Martin's Reason Why Discourse Semantic Units Cannot Be Defined As Categorically As Grammatical Units [New]

Martin (1992: 59):

A2 Can you get me a beer in light of my impending death from thirst? …

The point is that seen as process, any dialogue is an on-going site of textual dynamism. There is nothing to prevent an interlocutor digging in and negotiating information presented as non-negotiable: my impending death from thirst is not presented for grading in the first example above; but one can imagine contexts in which it is contested, light-heartedly or not.

Because of this dynamism it is not possible to define discourse units as categorically as grammatical ones. There is a system, but its potential for ongoing re-contextualisation means that there will always be rough edges for the analyst. Analysis in other words will inevitably involve interpretation.

Blogger Comments:

To be clear, this is a non-sequitur. The ability to negotiate wording that lies in the Residue of a clause (in this instance: my impending death from thirst) can have no bearing on what constitutes a unit at a higher level of symbolic abstraction. If such a wording is taken up for negotiation, it simply falls within the Mood or Residue of the following clause. In SFL Theory, the semantic unit that such clause realises is a move (proposition or proposal) in an exchange (Halliday 1985: 69-71).

Confusing Unmarkedness And Congruence

Martin (1992: 58-9):

Halliday's (1985) notion of congruence bears on the problems encountered here.  Just as one can argue that there is an unmarked relationship between grammar and phonology whereby tone groups are associated with a single clause, so one might suggest that a similar unmarked relationship holds between a move and a clause complex: generally speaking a move in the exchange will be realised by a clause and its dependents.

Blogger Comment:

This is not congruence.  In SFL theory, congruence refers to a non-metaphorical relation of meaning (semantics) to wording (lexicogrammar).

On the other hand, the unmarked option is 'the form we tend to use if there is no prior context leading up to it, and no positive reason for choosing anything else' (Halliday & Matthiessen 2004: 58), in contradistinction to marked, which means that the option is less frequent and 'carries a special interpretation' (Halliday & Matthiessen 2004: 207).

The realisation of a single clause as a single tone group is unmarked tonality — where tonality refers to the selection of the number and boundaries of tone groups (Halliday 1970).

Misrepresenting Structure, Metafunction And Stratum

Martin (1992: 56):

In a model of discourse structure such as that proposed by the Birmingham school, all aspects of text structure have to be incorporated into a single rank scale such as act-move-exchange-transaction-lesson.  This naturally puts a great deal of pressure on move structure to capture cohesive relations as multivariately structured act sequences.  Here on the other hand, identification, conjunction and ideation will be treated separately from negotiation as discourse structures in their own right.

Blogger Comments:

Cohesive relations cannot be "captured" by move structure as multivariately structured act sequences — whatever the pressure — for three reasons.

• First, cohesive relations are not structures, multivariate or otherwise.
• Second, cohesive relations are textual in metafunction, whereas act sequences are presented as interpersonal.
• Third, cohesive relations are lexicogrammatical in terms of level of symbolic abstraction, whereas act sequences are presented as semantic.

Misconstruing Stratification

Martin (1992: 55, 56):

Where grammar is conceived as making meaning, there is no need to add an extra layer of interpretation to the semantics to simply re-label these functions. …

The general point here is that if the grammar, or the phonology for that matter, does the work, so be it.  The model developed here does not dualise meaning and form and so does not have to re-state the contributions made by phonology and lexicogrammar to text structure at the level of semantics.

Blogger Comments:

[1] This confuses semogenesis with stratification.  The grammar making meaning is semogenesis.  In terms of stratification, the grammar realises meaning; grammar (wording) is construed as a lower level of symbolic abstraction than semantics (meaning).

[2] In proposing a level of symbolic abstraction above the grammar, it is necessary to make explicit both the system of choices at the level of semantics (meaning), and how they are congruently realised at the level of lexicogrammar (wording).  The reason it is necessary is because it is this that provides the baseline for examining incongruent realisations — grammatical metaphor — the major means of expanding meaning potential.

[3] SFL theory contrasts meaning with wording, as in the content plane strata, semantics and lexicogrammar, and form with function, as in the grammatical rank scale, where functions at a higher rank are realised by forms at the rank below.

Misconstruing Metafunctions As Modules [New]

Martin (1992: 55, 56):

For various reasons this multivariate approach to move structure will not be pursued here. These reflect in general the modular approach to text structure underlying English Text, …

Equally important, as far as modularity is concerned, is the fact that negotiation provides just one of four perspectives on text structure elaborated in the model of discourse semantics presented here.

Blogger Comments:

To be clear, here and elsewhere (pp90, 268, 269, 390, 488), Martin misconstrues each of the metafunctions and each of the strata proposed by SFL Theory as 'interacting modules'.

Importantly, metafunction and stratification are global dimensions of the theory (Halliday & Matthiessen 2014: 20, 32). Each are complementary perspectives on the same phenomenon. As different perspectives on the same phenomenon, they cannot "interact".

For example, the three metafunctional systems and structures of the clause are three perspectives on the same phenomenon (the clause).

Likewise, the two strata of the content plane, semantics and lexicogrammar, are two perspectives on the same phenomenon (the content plane), differing in terms of symbolic abstraction (identity + elaboration).

By the same token, the two planes of language, content and expression, are two perspectives on the same phenomenon (language), differing in terms of symbolic abstraction.

Confusing Paradigmatic Features With Syntagmatic Structure [New]

 Martin (1992: 53, 52):

To begin, it is useful to compare Burton's analysis of a basic action exchange with that developed in 2.3 above. The English Text analysis developed to this point is presented first, with full feature specifications at exchange and move rank.

Blogger Comments:

[1] This is misleading. Feature specifications for exchange rank were not developed in 2.3 above, and Martin does not provide a system network of features for exchange rank.

