Yaegan Doran On Structure Types
A meticulous examination of 'Factoring Out Structure: Nuclearity, Linearity And Iteration' (Doran 2021)
07 November 2024
25 November 2021
Overall Assessment
Blogger Comments:
Doran's paper is not concerned with modelling language, but with modelling one aspect of theory: SFL's model of structure.
Because Systemic Functional Linguistic Theory gives priority to the view 'from above', structure is seen as a means of realising meaning, with structure types varying with the metafunctional meaning they express. Halliday & Matthiessen (2014: 85):
Doran, however, is not concerned with structures as realisations of meaning, but with structures only, and his approach is to take a 'distinctive feature' ("factor") approach to classifying structures, the approach taken to classify phonemes in Formal phonology, limiting his attention to structure types favoured by the ideational metafunction.
The factors (features) he derives are [±iterative], [±nuclear] and [±linear]. Deriving the factor [+iterative] involved confusing iterative structures (logical metafunction) with segmental structures (experiential metafunction) that feature more than one occurrence of an element, such as Epithet.
The factors [+nuclear] and [+linear], however, unknown to Doran, are actually derived from Martin's misunderstanding of metafunctional structures, where Martin represents segmental structures as if hypotactic ("mono-nuclear"), and iterative structures as if paratactic ("multi-nuclear"):
Doran derives both [+nuclear] and [+linear] from Martin's orbital structure, with [+nuclear] corresponding to the relation between nucleus and satellite, and [+linear] corresponding to the relation between satellites.
The outcome of this distinctive feature approach is a table with cells that variously include whole structures, partial structures, and non-structures (cohesive relations, subjacency duplexes), with no suggestions as to the functions of the factors that classify them.
For the arguments on which these conclusions are based, see the individual posts, below.
18 October 2021
[45] Some Problems With Non-Iterative/Non-Nuclear/Linear Structures
To be clear, here Doran is looking for structures that fit the combination of factors that specify the last unfilled cell of his table.
Ignoring all the previously noted problems with the factors themselves, by the logic of the table, Doran should be looking for a structure that is:
- the non-nuclear counterpart of subjacency duplexes, and
- the non-iterative counterpart of covariate lexical relations, and
- the linear counterpart of relational clauses without a Process.
He tentatively suggests:
- correlative conjunctions in English, or
- clitic doubling in Spanish, or
- full reduplication in Sundanese.
There is no argument as to how these satisfy the three requirements above, and in the case of correlative conjunctions — which do not, in themselves, constitute a structure — the structure that they mark is a paratactic clause complex, a univariate structure, which, in terms of Doran's factors is:
- iterative rather than non-iterative, and
- non-linear rather than linear.
17 October 2021
[44] The Problem With Non-Linear/Non-Nuclear/Non-Iterative Structures
Doran (2021):
Blogger Comments:
16 October 2021
[43] Problems With 'Univariate', 'Covariate (Non-Structural)' And 'Subjacent'
Blogger Comments:
To be clear, the additional claims here are that
- parataxis, hypotaxis and phoric covariate are univariate structures,
- expectancy covariate, hypotaxis and phoric covariate are covariate and non-structural, and
- subjacency duplexes are subjacent.
15 October 2021
[42] Self-Contradiction: Multivariate As Nuclear
Blogger Comments:
See Martin's Orbital Model As Univariate Representation Of A Multivariate Structure
14 October 2021
[41] The Variations In Structure Afforded By Doran's 3 Factors
Blogger Comments:
[1] To be clear, this table presents the four cells of the preceding table as 'iterative'.
In the case of the top cell of the 'iterative' column, [non-linear, non-nuclear], this is consistent with SFL Theory, since parataxis is one type of interdependency relation in iterative structures of the logical metafunction.
In the case of the second cell of the 'iterative' column, [non-linear, nuclear], this is not consistent with SFL Theory, because, as previously explained, Doran's 'iteration' confuses multiple occurrences of (experiential) elements — Epithets, Qualifiers, circumstances — with iterative structures of the logical metafunction.
In the case of the third cell of the 'iterative' column, [linear, non-nuclear], this is not consistent with SFL Theory, not least because 'covariate' is not a type of structure, as Lemke (1989) acknowledges.
