PROBLEMS WITH THE COALFACE GRAMMAR

 

Emeritus Professor Rodney Huddleston,

Honorary Research Consultant, The University of Queensland

 

1. INTRODUCTION

In July 2007 my attention was drawn to an article by Lenore Ferguson about teaching grammar. It appeared in journal for English teachers called Words'Worth, and was the first in a planned series under the heading `Grammar at the coalface'. This initial article was entitled `The structural basics'. Since I regard the teaching of grammar in schools as very important, I turned to it with interest.

What I found was, to my considerable regret, a large amount of inconsistency, error and confusion. I accordingly suggested that I should write a reply. This article is the result.

By the time I began writing, the June issue of Words’Worth had appeared, with a second article, ‘Functional elements in a clause’. It was decided that this reply should cover them both. It does not, however, cover the September and December articles. I was able to send comments and suggestions on a preliminary draft of ‘Discourse elements in texts’, while few comparable problems arise with ‘Aesthetic and artistic elements in texts’.

It is a bit uncomfortable for me as a non-member of the English Teachers' Association of Queensland (ETAQ) to be criticising what was a major project for the 2007 volume of the Association's journal. I feel embarrassed to be pouring cold water on a plan that in principle I want to support. But as a grammarian I feel I have to report that the grammatical content of the first two articles is not a reliable resource for teachers of English grammar. I will lay out below my reasons for saying this. I can but hope that readers will not feel inclined to shoot the messenger.[1]

In email messages sent out to a number of academics inviting their comments on the ‘Grammar at the coalface’ articles, Lenore Ferguson spoke of ETAQ members seeking guidance in two areas: ‘on links between structural and functional approaches to grammar; and on how to marry grammar with other areas of the English curriculum’ [email 13.09.07]. She also stated that the grammar sketch offered ‘is not ‘pure’ but includes selected aspects of traditional grammar, of discourse theories, of aesthetics and of verbal art - all within a functional perspective’ [email 28.11.07]. And in an email to me [30.11.07] she wrote: ‘Our purpose was to provide a simplified overview of structural and functional grammar from a functional perspective and a functional linguistics framework’.

The functional framework adopted is, more specifically, that of Hallidayan Systemic-Functional grammar. There are aspects of Functional Grammar's analysis of English that are in my view unsatisfactory, but I'll try to deal with these separately from weaknesses specific to the Coalface articles. What follows will therefore be organised into three parts. The first deals with the inconsistency, error and confusion in the Coalface grammar itself. The second covers two related problems with Functional Grammar. And the third treats the issue of simplifying Functional Grammar.

There are a good number of cases where the Coalface grammar departs from Functional Grammar in ways that cannot be justified in terms of simplification, or in any other way, and some of these will be dealt with in Part 1.

There is a fair amount of traditional grammar in the two articles I'm considering, but to a large extent it is inconsistent with the main functional orientation of the work. Functional Grammar is very different from traditional grammar. I will deal with the contrasts as they arise in the topics covered in Part 1. As for ‘structural grammar’ it is difficult to find anything that derives specifically from structuralism in the two articles, so I will not address that topic any further.

For brevity, let me introduce some bibliographical abbreviations. I'll refer to the Grammar at the Coalface articles collectively as ‘CF’. To make a reference to the left column on p. 38 of Part 1 (‘The structural basics’) I'll write ‘CF1: 38L’; for the right column on p. 49 of Part 2 (‘Functional elements in a clause' I'll write ‘CF2: 49R’; and so on. I'll call Functional Grammar ‘FG’, and I'll refer to the main source on FG, the book Introduction to Functional Grammar, as ‘IFG’.[2] Finally I will refer to my Words'Worth paper, ‘Aspects of grammar’, as ‘AG’.

IFG has a useful convention of using a capital initial for function terms (e.g. Subject) and a lower-case initial for class terms (e.g. parts of speech such as noun), and I'll follow that practice here (as also in AG).

For reasons outlined in Section 9, I will use the term ‘noun phrase’ in preference to FG's ‘nominal group’ and CF's ‘noun group’, except where the context makes one or other of these latter terms more appropriate.

 

PART 1. INCONSISTENCY, ERROR AND CONFUSION IN CF ITSELF

 

2. THE PARTS OF SPEECH

Grammars of English typically begin by assigning the words of the language to word classes, traditionally known as parts of speech. Some of the most striking and obvious errors in CF concern the classes to which certain words and other expressions are assigned.