[2] To be clear, Figure 2.13 misrepresents paradigmatic features as syntagmatic structures (exchange structure), and posits the same features at both ranks ('negotiating', 'exchanging'). Cf. positing the PROCESS TYPE 'material' at both clause and group rank.

The Inconsistency In Treating Genre As A Connotative Semiotic [Revised]

Martin (1992: 51):

The rank scale at the level of discourse proposed originally by Sinclair and Coulthard included three additional ranks, two above the exchange (lesson and transaction) and one below (move).  Considerations at the ranks of lesson and transaction will be handled under the heading of genre in this book and will be taken up again in Chapter 7, along with a discussion of why genre is treated as a[n] underlying connotative semiotic rather than a higher rank at the level of discourse semantics…

Blogger Comments:

Treating genre as a connotative semiotic is inconsistent with the meaning of both genre and connotative semiotic.  Because 'genre' is a variety of language, and language is a denotative semiotic, 'genre' is a variety of a denotative semiotic. And because a connotative semiotic is a semiotic system whose expression plane is language (Hjelmslev 1961), 'genre', as a variety of language, is located on the expression plane of a connotative semiotic, not on its content plane (context).

Underestimating The Mood Grammar

Martin (1992: 50-1):

… Ventola's category of a linguistic service (1987: 115-7) functions semantically as both an action and a knowledge exchange, and can be initiated with an interrogative as in the example below:

Can you tell me your name?
— Yes, allright, John Smith

The responding move picks up on the grammar (Can you … Yes), the SPEECH FUNCTION (tell me … allright) and the exchange structure (your name … John Smith).  This can be captured by analysing the exchange structure as K2^K1 (since the exchange can only be completed by providing the appropriate information), with the K2 realised by a demand for services, which is in turn coded through the grammar as a modalised polar interrogative.  Note that linguistic services of this kind thus demonstrate that Berry's (1981a: 40) suggestion that the exchange be viewed as a lexicogrammatical rank consisting of clauses cannot be maintained since such a model could not show that Can you tell me your name? is initiating an exchange of information as a service.

Blogger Comments:

[1] The polar interrogative clause Can you tell me your name? realises a demand for goods–&–services — a command, in terms of speech function.  The speech function is demonstrated by its congruent realisation as the imperative verbal clause Tell me your name.  The service demanded is thus the giving of information.  However, here the command is realised metaphorically by a modalised polar interrogative, as a way of construing a particular tenor relation between the interactants.

The modal operator can checks the inclination of the addressee, which is, in congruent realisations, the function of an imperative mood tag.  Halliday & Matthiessen (2004: 627):

On the one hand, an ‘imperative’ clause imposes an obligation; on the other hand, the imperative tag checks the addressee’s inclination to comply… .

[2] What the response 'picks up on' is first the grammar — the polar interrogative: yes — and then the semantics — the command to tell: allright, John Smith).  This relates to grammatical metaphor as 'junctional'.

As Halliday & Matthiessen (1999: 283, 288) point out, the metaphorical form also embodies semantic features deriving from its own incongruent lexicogrammatical properties.  That is, grammatical metaphor is a means of simultaneously construing the meanings of both the congruent and incongruent grammatical realisations — in this instance: of imperative and of polar interrogative mood, respectively.

These two meanings, the speech functions command and question, are themselves in an elaborating token-value relation within the semantic stratum, with the metaphorical Token (question) realising the congruent Value (command).

Misunderstanding Stratification And Incongruence [New]

Martin (1992: 50):

The model of conversational structure developed to this point consists of three networks, two in the discourse semantics (NEGOTIATION at exchange rank and SPEECH FUNCTION at move rank) and one in lexicogrammar (MOOD at clause rank). Clauses selecting independently for MOOD are thus classified in three ways: by MOOD class, by SPEECH FUNCTION class, and with respect to their structural function in the exchange (i.e. with respect to their NEGOTIATION class). The picture is presented, by strata and rank, in Fig. 2.12.

The model thus makes provision for two types of incongruence: between NEGOTIATION and SPEECH FUNCTION and between SPEECH FUNCTION and MOOD.

Blogger Comments:

[1] To be clear, Martin does not provide a system network for NEGOTIATION, and the networks of SPEECH FUNCTION and MOOD are based on those previously devised by Halliday.

[2] This misunderstands the relation of semantics to grammar. To be clear, to claim that clauses (grammar) are classified by semantic system features ("classes") is to mistake an attributive relation (class membership) for an identifying relation (identity). That is to say, semantics (Attribute) does not classify grammar (Carrier); semantics (Value) is realised by grammar (Token).

[3] This misunderstands the notion of in/congruence in SFL Theory, where it describes realisation relations between strata. To be clear, to claim that in/congruence applies to ranks on the same stratum is to mistake composition (extension) for realisation (elaboration). See, for example, Halliday & Matthiessen (1999: 145-6).

Misrepresenting Speech Function [New]

Martin (1992: 46):

It was further noted that through ellipsis and the grading systems associated with MODALISATION, MODULATION and attitude the grammar is structured in such a way as to pair off the interacts, giving rise to the seven basic adjacency pairs suggested above. The major limitation of the description to this point is that it is not explicit about how in dialogue this pairing takes place. Both MOOD and SPEECH FUNCTION classify individual interacts, not sequences. It is to the question of sequencing interacts that this section now turns.

Blogger Comments:

[1] This is misleading, because it is not ellipsis, attitude and modality that give rise to adjacency pairs, but the fact that the giving or demanding of information or goods-&-services (SPEECH FUNCTION) invites a response from the addressee.

[2] This is misleading, because it is not true. The system of SPEECH FUNCTION specifies both initiating and responding moves. Halliday (1985: 69) listed the responses as follows:

and Halliday & Matthiessen (2004: 108) later provided the following system network:

Misrepresenting Stratification [Augmented]

Martin (1992: 46):

To this point a stratified approach to the function of interacts has been proposed, with MOOD and SPEECH FUNCTION each contributing one layer of meaning on the content plane.