In the case of the bottom cell of the 'iterative' column, [linear, nuclear], this would be consistent with SFL Theory, if hypotaxis had been what Doran argued for, since hypotaxis is one type of interdependency relation in iterative structures of the logical metafunction. However, Doran's argument for this classification was actually concerned with the relation between Classifiers in the experiential structure of the nominal group.
This bottom cell also features a category for which there has been no argumentation, 'phoric covariate'. Again, the problem here is that 'covariate' is not a type of structure, and so not iterative in a structural sense.
[2] The 'non-iterative' column of the table has only two of its four cells filled.
The upper cell, [non-linear, nuclear], is consistent with SFL Theory in as much as experiential structures are multivariate and so non-iterative.
However, the lower cell, [linear, nuclear], identifies the subjacency duplex. On the one hand, there has been no argument for classifying this structure as linear, and on the other hand, the subjacency duplex is not a structure, not least because it is based on misunderstandings of constituency. See, for example, the previous posts:
13 October 2021
[40] Problems With Linearity x Nuclearity
[5] In short, Doran has here categorised three of Martin's misunderstandings of structure types — covariate, orbital, serial — in terms of distinctive features. In doing so, he essentially provides a flawed system to specify classes — not functions — in metalanguage — not language — without regard to the metafunctions that the structure types express.
12 October 2021
[39] Problems With The Factors Of Covariate Structures
Doran (2021):
Blogger Comments:
[1] To be clear, as previously demonstrated, Doran's linearity corresponds to the relation between satellites in Martin's orbital (mono-nuclear) model of structure, which, in turn, misrepresents the multivariate structure of the experiential metafunction as a univariate hypotactic structure of the logical metafunction.
[2] To be clear, in Doran's own terms, it could be argued that there is indeed "difference in status" in this example: between the whole (dog) on the one hand, and the parts (mouth, teeth, neck), on the other.
(Note that conformation is not a part of body or dog, since it refers to the form or shape of the dog.)
11 October 2021
[38] Problems With Lexical Relations As Structures
Blogger Comments:
10 October 2021
[37] The Problem With Covariate Structures
Blogger Comments:
“discourse” is text that is being viewed in its sociocultural context, while “text” is discourse that is being viewed as a process of language.
The system and the text are not two different phenomena: what we call the “system” of a language is equivalent to its “text potential”. Analysing discourse means, first and foremost, relating the text to the potential that lies behind it.
A text is meaningful because it is an actualisation of the potential that constitutes the linguistic system; it is for this reason that the study of discourse (‘text linguistics’) cannot properly be separated from the study of the grammar that lies behind it.
The “textual” metafunction is the name we give to the systematic resources a language must have for creating discourse: for ensuring that each instance of text makes contact with its environment. The “environment” includes both the context of situation and other instances of text.
My own 'covariate structure' (Lemke 1985), which includes Halliday's univariate type, is for the case of homogeneous relations of co-classed units, and should perhaps be called a 'structuring principle' rather than a kind of structure.
Martin's covariate structures are not structures in the sense of units with internal structure, nor in the sense of units forming complexes.
09 October 2021
[36] The Non-Structural Relations Of Cohesion
Doran (2021):
Blogger Comments:
[1] To be clear, because cohesive relations are non-structural, they do not qualify as a type of structure.
[2] To be clear, for Halliday & Hasan (1976), a text is not a semantic unit with respect to the lexicogrammatical systems of cohesion. However, in terms of semantic systems, a text is a structured semantic unit. Halliday (1985: 318):
A text has structure, but it is semantic, not lexicogrammatical.
[3] To be clear, in SFL Theory, cohesive relations obtain in the unfolding of text, as patterns of instantiation; see Halliday & Matthiessen (2014: 594ff).
[4] To be clear, in præsentia means 'for the present'.
[5] To be clear, cohesive relations are neither syntagmatic relations (see [3]), nor types of structure.
08 October 2021
[35] Cohesive Relations / Covariate Structures
Blogger Comments:
My own 'covariate structure' (Lemke 1985), which includes Halliday's univariate type, is for the case of homogeneous relations of co-classed units, and should perhaps be called a 'structuring principle' rather than a kind of structure.