 

2.1 Adverbs and adjectives.

Perhaps the worst cases have to do with the classification of won't and capable of as adverbs (CF1: 37R), and of a pair and set of as adjectives (CF1:37L; 2:56, Table 2). The two so-called adverbs appear in a set of five examples partially reproduced in [1]:

 

[1]       i         The small boy probably ate his lunch.

ii         The small boy won't eat his lunch.

iii                                                                         The small boy is capable of eating his lunch.

 

The traditional analyses of these words are very well known, and they are basically correct.

 

o       Probably is an adverb;

o       won't is a verb (one of the special verbs that many grammars call auxiliary verbs);

o       capable of is not a grammatical unit at all;

o       capable is an adjective (capably is the corresponding adverb); and

o       of is a preposition (and of eating his lunch is a preposition phrase functioning as a Complement of capable - in the sense of Complement explained in AG, Section 3).

 

What makes CF's errors here so serious is that the adverb analysis is quite unprincipled. Normally when we classify words together into a single part of speech we are saying that they share certain grammatical features. But the underlined expressions in [1] have quite different grammatical properties.

Let's first consider probably and won't, comparing [1i-ii] with [2]:

 

[2]       i         Probably the small boy ate his lunch.

ii         Won't the small boy eat his lunch?

 

Moving the adverb probably to the front of the clause merely gives it a little more prominence, but moving the auxiliary verb won't to the front has a completely different effect: it changes a declarative clause into an interrogative. This is what happens with other auxiliary verbs too: compare declarative He had eaten his lunch and interrogative Had he eaten his lunch?


Now compare probably and capable of. Although we can replace capable of by probably in [1iii] to give The small boy is probably eating his lunch this replacement is not generally possible, as evident from examples like [3]:

 

[3]       i         The small boy is capable of great achievements.

ii         The small boy seems capable of eating his lunch.

 

If probably and capable of belonged to the same grammatical class we would expect to be able to substitute probably in such examples, but of course we can't: *The small boy is probably great achievements and *The small boy seems probably eating his lunch are quite impossible (the asterisk indicates that what follows is ungrammatical)

Arguments like these - and numerous others could be added - demonstrate that won't and capable of behave grammatically quite differently from the adverb probably. I can only assume that the the three expressions were grouped together on the basis of meaning. This, however, doesn't provide a valid basis for grammatical classification. If you want to know what part of speech a word belongs to you have to examine its grammatical behaviour, and you need to be prepared to look beyond the sentence you're actually analysing in order to find relevant evidence.

The same applies to the so-called adjectives a pair and set of. CF groups these together with ten and second, but though the meanings may all have to do with quantification, the grammatical properties are very different.

 

o       A pair is a noun phrase, with pair the Head noun: it can be manipulated like other noun phrases, so that we can, for example, replace a by other members of the determiner class such as the and this, and the Head can be modified by adjectives, as in a new pair, and so on.

o       Set of is again not a grammatical unit. It is cited in the example a set of bowls, a noun phrase, with set as Head noun and of bowls a preposition phrase functioning as a Complement. The status of set as a noun is evident from the fact that it can be made plural (two sets of bowls) or modified by an adjective (a new set of bowls); note also that the Complement is optional, for we can have a set on its own.

 

The fact that CF analyses won't and capable of as adverbs, a pair and set of as adjectives is enough to show that it cannot be used as a model for the student. You can't do a grammatical analysis of a sentence simply on the basis of intuitions about meaning: anyone trying to go about it this way will be led into countless similar errors.

Another serious part-of-speech error that should be mentioned in this subsection is the misclassification of the underlined adverbs in more swollen and most swollen as adjectives (1: 36R). In general, it is easy to distinguish adjectives from adverbs because there are thousands of adjective-adverb pairs like extreme and extremely, where the adverb is derived from the adjective by adding the suffix -ly. This provides a simple test: see which member of such a pair would be required in the context in question. Applying this to the swollen examples we find that more and most can be replaced by the adverb extremely, not the adjective extreme.

In addition, CF follows the common practice in school grammars of misclassifying nouns as adjectives when they are functioning as Modifier of another noun, as in porcelain figurine or school yard (1: 37L), I discussed this issue in AG, Section 2.3, and here need only add that this is one of the places where CF departs from the FG model: FG recognises that the Classifier function (a subtype of Modifier) can be realised by adjectives or nouns.