 

Blogger Comment:

[1] This is misleading, because it misrepresents Halliday's prior theorising as Martin's innovation. To be clear, firstly, Martin is claiming in this publication that stratifying the content plane is his innovation, despite the fact that Halliday & Hasan (1976: 5) present just such a stratification:

And secondly, Martin is here claiming that stratifying SPEECH FUNCTION (semantics) with respect to MOOD (lexicogrammar) is his innovation, despite this distinction being made in Halliday (1985: 71, 342), if not before:

The semantic function of a clause in the exchange of information is a proposition; the semantic function of a clause in the exchange of goods-&-services is a proposal. …

The other main type of interpersonal metaphor is that associated with mood. Mood expresses the speech function; and as we saw in Chapter 4 the underlying pattern of organisation here is the exchange system — giving or demanding information or goods-&-services, which defines the four basic speech functions of statement, question, offer and command.

[2] In SFL stratification, the content plane consists of one level of meaning, semantics, and one level of wording, lexicogrammar, not two levels of meaning. The grammatical system of MOOD is a system of wording that realises meaning: the semantic system of SPEECH FUNCTION. The semantic system of SPEECH FUNCTION is a system of meaning that is realised by wording: the grammatical system of MOOD.

Misrepresenting The Potential Of Ellipsis To Negotiate Attitude [New]

Martin (1992: 45):

Significantly, attitude is not realised through the Mood function in English, unlike MODALISATION and MODULATION which can be expressed through modal verbs and adjuncts. For this reason ellipsis in English does not facilitate the negotiation of attitude — the attitude to be graded is commonly realised in the Residue itself. 

It thus follows from the grammar that how a speaker feels is less commonly negotiated through pair parts in dialogue than is inclination, obligation, probability and usuality and that where it is negotiated, MODALISATION and MODULATION are typically being negotiated as well since the Mood element is normally present (cf. Yes he is rather above).

Blogger Comments:

[1] The argument here is:

Premiss: attitude is realised in the Residue not the Mood element
Conclusion: ellipsis does not facilitate the negotiation of attitude

To be clear, this is a false conclusion invalidly argued from a false premiss. The premiss is false because attitude can be realised in the Mood element as well as the Residue, as demonstrated by:

The conclusion is false because ellipsis can "facilitate the negotiation of attitude", as demonstrated by exchanges like:

Is he scrupulously honest?

— No, he isn't scrupulously honest.

— Yes, he is scrupulously honest

The argument is invalid because the location of attitude in the clause, has no bearing on whether or not ellipsis can "facilitate the negotiation of attitude".

[2] To be clear, this does not follow from the grammar, see [1] above. Moreover, it is bare assertion, unsupported by evidence from corpora.

[3] The argument here is:

Premiss: the Mood element is normally present.
Conclusion: MODALISATION and MODULATION are typically being negotiated.

To be clear, this argument is invalidated by clauses that normally feature a mood element, but don't typically negotiate modality, such as:

The reason why Martin thinks that all negotiation is the negotiation of modality is that he (p39) misunderstands modality as including positive and negative polarity, rather than just the space between them.

The Line Between Exclamations and Statements [New]

Martin (1992: 43-5):

This raises the question of how to draw the line between Exclamations and Statements. On the one hand, exclamatives have affirmative syntax (Subject^Finite) and so are definitely negotiable; at the same time, declarative clauses may encode reaction through attitudinal adjuncts (e.g. regrettably) or through group rank systems (e.g. can't stand and silly bugger). The strategy developed here will be to treat attitudinal minor clauses and exclamatives as Exclamations:

Exclamation [minor/exclamative]     What a fooI!/What a fool he's been!

Response to Exclamation                  — Quite so.

and to include other MOOD classes only where attitude is explicitly graded in a response:

Statement [declarative]      He's been such a fool.

Acknowledge Statement     — Oh, has he?

But:

Exclamation [declarative]     He's been such a fool.

Response to Exclamation     — Absolutely.

The relevant section of the MOOD network needs to be adjusted as follows, incorporating the feature [exclamative] and adding the feature [affirmative] to generalise across the Subject^Finite st[r]ucture in both exclamative and declarative clauses.

 

Blogger Comments:

[1] To be clear, exclamations and statements are speech functions at the level of semantics, whereas exclamatives and declaratives are mood types at the level of lexicogrammar. The "line between" exclamations and statements is semantic, not lexicogrammatical.

[2] To be clear, "encoding reaction through attitudinal adjuncts" is irrelevant here because it does not distinguish either speech functions or mood types. For example:

Exclamation [realised as exclamative]     What a fool he's been!

Acknowledgement                                       — Regrettably so.

Statement [realised as declarative]      He's been such a fool.

Acknowledgement                                   — Regrettably so

[3] On the one hand, this is a non-sequitur, since it is not entailed as a conclusion by the preceding propositions. On the other hand, the realisation of exclamations (semantics) as exclamative clauses or by minor clauses (lexicogrammar) is Halliday's model; see Halliday (1985: 84-5; 1994: 84, 95).

[4] To be clear, this is a statement, not an exclamation. The exclamatory response absolutely has no bearing on the speech function of this initiating clause, since it can be used to respond to either statements or exclamations, as well as questions (yes/no), commands and offers.

[5] To be clear, this MOOD network does not address the "line between exclamations and statements", the question raised at the beginning of this section, and does not address the realisation of exclamations as either exclamative or minor clauses, not least because it does not include minor clauses as an option.

Greetings and Calls [New]

Martin (1992: 42):

Minor clauses lack Subject, Finite and Predicator functions, so the question of negotiating MODALISATION and MODULATION does not directly arise. This raises the question of the function of minor clauses as interacts. For one thing, minor clauses are used to initiate two types of adjacency pair: Greetings and Calls. Greetings (subsuming leave-takings) are found at the beginning and end of conversations, making way for or closing down negotiations:

Greeting                             G'day.