07 October 2021
[34] Problems With The Three Factors: Iteration, Nuclearity And Linearity
Doran (2021):
Blogger Comments:
[1] To be clear, as previously demonstrated:
- 'iteration' confuses multiple occurrences of multivariate elements with iterative structure;
- 'nuclearity' corresponds to a hypotactic relation between Martin's nucleus and satellite;
- 'linearity' corresponds to a hypotactic relation between Martin's satellite and satellite
[2] To be clear, on this model, the experiential structure of a nominal group like new two-storey brick house involves all three factors:
- iteration: 2 Classifiers;
- nuclearity: Thing–Epithet "status distinction" (logical hypotaxis);
- linearity: Classifier–Classifier "interdependency" (logical hypotaxis).
06 October 2021
[33] The Basis Of Doran's Nuclearity vs Linearity Distinction
[1] To be clear, although it is not acknowledged here, Doran's distinction between nuclearity (nuclear dependency) and linearity (linear dependency) actually draws on Martin's notion of orbital structure:
Doran's nuclear dependency corresponds to the relation between nucleus and satellite (e.g. Thing and Epithet) in Martin's model, whereas his linear dependency corresponds to the relation between satellite and satellite (e.g. Classifier and Classifier).
As previously explained, Martin's orbital model misconstrues experiential structure as a hypotactic logical structure, with nucleus as dominant and satellite as dependent. So Doran's model continues this misapplication of hypotaxis to a multivariate structure. However, Doran compounds the error by proposing two different types of structure within the same structure:
- nuclear dependency: Thing–Epithet
- linear dependency: Classifier–Classifier
05 October 2021
[32] Some Problems With Doran's Notion Of Linear Dependency
Blogger Comments:
… for the purposes of the nominal group we need to take account of just one such relationship, that of subcategorisation: ‘a is a subset of x’. … The basis of the subcategorisation, of course, shifts as we move to the left: ‘what type of ...?’, ‘what quality of ...?’, ‘how many ...?’ and so on – this is the principle underlying the experiential structure.
So in Doran's example:
- what type of neutrinos? electron neutrinos.
- what type of electron neutrinos? solar electron neutrinos.
[4] To be clear, there are two degrees of interdependency: hypotaxis and parataxis. The relation that obtains in the logical structure of a nominal group is hypotaxis, and the relation between the two Classifiers in Doran's nominal group is thus hypotactic. In terms of Doran's own model, this makes it a relation between two satellites in a nuclear structure, not a relation of linear dependency.
04 October 2021
[31] Some Problems With Doran's Notion Of Nuclear Dependency
Blogger Comments:
[1] To be clear, before Doran, Halliday (1985: 167) identified the Thing as 'the semantic core of the nominal group'.
[2] To be clear, in SFL Theory, in the experiential structure of the nominal group, all the other elements have 'the function of characterising the Thing' (Halliday 1985: 167). The Thing specifies the class of thing, and the other elements specify some category of membership within this class (Halliday 1985: 160).
[3] To be clear, the first nominal group is the artificial invention of a non-native speaker of English, and the second is a reworking of it by Doran. To the extent that the Epithets can be swapped in this unnatural nominal group, it is because they vary little in terms of 'permanence'. Halliday (1985: 166):
By and large, the more permanent the attribute of a Thing, the less likely it is to identify it in a particular context. So we proceed with the very impermanent, quantitative characterisation that is nearest to a Deictic, e.g. three in three balls; through various qualitative features such as new in new ball; and end up with the most permanent, the assignment to a class, e.g. tennis ball. Within the qualitative characteristics, if more than one is specified there is, again, a tendency to more from the less permanent to the more permanent; e.g. a new red ball rather than a red new ball.
On the other hand, the reason why the prepositional phrases serving as Qualifier can be reversed is because they form a paratactic complex, as shown by the fact that either can be omitted.
[4] To be clear, a structure is a structure of a whole unit. Incongruously, Doran here proposes different structures obtaining between different elements within the structure of the same unit. That is, he proposes one structure for the relation between Epithet and Thing, and between Qualifier and Thing, but another structure with regard to the Classifier. (see the next post).