 

2.2 Determiners


CF's account of the determiner class is inconsistent and confused. The demonstratives this, that, etc. (as used with a following noun, as in this book, that car) are sometimes treated as adjectives, sometimes as determiners (see, for example, 1: 39L) - and the statement that ‘determiners appear as determining adjectives in noun groups’ (also 1: 39L) is incompatible with CF's part-of-speech system. Adjective and determiner are presented as distinct parts of speech, along with noun, verb, etc. This means that no (unambiguous) instance of a word can be assigned to two different parts of speech: it doesn't make sense to say, for example, that in this book the word this is both a determiner and an adjective.

In the second article (CF2: 56) the determiner class is shown as including not only articles and demonstratives, as in the first, but also possessive pronouns, as in your folder. That, however, is again not a permissible analysis within the framework of CF: determiner and pronoun are distinct parts of speech, so this your cannot belong simultaneously to both classes. A further error here is that Sam's is included among the possessive pronouns: it is of course the possessive form of a proper noun.

The problem CF gets into with these possessive forms results from its failure to distinguish between functions and classes. We need to distinguish between the Determiner function and the determiner word-class. The your of your folder and the Sam's of Sam's folder are like the in the folder in function, but not in class: they are Determiners in function but noun phrases in class - possessive noun phrases consisting respectively of a pronoun and proper noun as Head. This analysis is presented more fully in AG, Section 2.2.

 

2.3 Pronouns

Besides this issue of the treatment of possessives there are three other problems concerning CF's treatment of pronouns.

 

(a) CF follows traditional grammar in taking the pronoun to be a distinct part of speech. FG and most other modern approaches, by contrast, take the pronoun to be a subclass of noun, along with common noun and proper noun. This difference correlates with the fact that traditional grammar has no concept corresponding to the modern noun phrase (nominal group or noun group). In a framework which, like CF, does have this concept, we need pronouns to be a subclass of nouns because a noun phrase is a phrase that (in the absence of ellipsis and the like) has a noun as Head. Pronouns are like common and proper nouns in that they function as Head in noun phrase structure.

 

(b) What and who in They saw [what lay before them] and the explorer [who saw the carnage] are incorrectly given as examples of conjunctions (1: 40L). They are in fact pronouns, for they are functioning as Head of the Subject noun phrase in the bracketed subordinate clauses. The analysis of who as a conjunction is, moreover, inconsistent with the earlier correct classification of it as a relative pronoun (1: 38R).

 

(c) There is inconsistency between the analysis of something as a pronoun (1: 39L), and everything as a noun (1: 43R).

 

2.4 Verbs

‘Verb’ is the name of a class of words, but CF also applies it to such word sequences as have a peep (1: 37R), is faced with, wants to help (1: 43R).

 

o       Have a peep is not a verb: it is a verb + Object construction. It differs from have a car, say, in that have here has little content, the meaning of the whole being much the same as that of peep on its own, used as a verb, but that doesn't make have a peep itself a verb.

o       In Kit is faced with a terrifying situation the with is a preposition which forms a unit with the noun group following: with a terrifying situation is a preposition phrase. This is obvious from the fact that we can say, for example, the situation with which she is faced: the underlined unit is a relative clause introduced by a preposition phrase, just like that in the knife with which she cut it.


o       In she wants to help James we have the verb wants followed by a non-finite clause. This is an extremely frequent construction: compare seemed to evaporate, intend to help, start reading, got killed, and so on. It may be that Lenore felt that FG's account of it was too complicated for CF (see Sections 4, 10 below), but in that case some other description must be found, for you will be constantly coming across examples like these if you attempt any textual analysis. You can't avoid the issue by treating such sequences as single verbs - as having no internal syntactic structure.

 

2.5 Numerals

FG has numeral as a distinct part of speech. CF, however, is inconsistent in its treatment of these words: numeral is not included in the initial list of the parts of speech (1: 35R), but in the next paragraph it is listed along with noun, adjective, pronoun, and determiner among a group of parts of speech called ‘nominals’. On p. 37L, however, there is a switch over to the term ‘numerative’ (which in FG is the name of the function in nominal group structure filled by numerals), and this is treated as a subclass of adjective.

 

3. FUNCTIONS IN CLAUSE STRUCTURE

In this section I draw attention to problems concerning CF's treatment of Theme, Subject, Complement and Participant.