Response to Greeting       — G'day.

Calls summon the attention of potential negotiators where this attention has not been secured or has wandered:

Call                                     Bill.

Response to Call               — What?

Both Calls and Greetings can be realised through major clauses, although these are for the most part lexicalised formulas: How's it going?, Nice weather we're having!, We'll be seeing you., Got a minute?, Listen to this. etc.

Blogger Comments:

[1] As previously observed, MODALISATION and MODULATION are only potentially "negotiated" in clauses that feature MODALISATION and MODULATION, and not all exchanges involve negotiation.

[2] To be clear, this question had already been answered by Halliday in Dimensions of Discourse Analysis (1985), where minor clauses were said to realise minor speech functions: exclamations, calls, and greetings. Here Martin once again misleads the reader into thinking that this is his original thinking.

Misrepresenting Nonfinite Clauses [New]

Martin (1992: 42):

With nonfinite clauses, meaning is presented not so much as having been negotiated as non-negotiable: the clauses have been nominalised, and thereby taken out of the realm of potential interacts and into that of things (cf. They watched Manly winning/the game/their favourite player).

Blogger Comments:

This seriously misunderstands nonfinite clauses. Although his MOOD network (Fig 2.6) contrasts nonfinite clauses with dependent and embedded clauses, Martin here presents an embedded nonfinite clause, while, at the same time, ignoring the possibility of nonfinite dependent clauses.

To be clear, nonfinite clauses may be ranking (dependent) or embedded, and it is only embedded nonfinite clauses that are nominalised to served as the Head or Postmodifier of a nominal group. Examples of ranking dependent clauses include:

• The hairy coat holds a layer of air close to the skin, insulating the body against changes in the outside temperature. [elaboration]
• We used to go away at the weekend, taking all our gear with us. [extension]
• Bacteria can also aid chemical precipitation of calcite by making the water more alkaline. [enhancement: manner]
• This view was not empirically based, having arisen from an a priori philosophy. [enhancement: cause]

See Halliday & Matthiessen (2014: 468, 475, 485).

Dependent Clauses, Speech Function And Negotiability [New]

Martin (1992: 41-2):

Hypotactically dependent clauses may turn out to be better taken from a discourse perspective as an intermediate case (between embedded and independent clauses) — quantative [sic] studies might well show them to be more negotiable than embedded clauses, though less likely to be responded to than independent ones. Note in this connection that when projection is used to express modality metaphorically (Halliday 1985a:336), it is the structurally dependent clause that is in fact being negotiated:

I think he'll be there. (meaning 'maybe he'll be there'.)

— Will he? (more likely than Oh, do you?)

This need not of course block the 'sardonic' interlocutor from negotiating the projecting clause as if it deserved a congruent reading:

"I'm inclined to think—" said I.

"I should do so," Sherlock Holmes remarked impatiently.

I believe that I am one of the most long-suffering of mortals; but I'll admit that I was annoyed at the sardonic interruption. "Really, Holmes," said I severely, "you are a lit[t]le trying at times." (Doyle 1981:769)

Pending quantitative investigation of these patterns, hypotactically dependent clauses will be grouped with embedded ones as in the network above. The most negotiable clause type in this category involves dependent elaboration as these clauses can be tagged:

α Sherlock put Watson down,

= β which was mean, wasn't it?

Blogger Comments:

[1] As previously observed, Martin is here trying to determine the grammatical unit that realises speech function, but is instead concerned with arguability ("negotiability"), which is a separate issue from speech function. Embedded and some dependent clauses may be presented as unarguable, but that does not necessarily entail that they do not realise a statement, question, command or offer. For example, in this hypotactic complex:

I asked if you like Joy Division

the dominant clause realises a statement, while the dependent clause realises question, as demonstrated by its paratactic agnate:

I asked "do you like Joy Division?"

[2] To be clear, thirty years on, these quantitative investigations remain undone.

[3] To be clear, this needs to be accounted for in the model, not just mentioned and forgotten.

[4] Strictly speaking, despite the relative pronoun which, the expansion relation in this instance is extension, rather than elaboration, since

• the dependent clause adds to the meaning of the dominant clause, rather than describes it;
• the paratactic agnate of the dependent clause is and that was mean, wasn't it?;
• it does not involve tone concord, since the unmarked tones would be tone3 and tone1.

Halliday & Matthiessen (2014: 468):

There is one group of non-defining relative clauses that strictly speaking would belong with extension rather than elaboration; for example,

||| She told it to the baker’s wife, || who told it to the cook. |||

Here the who stands for ‘and she’ and the clause is semantically an additive: the agnate paratactic variant would be ... and she told it to the cook. …

Note that such instances are not characterised by tone concord.

Misrepresenting The Interpersonal Function Of Independent, Dependent And Embedded Clauses [New]

Martin (1992: 41):

These classes can be interpreted semantically as follows. Independent clauses negotiate MODALISATION and MODULATION — speakers typically work towards consensus as far as grading probability, usuality, inclination and obligation are concerned. Dependent and embedded clauses on the other hand code meanings as already negotiated — the responses in the following example are to the main clause, not the clause embedded in it (Did you love the beer? not Did you try the beer?):

I loved the beer I tried last time.

— Did you?

Blogger Comments:

[1] This is misleading, because it is not true. On the one hand, speakers do not necessarily "work towards consensus" in any respect, let alone with respect to MODALISATION and MODULATION, as demonstrated, for example, by discussions on academic email lists. On the other hand, not all independent clauses feature either type of MODALITY, as demonstrated by Martin's own example:

I loved the beer I tried last time.

— Did you?

