English, Tagalog, Spanish and Korean), I'll suggest that the association of multivariate structure with non-recursive systems and univariate structure with recursive systems needs to be relaxed. Doing so makes room for recognition of non-iterative dependency structure, which I'll refer to as subjacency structure (first foregrounded as duplexes in Rose's work on Pitjantjatjara; 2001) – a structure which can be usefully applied to the analysis of what are often fudged as 'structure markers' in SFG descriptions (i.e. adpositions and linkers).

Halliday, M A K 1981 (1965) Types of Structure. M A K Halliday & J R Martin [Eds.] Readings in Systemic Linguistics. London: Batsford. 29-41.

Halliday, M A K 1979 Modes of meaning and modes of expression: types of grammatical structure, and their determination by different semantic functions. D J Allerton, E Carney, D Holcroft [Eds] Function and Context in Linguistics Analysis: essays offers to William Haas. Cambridge: Cambridge University Press. 57-79

Rose, D 2001 The Western Desert code: an Australian cryptogrammar. Canberra: Pacific Linguistics.

Blogger Comments:

[1] To be clear, the notion of 'entity' in semantics is from Halliday & Matthiessen (1999: 172):

The nominal group, as we have seen, construes an entity — something that could function directly as a participant.

[2] To be clear, on the one hand, Martin takes Halliday (1965) as his point of departure rather than more recent publications. On the other hand, it is simply not true that Halliday (1965) equates the univariate/multivariate distinction with the recursive/non-recursive distinction. Halliday associates univariate structures with lineal recursion and multivariate structures with both cyclical recursion and non-recursion. Halliday (1981 [1965]: 45):

[3] To be clear, the term 'subjacency' is from Chomsky (1973), and the term 'duplex' is from Matthiessen (1995: 161) and simply means a univariate complex of two units. In his talk, Martin attributed 'duplex' to Rose, instead of Matthiessen, and falsely claimed that it occurs all through Rose's thesis, whereas, in fact, it only occurs 4 times (pp367-8, 400), and is only applied to verbal group complexes.

More importantly, Martin's notion of a 'subjacency duplex' structure rests on the claim that it is 'non-iterative' — meaning that the structure cannot be further modified through lineal recursion. The final example that Martin used to illustrate this was:

To be clear, the reason why this 'subjacency duplex' cannot be further modified is that it is not a duplex. That is, it is not a 2-unit complex: it is neither a nominal group complex nor a 'clitic complex', and the clitic neither modifies (subcategorises) the nominal group — see Halliday & Matthiessen (2014: 389) — nor hypotactically expands the nominal group in terms of elaboration, extension, or enhancement.