 

3.1 Theme

In the first paragraph on information packaging (CF2: 52R) we read ‘What comes first [in a clause] is ‘known’ and leads to ‘new’ information. What comes first in a Textual analysis is the grammatical Theme.’ The Theme is identified as in FG, but the statement that the Theme is ‘known’ (= ‘given’ in FG terminology) is false. Very often the Theme is given, but it is not at all unusual for it to be new, as in Something strange is going on. Or compare B's answers in the following very simple question-answer sequences:

 

[4]       i         A: What did Sue do?                  B: She called the police                     [Theme given]

ii         A: Who called the police?           B: Sue did.                                          [Theme new]

 

In [i] the Theme she represents given information: it refers to the person mentioned in A's question; the rest of the clause, the Rheme, gives new information. But in [ii] we have the reverse situation: the Theme Sue is new, for this gives the information that A was asking for; the Rheme is given, with did interpreted from A's question as ‘called the police’. CF is thus identifying two different contrasts, that between Theme and Rheme and that between given and new. This error is particularly serious in view of the fact that IFG emphasises the importance of distinguishing them: Section 6 of Ch. 3 is devoted to precisely this issue.

 

3.2 Subject

3.2.1 Definitions

Early in Part 1, CF issues a warning that traditional school textbooks need to be used with caution because they frequently define grammatical terms on the basis of meaning rather than grammatical properties (1: 35R-36L). Yet CF itself uses this kind of definition for the Subject. There are, in fact, two such definitions of the Subject commonly found in school textbooks, and both of them appear in CF. One is that the Subject is the ‘doer of the action’ (1: 41R), while the other is that the Subject ‘denotes that about which an assertion is made’ (1: 42L). It is easy to show that neither will work.

 

(a) Subject as ‘doer’. There are two reasons why a definition of the Subject as the ‘doer of the action’ won't work satisfactorily.

 

o       In the first place, many clauses don't describe actions. Sue is like her mother and I don't know the answer, for example, describe states, not actions, so the Subjects (underlined) aren't ‘doers of the action’.

o       Secondly, in an active-passive pair describing an action, such as Sue shot the intruder and The intruder was shot by Sue, the doer of the action is the same in both, namely Sue, but only in the active version is Sue Subject. CF itself implicitly refutes the definition when it notes (2: 53R) that in passives like The race was won by the boy the doer is ‘relegated’ to the by-phrase.

 


There is certainly a significant correlation between Subject and actor (doer): In the grammatically most elementary constructions (which exclude passives) the Subject of a clause expressing an action will be associated with the actor role. But that is very different from a test that can be used in textual analysis to identify the Subject.

 

(b) Subject as ‘what an assertion is about’. An initial problem here is that what an assertion is about - the ‘topic’ - is much less clear than what its Subject is. Nevertheless, there are lots of examples where it is quite evident that the Subject does not identify the topic. Something has gone wrong with the amplifier, for example, would surely be construed as saying something about the amplifier, not ‘something’; Nobody understood Ed's lecture similarly tells us something about Ed's lecture, not ‘nobody’; and so on.

I should add that this definition is similar to what FG gives as the meaning of the Theme (or one major sub-type of Theme). Very often Theme and Subject coincide, but the Theme can also precede the Subject, as in The remaining problems we can deal with next week, where the Theme is the remaining problems and the Subject is we. And FG would say that it is here the Theme, not the Subject, that identifies what the clause is about.

 

3.2.2 Distinctive grammatical properties of the Subject.

To pick out the Subject in textual analysis we need to look at the grammar rather than the meaning, and this is generally easy because it is distinguished from other elements in the clause by a whole cluster of properties:

 

o       It usually has the form of a noun phrase.

o       Its default position (i.e. the position it occupies unless there are special reasons for it to be located elsewhere) is before the verb.

o       Pronouns with a nominative-accusative case contrast (I vs me, etc.) appear in nominative case when Subject of a main clause: I saw him.

o       Person-number properties of the verb are normally determined by agreement with the Subject: The dog is barking vs The dogs are barking.

o       Certain types of interrogative clause are marked by Subject-auxiliary inversion. The Subject thus occurs after the first auxiliary verb rather than in its default position: compare declarative Sam has arrived and interrogative Has Sam arrived?

o       An interrogative tag (CF's ‘mood tag’) contains a pronoun which relates back to the Subject of the clause to which it is attached: Sue is in the garden, isn't she? (This property is given, along with the ‘doer’ definition, in the glossary entry for Subject in CF2: 57.)