[2] To be clear, dependent and embedded clauses present speech functions as presumed, regardless of whether or not the meanings have been "already negotiated".

[3] To be clear, the embedded clause realises a statement which can be "negotiated" as follows:

I loved the beer I tried last time.
— I didn't realise you tried it last time.

Martin's MOOD Network For Clauses Realising SPEECH FUNCTION [New]

Martin (1992: 40-1, 91n):

The MOOD network underlying this definition is presented below, along with examples of realisations for terminal features. [Ventola's 1988a criticisms of the unit proposed will be taken up in 2.3 below.] 

² The I/T superscript notation shows that dependent/embedded clauses are declarative by default; according to this network projected proposals have both nonfinite (I asked him to come) and dependent (I asked him if he could come) realisations.

Blogger Comments:

[1] To be clear, this MOOD network falsely proposes that all dependent and embedded clause are finite. This is falsified by Martin's own example I noticed you making a small one which features a nonfinite embedded clause. Halliday & Matthiessen (2014: 171):

Whether they are dependent or downranked, ‘bound’ clauses may be either ‘finite’ or ‘non-finite’.

As Halliday & Matthiessen (2014: 23-4) explain:

A major clause is either indicative or imperative in MOOD; if indicative, it has a Finite (operator) and a Subject. An indicative clause is either declarative or interrogative (still in MOOD); if declarative, the Subject comes before the Finite. An interrogative clause is either yes/no type or WH-type; if yes/no type, the Finite comes before the Subject; if WH-type, it has a Wh element.

[2] To be clear, Martin is concerned here with determining the grammatical unit that realises a proposition or proposal, despite the fact that Halliday had already identified the unit as the clause in its interpersonal guise. Martin has determined that only clauses "selecting independently for mood" realise these major speech functions, but falsely claims that this excludes dependent and embedded clauses; see previous post.

Essentially, Martin here confuses speech function with arguability. Dependent and embedded clauses can realise propositions and proposals; it's just that, in such forms, they are presented as less arguable than propositions and proposals realised by ranking independent clauses.

[3] This is misleading, because it is untrue. To be clear, this network says nothing whatsoever about projected proposals. Moreover, non-finite vs dependent is a false dichotomy: both projected clauses are dependent, the first being non-finite, the second finite. Note also that the non-finite projected clause is simply to go, because, in both projection nexuses, him serves as the Receiver of the projecting clause I asked him.

Misrepresenting Speech Function And Mood [New]

Martin (1992: 40):

The question of how many layers of meaning to recognise raises the problem of units: just what unit is it to which speech function is being assigned? Is for example the response Yes, I would, thank-you, but make it a small one. one speech act or two (or more)? Given what has been stressed to this point about the grammar making available resources for structuring dialogue the most appropriate unit would appear to be a clause selecting independently for MOOD. This rules out the embedded and hypotactically dependent clauses illustrated below:

They loved the team that won. (defining relative)

They defeated whoever they met. (nominalised wh clause)

They watched Manly winning. (act)

It pleased them that Balmain lost. (fact)

They wondered if they'd win. (hypotactic projection)

They won, which surprised them. (hypotactic expansion)

But it does admit paratactically dependent clauses, which do select independently for MOOD. Note the variation in MOOD possible after but, showing that the but is introducing a new move:

Yes, I would, thank-you, but make it a small one.

Yes, I would, thank-you, but I'd like a small one.

Yes, I would, thank-you, but could you make it a small one?

Blogger Comments:

[1] As previously explained, Martin misunderstands all strata as "layers of meaning" (even phonology!), because he confuses strata (meaning-wording-sounding) with semogenesis ("all strata make meaning").

[2] This is a non-sequitur. The number of strata ("layers of meaning") has no bearing on "the problem of units", since the latter is only concerned with one stratum, irrespective of the number of strata.

[3] To be clear, Halliday (1985: 68) "assigns" speech function to the clause as interact or exchange.

[4] To be clear, this instance involves two major speech functions:

Yes, I would [statement]
but make it a small one [command] 

[5] Although it is true that such clauses are presented as presumed rather than negotiable (Halliday & Matthiessen 2014: 172), embedded and dependent clauses can realise speech functions, and can "select independently" for MOOD. For example, in:

Did anyone see [[ if the policeman shot him ]]? 

the ranking interrogative clause realises a question, whereas the rankshifted declarative clause realises a statement; and in:

Was anyone here || when the policeman shot him? 

the dominant interrogative clause realises a question, whereas the dependent declarative clause realises a statement. Other combinations are, of course, possible, as demonstrated by the imperative clause realising a command and declarative clause realising a statement in:

Don't shoot every policemen just because one of them is a murderer.

Moreover, in an exchange, an interlocutor can respond to a dominant clause, dependent clause, or embedded clause:

Robin said that Lise saw Tom cheating at cards.

— He said no such thing! (reply to dominant clause)

— She couldn't have! (reply to dependent clause)

— But he wasn't! (reply to embedded 'act' clause)

Confusing Semogenesis And Stratification

Martin (1992: 40):

It should also be noted in passing that in the model being developed here both grammar and discourse semantics make meaning. Levinson (1983: 290) comments that the first move in the following text is both "a question and an offer, as indicated by the response".

Would you like another drink?

— Yes, I would, thank-you, but make it a small one.

This is just what would have been predicted by the model developed to this point. Grammatically the first move is a demand for information (interrogative) while semantically it offers goods (Offer); both MOOD and SPEECH FUNCTION contribute a layer of meaning to the clause.

Blogger Comments:

 [1] It is true that both grammar and semantics make meaning — making meaning is semogenesis.  However, this is quite distinct from the stratification of content plane as meaning (semantics) and wording (lexicogrammar), which construes two levels of symbolic abstraction in an intensive (elaborating) identifying relation.

