Mick O'Donnell On Following SFL Analytical Principles

On 27 June 2013, on the sysfling list, Mick O'Donnell wrote: 

We are doing grammar, not notional semantics (well, I think so at least), so we need to use grammatical tests to determine process types. Not a reference to "same meaning". 

and then again on 28 June 2013: 

We need to follow the grammatical PRINCIPLES that Halliday established (how to argue Linguistically) rather then the grammatical FACTS that he put forward on the basis of applying these principles.

Blogger Comment:

Halliday's 'principle' for grammatical analysis is to take a trinocular perspective, which means also looking at the grammar 'from above' (what meaning is being realised) and 'from below' (how the wording is realised) — not just 'from roundabout' (the level of grammar).  As Halliday & Matthiessen (2004: 31) make clear:

We cannot expect to understand the grammar just by looking at it from its own level; we also look into it ‘from above’ and ‘from below’, taking a trinocular perspective. But since the view from these different angles is often conflicting, the description will inevitably be a form of compromise.

O'Donnell's application of the term 'facts' to theory, rather than data, betrays an epistemological confusion.  'Facts' are the construals of experience from which theory is designed.

Tom Bartlett On Ellipsis

On 19/6/13 Claudia Stoian asked on the sys-func and sysfling lists:

In clause complexes as the following, I could consider theme of the second clause either the dropped subject or the predicator. Which variant would you choose?

Canterbury Cathedral was founded in AD597 and [ø: Canterbury Cathedral] is the Mother Church of the Anglican Communion

David Banks then replied on sys-func:

I am averse to adding in things which are not in the original clause. So, I consider "Canterbury Cathedral" to be the single theme of the two coordinate clauses. This goes against the usual principle of one clause - one theme, but I feel it works better. However, I may be alone in holding this point of view.

Tom Bartlett then added on sys-func:

I agree with David. There is a very good case for seeing such cases as a single clause with coordinated Residues/propositions/processes (Finite + Residue often, so no cover-all term in SFL). If you put a tag at the end of them then both propositions usually need to be true for a "yes" response.

If we treat "I saw Peter and David" as a single proposition and "Peter and David saw me" as one, why not "I went to the shops and bought a cake"? In other words, not ellipsis at all.

Blogger Comments:

[1] Bartlett says he agrees with Banks, even though Bartlett sees the example as a single clause and Banks sees it as two clauses.

[2] The case for seeing the clause nexus as a single clause is not "very good", as demonstrated, for example, by a transitivity analysis: it includes two distinct processes, the first of which is material, the second relational.  See analysis here.  See related theoretical quote here.

[3] Coordination is a logical relation (paratactic extension), that holds between forms: clauses, phrases/groups, words. Here Bartlett misapplies the relation to functions: Residues/propositions/processes. This (unwittingly) complicates the theory without adding explanatory power.

[4] Any proposition regarded as true by someone will yield a 'yes' response from that person.  This is not an argument that supports the case of treating the example as a single clause.

[5] The clause nexus 'I went to the shops and bought a cake' enacts two propositions: 'I went to the shops (didn't I?)' and 'I bought a cake (didn't I?)', either of which can be separately affirmed or denied, whereas each of the clause simplexes 'I saw Peter and David (didn't I?)' and 'Peter and David saw me (didn't they?)' enacts just a single proposition.

Tom Bartlett On Attribute, Circumstance, Adjunct & Complement

On May 31, on the sysfling and sys-func lists, Claudia Stoian wrote:

I have a question related to the interpersonal analysis of clauses with relational processes. I know that in the case of intensive attributive relational, the Attribute is considered Complement. But what happens in the other cases, possessive and circumstantial? Is the Attribute a Complement or an Adjunct? I'd consider it an Adjunct since it can't become a Subject. What about identifying relational processes? The Identifier/Identified in the 2nd position is a Complement, since it can become a Subject, right? Here are some examples I found in my corpus, for the possesssive and circumstantial cases:

1. their extensive and well-preserved remains stand in spectacular natural surroundings
2. The only time the ceremony has been interrupted in the last 700 years was during the Second World War.
3. It has a perpendicular nave, 12th-century Gothic choir, stunning stained glass windows and a Romanesque crypt.
4. Your tour includes tickets to Edinburgh Castle.