[5] To be clear, Doran's notion of nuclear dependency misconstrues the elements that characterise the Thing, Epithet and Qualifier, as elements that are dependent on the Thing — simply because they relate to the Thing. This misunderstanding derives from confusing interdependency (hypotaxis) with the 'dependency construct' in Halliday (1979):
03 October 2021
[30] Recapping And Looking Ahead
- Martin's orbital interpretation is a univariate representation of a multivariate structure.
- Doran distinguishes Martin's orbital structure from multivariate structures.
02 October 2021
[29] Misconstruing Orbital Structure As Not Multivariate
Blogger Comments:
[1] To be clear, here Doran misconstrues Martin's two different ways of representing one type of structure — the multivariate structure of the experiential metafunction — as two different types of structure, treating the tree schema representation as multivariate, and the co-tangential ellipses representation as orbital (but not multivariate).
The reason why both are representations of multivariate structure is because both represent the same elements of structure and both represent 'a configuration of elements each having a distinct function with respect to the whole' (Halliday & Matthiessen 2014: 390).
[2] To be clear, there are problems with both representations in Martin's Figure 7. Firstly, contrā Doran, the co-tangential ellipses representation does not represent an orbital structure, because it does not represent the non-nuclear elements as satellites of a nucleus.
Secondly, the tree schema representation misunderstands grammatical constituency. In SFL theory, grammatical constituency is modelled as a rank scale of forms, such that clauses consist of groups ± phrases, which consist of words, which consist of morphemes.
So, in Figure 7, the whole is the clause, the parts of the clause are the groups that constitute the clause. The structure, on the other hand, is the relationships between experiential functions assigned to the syntagm of groups.
The relation between a whole and its parts is composition (extension), whereas the relation between function and form is realisation (elaboration + identity).
01 October 2021
[28] Martin's Orbital Model As Univariate Representation Of A Multivariate Structure
Blogger Comments:
30 September 2021
[27] Misrepresenting Halliday's Explanation of Ergative Nuclearity As Doran's Argument For Martin's Orbital Structure Model
Except in the special case of the medio–receptive voice, the Medium is obligatory in all processes; and it is the only element that is, other than the process itself. … The Medium is also the only element that is never introduced into the clause by means of a preposition (again with the same exception of medio–receptives); … The Process and the Medium together form the nucleus of an English clause; and this nucleus then determines the range of options that are available to the rest of the clause.
29 September 2021
[26] Problems With Martin's Orbital Model Of Experiential Structure
Blogger Comments:
[1] To be clear, Martin (1996) misunderstands Halliday (1985, 1994) and Matthiessen (1995) by misconstruing the (multivariate) experiential structure of a clause as a (univariate) logical structure in which the Process and Medium constitute the dominant element (nucleus), with the Agent and Location as dependent (satellites):
In SFL Theory, interdependency relations obtain between units in unit complexes, not between elements in a unit.
but also with the model in Martin & Rose (2007: 95), which followed both Matthiessen (1995) and Martin (1996):
28 September 2021
[25] Martin's Orbital And Serial Structures
[1] To be clear, before Matthiessen (1995), Martin's (1992: 22) model was:
and after Matthiessen (1995), Martin's (1996) model became:
[2] As previously explained, Martin's orbital structure misconstrues the multivariate structure of the experiential metafunction as the univariate relation of hypotaxis. As Doran confirms, "there is status distinction" (nucleus vs satellite).
[3] To be clear, Martin's serial structure, which, contrā Doran, he does not class as iterative, is not just similar to univariate structure, it is Martin's model of univariate structure. However, as previously explained, in modelling logical structure as multi-nuclear, it construes each unit as having the same status, and so reduces all univariate structure to parataxis.
27 September 2021
[24] The Nuclear Model Of Clause Transitivity
Doran (2021):
Blogger Comments:
[2] Importantly, 'dependency construct' refers to a tree structure schema, not to interdependency in the sense of tactic relations between units. It will be seen in future posts that Doran confuses the two.
[3] To be clear, the notion of a clause nucleus was taken up in Halliday (1985: 147) — and every edition of IFG since — but applied to the ergative model, not the transitive model, of clause transitivity:
The Process and the Medium together form the nucleus of an English clause; and this nucleus then determines the range of options that are available to the rest of the clause. Thus the nucleus ‘tear + cloth’ represents a small semantic field which may be realised as a clause either alone or in combination with other participant or circumstantial functions.