[2] Grammatically, the first move is an interrogative, which is a congruent realisation of a demand for information (question) at the level of semantics.  However, this interrogative clause realises an offer.  Offers are not grammaticised in the mood system and are realised by interrogatives, declaratives or imperatives, depending on the tenor (context) being realised.

[3] This requires clarification.  The mood of a clause, at the level of wording (lexicogrammar), realises speech function, at the level of meaning (semantics).  The meaning of a clause consists of the semantic choices that the grammatical choices realise.  The clause itself is at the level of wording, not meaning.

Blurring Context And Material Setting

Martin (1992: 39-40):

The final set of factors to be taken into account when determining speech function are considerations of context.  Can you open this window? when inspecting a property is likely to be heard differently than when sitting in a stuffy room.  Field, mode, tenor, genre and ideology are all relevant as participating levels of semiosis; it is not possible simply to map speech function directly from 'words on the page'.

Blogger Comments:

[1] This blurs the distinction between context (the field of property inspection) and material setting (a stuffy room).  The context is the semiotic system that is realised by language.  In these instances, the speech function realised by the interrogative clause — question or command— varies with the situation being realised in language.

[2] While field, tenor and mode are metafunctional systems of a level of semiotic systems — the stratum of context — genre, as the name suggests, is a type of semiotic system.  A type of x is not a level of x.  A type of bird is not a level of bird.

[3] It is worth clarifying that the four most general speech functions are defined in terms of the cross-classification of giving vs demanding and information vs goods-&-services, and that three of these have congruent realisations in the mood system — the level of wording.  Metaphorical renderings can be identified by testing the mood choice against the speech function cross-classification.  Congruent and metaphorical renderings realise, and construe, different tenor values at the level of context.

Misunderstanding Modality In Responses To WH- Interrogatives [New]

Martin (1992: 39):

The one clause type to which the notion of negotiation through grading does not transparently apply is the wh interrogative. One might argue that in an adjacency pair like the following, the response simply supplies the missing experiential meaning; grading is not an issue:

What do you reckon would be good for a five-year-old kid?

— Fairy tales.

Note however that the elliptical response does take over the degree of probability coded in the wh interrogative; it's [sic] full form is She would like fairy tales, with modality probable, as opposed to possible or certain. The actual answer to this question is taken up in 2.8 below (text 2) and demonstrates that interlocutors are prepared to negotiate modality in the context of wh interrogatives where they feel uncomfortable with what an elliptical response might imply: — She'll like fairy tales, does she? The nonellipitical reply enables its speaker to adjust modality in the tag.

Blogger Comments:

[1] To be clear, this is all utter nonsense. The grading of modality in responses to WH- interrogatives is not at all problematical, as demonstrated by

What do you reckon would be good for a five-year-old kid?

— Fairy tales, possibly/probably/certainly. (elliptical)

— She'd possibly/probably/certainly like fairy tales. (non-elliptical)

[2] To be clear, because the Mood tag reprises the Finite element, the only way to "adjust modality in the tag" is to first "adjust" it in the Finite:

What do you reckon would be good for a five-year-old kid?

— She could like fairy tales, couldn't she? (low)

— She would like fairy tales, wouldn't she? (median)

— She must like fairy tales, mustn't she? (high)

Cf

What do you reckon would be good for a five-year-old kid?

— She'll possibly like fairy tales, won't she? (low)

— She'll probably like fairy tales, won't she? (median)

— She'll certainly like fairy tales, won't she? (high)

Using Modality To "Determine" Speech Function [New]

Martin (1992: 38-9, 91n):

A second set of evidence relevant to determining speech function is evidence from the co-text, either through the presence of one of the indexical markers noted above in an adjacent interact, or through the expression of gradations within probablity, usuality, inclination and obligation. Halliday (1985a:335) associates degrees of MODALISATION (probability and usuality) with propositions and degrees of MODULATION (inclination and obligation) with proposals; and within proposals, inclination is associated with Offers and Response Offers to Commands and obligation with Commands and Acknowledge Offers.

Examples of this grading in the context of the adjacency pairs considered to this point are provided below.

OFFER ^ ACKNOWLEDGE OFFER (obligation)
Shall I mark them then?
— You're required/supposed/allowed to.

COMMAND ^ RESPONSE OFFER TO COMMAND (inclination)
Get me a drink, would you?
— I'm willing/keen/determined to.

QUESTION ^ RESPONSE STATEMENT TO QUESTION (probability)
Will she win then?
— Possibly/probably/certainly she will.

STATEMENT ^ ACKNOWLEDGE STATEMENT (usuality)
She wins then.

— Sometimes/usually/always¹ she does.

The presence of realisations along these scales (with positive and negative polarity as outer poles), including their metaphorical variants (see Halliday 1985a:336), provides additional evidence for assigning speech function to grammatical classes. The metaphorical realisation of inclination I'd love to in the following pair, clearly associates it with exchanging goods and services, not information:

Why don't you get us a beer?

— I'd love to.

Similarly the scale of usuality associates the first pair below with the negotiation of propositions, while that of inclination implies the negotiation of a service.

Can you open this window?
— Sometimes/usually/always.

Can you open this window?

— I'm willing/keen/determined to.

¹ Always would of course be realised between Subject and Finite (i.e. she always does).

 
Blogger Comments:

[1] This is very misleading indeed. To be clear, here Martin is continuing the fiction that he is establishing SPEECH FUNCTION as a semantic system, despite the fact that Halliday's SPEECH FUNCTION is already a semantic system, as part of the more general fiction that he is in the process of stratifying the content plane into semantics and lexicogrammar, despite the fact that this stratification was proposed in his chief source, Halliday & Hasan (1976: 5):

[2] To be clear, here Martin is trying to establish SPEECH FUNCTION as a semantic system by arguing 'from below': how it is expressed. This is opposite to SFL methodology, which takes the view 'from above': the meaning that is expressed.