To which Tom Bartlett replied:

In your examples 3 and 4 there is no problem with Attribute and Value as Complements, as you've highlighted. I think 2 is also an identifying clause, however.

I think 1 is best analysed as Circumstance conflated with Complement (though this is not in IFG). There is a tendency for SFLers to conflate the two categories of Circ and Adjunct all the time in analysis, which would make the distinction rather pointless. In this way it is often said that Circumstances can appear just as easily at the beginning or end of the clause. I think this is wrong: it is a property of Adjuncts (as non-nuclear, as "join-ons") that they have freer positioning around the nucleus. As most of the time Adjuncts are conflated with Circs, this might appear to be a property of Circs; but that would be wrong, as example 1 shows. Here thematising the pphr would be as marked as fronting with other Complements:

Similarly, the core property of Complements is that they "complete" the nucleus; and as the Circ is required in example 1 it fits this category.

So, for me your example 1 has a Circ as Complement. It is therefore neither an Attribute in experiential terms nor an Adjunct in interpersonal terms

and later:

P.S. I should make clear that the concluding sentences of my previous e-mail were my analysis, not that of IFG:

So, for me your example 1 has a Circ as Complement. It is therefore neither an Attribute in experiential terms nor an Adjunct in interpersonal terms

For IFG the highlighted section of example 1 is At:Circ

1. their extensive and well-preserved remains stand in spectacular natural surroundings

The point I was making is that this analysis in some way helps perpetuate the tendency for analysts to conflate the concepts of Adjunct and Circumstance as it obscures the Circ function of the phrase in question. It also seems to conflate two experiential elements that should not really be conflated - a Circumstance and a participant. Does Halliday mean Circ here as a more delicate category of At (so not the same as a full Circ?) or is he suggesting conflation? I see each one as having its own problems.

Blogger Comments:

[1] On the SFL model, the prepositional phrase in spectacular natural surroundings functions experientially as a (Location) circumstance, and

• experientially as Attribute, and 
• interpersonally as Adjunct (not Complement).  

The reason why it can be said to function as Attribute is that it (in spectacular natural surroundings) is the element of an attributive clause that realises the class to which the Carrier (their extensive and well-preserved remains) is assigned as a member.  One reason why Bartlett's analysis is not preferable to Halliday's is that his notion of an attributive relational clause with a Carrier but no Attribute is nonsensical.  The function of an attributive relational clause to relate a Carrier to an Attribute.

The reason why this prepositional phrase can be said to function as an Adjunct is that, by definition, 'an Adjunct is an element that has not got the potential of being Subject' (Halliday & Matthiessen 2004: 123).

See transitivity and mood analysis here.

[2] Circumstance and Adjunct are distinct categories viewed from the angle of distinct metafunctions: the former is experiential, the latter is interpersonal.  Adjuncts are not restricted to those that conflate with circumstances; Adjuncts include not only circumstantial Adjuncts, but also modal Adjuncts and conjunctive Adjuncts.  It is not the concepts that are conflated, but the functions, as when one element of clause structure functions experientially as a circumstance and interpersonally as an Adjunct.

[3] Complements do not "complete" the nucleus, nor are Adjuncts "non-nuclear".  Whereas Complements and Adjuncts are interpersonal angles on the clause, nuclearity is an experiential angle, specifically that of the ergative model (Process, Medium, Range etc).

[4]  Circumstances do conflate with participants if they are presented as participating in a process.  A common environment for the conflation of circumstance with participant is the circumstantial identifying clause (see Halliday & Matthiessen 2004: 242), as demonstrated by Claudia's 2nd example above:

• The only time the ceremony has been interrupted in the last 700 years was during the Second World War

which, incidentally, is a clause that does genuinely feature a Complement that is realised by a prepositional phrase, since it can be made Subject.  See transitivity and mood analysis here.

[5] There is no Value participant in the 4th clause above, because the relational Process is attributive, not identifying.