[3] To be clear, this misrepresents Halliday (1985: 335), who actually wrote:

[4] This is misleading, because it is not true. Expressions of MODALITY do not "provide evidence for assigning speech function to grammatical classes". On the contrary, Halliday (1985: 335) uses an already "determined" SPEECH FUNCTION to differentiate MODALITY into MODALISATION and MODULATION.

[5] This is misleading, because positive and negative polarity are not the outer poles of modality. Modality lies between these poles, but excludes them.

[6] To be clear, what can associate expressions of inclination with the exchange of goods-&-services, not information, is their being responses to demands for goods-&-services, not information. Such expressions are no identifier of the commodity being exchanged because the same expressions can be used in responses to demands for information, not goods-&-services:

Would you like to be a millionaire?

— I'd love to.

— I'm willing/keen/determined to.

[7] Similarly, expressions of usuality are no identifier of the commodity being exchanged because they can be used with goods-&-services as well as information:

Obey the rules, whatever your personal ethics!

— sometimes/usually/always

[8] To be clear, always can occur in any position, depending on textual considerations:

She always does (least marked: interpersonal Theme)
She does always (focus of New information)
Always she does (most marked: focused interpersonal Theme)

Misrepresenting Hasan's Work On Speech Function [New]

Martin (1992: 37-8):

Beyond this, the most significant work in the area has been undertaken by Hasan (forthcoming), working on the category of Offer. Her strategy is to extend the SPEECH FUNCTION network in delicacy to the point where it makes more categorical predictions about the realisation of discourse semantics in lexicogrammar. As a first step Hasan adds the system [conclusive/nonconclusive] to the SPEECH FUNCTION systems noted above. This allows her to distinguish between Offers which accompany the handing over of goods or performance of a service (proffers) and Offers which foreshadow such {pre-offers). …

But it is clear that the realisation relationship between discourse semantics and lexicogrammar has been considerably clarified simply by taking one step — subclassifying [giving/goods & services] exchanges as [conclusive/nonconclusive]. …

It provides a mechanism for systematically relating SPEECH FUNCTION to MOOD, and at the same time establishes a set of SPEECH FUNCTION classes that is clearly limited, and at the same time indefinitely extendable (through the scale of delicacy).

 

Blogger Comments:

[1] This is misleading. The term 'discourse semantics' here gives the false impression that Hasan is using Martin's model — and so: endorsing it — whereas, in fact, she is using Halliday's previously devised model of semantics.

This attempted deception is aided by Martin's citing of Hasan's paper as (forthcoming), despite the fact that it has been variously dated to 1985 and 1987 — 5-7 years before Martin's publication. 

Together, these two misrepresentations conspire to conceal the fact that the content plane was stratified into lexicogrammar and semantics before Martin came along to try and take credit for it.

[2] This is misleading. Hasan's work is concerned with extending the delicacy of an already established SPEECH FUNCTION system on the already established stratum of semantics, not with "systematically relating SPEECH FUNCTION to MOOD". The implication here is that the content plane is only now being carried out, here in this work, by Martin, in drawing on the work of others.

[3] Here once again Martin mistakes systemic features for classes.

Trying To Classify Speech Function 'From Below' [New]

Martin (1992: 36):

As far as determining speech function is concerned, there are a number of factors to take into account. First of all, there are several indexical markers which clearly distinguish proposals from propositions. These include please, kindly, allright, okay and thank-you. Please and kindly are found in Commands, allright and okay in Response Offers to Commands and thank-you in responses to Offers or Commands.

Could I have a midi [middy] of Coopers, please?

— Okay.

— Thanks.

Beyond this, it is difficult to find unique grammatical criteria for recognising very general categories such as Offer, Command, Statement and so on.

 

Blogger Comments:

[1] To be clear, here Martin is trying to argue semantic distinctions 'from below' — in terms of grammatical expressions. This is the direct opposite of the methodology of SFL Theory, which gives priority to the view 'from above' — the meanings that are expressed. See the previous post.

[2] This is misleading. To be clear, three of the four primary speech functions can be recognised by their congruent grammatical realisations:

• commands realised in imperative mood,
• statements realised in declarative mood, and
• questions realised in interrogative mood.

Misrepresenting The Realisation Of Speech Function In Mood [Revised]

Martin (1992: 36):

To this point a model has been outlined in which SPEECH FUNCTION (discourse semantics) has been stratified with respect to MOOD (lexicogrammar) on the content plane.  This immediately raises two questions: (i) the determination of speech function in the absence of a one to one correlation between general SPEECH FUNCTION categories and those of MOOD; and (ii) the nature of the units to which speech function is assigned.

Blogger Comments:

[1] This is very misleading indeed. On the one hand, it gives the false impression that Halliday's system of SPEECH FUNCTION was not already a semantic system, and not already "stratified with respect to MOOD (lexicogrammar) on the content plane" before Martin came along to try and take the credit.

On the other hand, it gives the false impression that Martin has provided argument that SPEECH FUNCTION is specifically a discourse semantic system, whereas, in fact, no such argument has been made. Moreover, as previously demonstrated, the "argument" that has been made confuses the non-structural textual system of ELLIPSIS-&-SUBSTITUTION with the structural interpersonal system of MOOD.

[2] To be clear, the absence (or presence) of a "one to one correlation between general SPEECH FUNCTION categories and those of MOOD" has no bearing on the determination of speech function. Each of the four categories realises the feature bundle that specifies it:

• 'statement' realises [give, information];
• question' realises [demand, information];
• 'offer' realises [give, goods-&-services];
• 'command' realises [demand, goods-&-services].

In other words, if the meaning is 'give information' etc., then the speech function is 'statement' etc.

[3] To be clear, Halliday (1981) identifies the semantic unit "to which speech function is assigned" as a move in an exchange. Halliday (2002 [1981]: 240):

Each clause is in this sense a kind of gift, one move in an exchange, symbolised by the change of perspective from me to you.

Serious Problems With Martin's Mood System [New]

Martin (1992: 34, 35):

The MOOD systems considered to this point are outlined below (Fig. 2.2)…

Blogger Comments:

To be clear, there are several serious flaws with Martin's MOOD network that make it entirely untenable. 

Firstly, it misrepresents a non-structural textual grammatical system, ellipsis-&-substitution, as part of the structural interpersonal grammatical system of MOOD. Martin later (p100-1)) compounds his error by relocating this "interpersonal" system back to the textual metafunction in his system of IDENTIFICATION on his stratum of discourse semantics. That is, Martin incongruously claims that ellipsis-&-substitution functions interpersonally at the level of grammar, but textually at the level of discourse semantics.

Secondly, the upper network bizarrely proposes that a clause can be both a declarative and a WH- interrogative, and such a clause has the option of WH- ellipsis, which, contrary to Martin's claim, has not been "considered to this point". To be clear, the selection of WH- ellipsis is limited to WH- clauses.

Thirdly, the lower network bizarrely proposes that 

• mood ellipsis ("implicit") is a subtype of residue ellipsis,
• clause ellipsis is a subtype of mood ellipsis, and
• substitution is a subtype of mood ellipsis.

Strategically Misrepresenting The Relation Between Speech Function And Mood [New]

Martin (1992: 34):

Following Halliday (1984a) the semantic inventory of interacts outlined above can now be expanded into four pairs, which will be referred to provisionally, following work in ethnomethodology (e.g. Schegloff & Sacks 1973), as "adjacency pairs":

The grammar then makes available resources for tying an initiation to a response (ellipsis and substitution) and for orienting the exchange to goods and services or information and to giving or demanding (declarative, interrogative and imperative). These resources do not however stand in any biunique relation with a particular move in dialogue, so two levels of analysis are needed to relate system and text.

Blogger Comments:

[1] To be clear, Halliday (1985: 69) provides two types of response, expected vs discretionary, with different terms for the expected responses of offers (acceptance), commands (undertaking) and questions (answer), stipulating that only the last of these is essentially a verbal response:

[2] To be clear, ignoring the fact that there may be no verbal response to offers, commands and statements, and so no adjacency pairs, the tying of an initiation to a response by ellipsis-&-substitution is a cohesive tie, which is a resource of the textual, not the interpersonal metafunction. See Halliday (1985: 295-302).

[3] This is misleading. The grammatical MOOD selections (declarative, interrogative and imperative) realise selections in semantic SPEECH FUNCTION systems of COMMODITY (goods-&-services vs information) and INITIATING ROLE (giving vs demanding). 

However, if Martin had used the word 'realise', it would have disclosed the fact that the content plane had already been stratified by Halliday — a fact he is trying to keep from the reader — since realisation is the relation between strata.

[4] This is misleading. With the exception of offers, there is a "biunique" relation between SPEECH FUNCTION selections (semantics) and MOOD selections (grammar) except in the case of interpersonal metaphor. It is grammatical metaphor that motivates the stratification of the content plane into semantics and lexicogrammar (Halliday & Matthiessen 1999: 237), which is the main reason why Halliday stratified the content plane before Martin came along to try and take the credit.

Using A Textual Grammatical System To Argue For An Interpersonal Semantic System [New]

Martin (1992: 33-4):

Furthermore, as can be seen from the examples above, the relevance of the co-text is also grammaticalised in English. The relevant systems have to do with clause ellipsis and substitution and are described in detail in Halliday and Hasan (1976). There are two major patterns to be considered. The first can be established with respect to the more general MOOD functions Mood (including Subject, Finite and Mood Adjuncts) and Residue (including Predicator, Complement and other Adjuncts) and illustrated as follows:

Get me the new one, please. — Allright, I'll get it for you.

Get me the new one, please. — Allright, I will Residue.

Get me the new one, please. — Allright Mood Residue.

The first response (Allright, I'll get it for you.) is full, though potentially elliptical. The second (Allright, I will.) ellipses the Residue. The third (Allright.) ellipses both Residue and Mood functions. Alternatively, the Mood and Residue functions together may be substituted with so or not.

Will he make it? — Maybe so.

Will she win it? — Perhaps not.

The second pattern is found in the environment of wh interrogatives of both the first order and second order (echo) types; again, both elliptical and potentially elliptical structures are found:

Which is the new one? — This one's the new one.

Which is the new one? — This one.

This is the new one. — Which one's the new one?

This is the new one. — Which one?

As before, there is no one to one relation between grammatical class and discourse function. Responses may be either elliptical or not, as illustrated. In addition, elliptical clauses may initiate dialogue, with ellipsed information typically recoverable from the non-verbal context:

//2 Coming? // — Yes, I am.

//1 Leaving now. // — Oh, are you?

Blogger Comments:

[1] To be clear, this is nonsensical. The "relevance of the co-text" cannot be "grammaticalised". Grammaticalisation is essentially a move from the lexical end of lexicogrammar to the grammatical end, as when 'content words' become 'function' words.

[2] To be clear, Martin is here arguing that there is no 'one to one relation' between interpersonal semantics and interpersonal lexicogrammar. This would require demonstrating that there is no 'one to one relation' between the selection of SPEECH FUNCTION features and their realisation in the selection of MOOD features. However, instead, Martin presents only textual systems of the grammar, ellipsis-&-substitution, and does so without regard to the semantic systems they realise. 

Moreover, Martin cannot relate ellipsis-&-substitution to his interpersonal semantics because he subsumes ellipsis-&-substitution within his textual system of IDENTIFICATION (p100-1),  his rebranding of Halliday & Hasan's cohesive reference.

[3] Again, Martin misunderstands grammatical functions — here: ellipsis-&-substitution — as grammatical classes (forms).