 

3.2.3 Subject in non-finite clauses

There's one other point to be made about CF's account of the Subject - a factual one. Contrary to CF (1: 41R), it is possible to have a Subject in non-finite clauses. This is shown in the following examples, where bracketing marks the non-finite clause and underlining its Subject:

 

[5]       i         [For you to give up now] would be a great pity.

ii         We objected to [her children being given special privileges].

iii         [All things considered,] it was a reasonable success.

 

3.3 Complement

The term Complement has a narrower sense in traditional grammar than in FG. Compare:

 

[6]                                                                                Trad grammar          FG

  i         Sue is a nanny.                                       Complement            Complement

ii         The cat caught a mouse                          Object                     Complement

 


In traditional grammar the underlined noun phrase is Complement only in [i]: in [ii] it is Object. In FG, on the other hand, it is a Complement in both; FG doesn't use the term ‘Object’, but would distinguish between the examples on the ideational dimension with a nanny an Attribute and a mouse a Goal.

CF is inconsistent in that it initially introduces Complement in the traditional sense, where it contrasts it with Object (1:36L), and then defines it in the Glossary (1: 41L-R) in the broader FG sense - even though the Glossary also contains an entry for Object.

The Glossary definition, moreover, does not give a faithful account of what FG says. CF says that a Complement ‘has the potential to be the subject of the verb’, which means that it is not Subject in the clause containing it but that there is a related clause in which the noun group concerned is Subject: compare active-passive pairs like The Smiths hired a nanny (Complement) and A nanny was hired by the Smiths (Subject). This matches IFG: 122, but CF fails to mention the crucial point that FG recognises a systematic exception to the general rule that a Complement has the potential to be Subject. This is illustrated in such examples as Sue is a nanny or Sue had been ill, where the underlined items are Complement but do not occur as Subject in any related clause. For these have no passive counterpart, and nor can we say *A nanny is Sue or *Ill had been Sue. FG calls clauses like Sue is a nanny and Sue had been ill attributive: they attribute some property, quality or characteristic to the Subject; the Attribute expression is a Complement, although there is no corresponding clause in which it is Subject.

The CF glossary entry for Complement does contain an example with the verb be, but it is one where be is used in an identifying rather than attributive sense. The example is Ms Jago is the teacher, interpreted as identifying who Ms Jago is or who the teacher is. Here it is possible to switch the functions of the two elements, giving The teacher is Ms Jago, with the teacher now Subject. The teacher in Ms Jago is the teacher is presented in CF, therefore, as fitting the definition of a Complement as a noun group that isn't Subject but could be. But clearly it won't do to ignore the (much more frequent) attributive use of be: the entry is seriously deficient for there is nothing to indicate that the final element in Sue is a nanny, Sue had been ill and the like is a Complement.

 

3.4 Participant

Subject and Complement are functions on the interpersonal dimension of clause structure; on the ideational dimension, CF distinguishes Participants, Processes and Circumstances (2: 50ff). These are said to be realised respectively by noun groups, verb groups and adverbial groups. Later, noun groups are actually identified with Participants: Tables 2 and 3 have titles that include the expression ‘structural patterns of noun groups (Participants) in English’. This is a serious mistake, for noun groups and Participants are different kinds of concepts and do not stand in a one-to-one relation.

Consider the following examples:

 

[7]       i   a.    He wasted the whole morning.                      b.   He watched TV the whole morning.

ii         The house opposite the post-office burned down.

iii         He gave the parcel to his brother at home.

 

o       In the first place, noun groups do not always function as Participant. In [i] the underlined noun group is functioning in clause structure, but while it is a Participant in [a], it is a Circumstance in [b]. And in [ii] the post-office is embedded in a larger noun group and is itself neither Participant nor Circumstance.

o       Conversely, Participants are not always realised by noun groups, as shown by [iii]. This is one of CF's examples (2: 50L), with to his brother labelled Participant, but without ackowledging that this is a preposition phrase, not a noun group. And it won't do to slip this in without comment: if preposition phrases can be either Participants or Circumstances the reader needs some guidance on how to tell the difference.

 

4 RANK

CF begins its account of the grammar by presenting a hierarchical arrangement of units referred to

as the rank scale (1: 35L):

 


[8]          clause complexes

     clauses

     word groups and phrases

     words

     morphemes

 

The rank scale is a highly controversial feature of FG, and there are several aspects of CF's treatment that need comment.

 

4.1 Rankshift

CF says that ‘Elements of each rank combine to form the next highest rank’, but Part 1 fails to mention an essential qualification, namely that a unit of one rank may be rankshifted, or embedded, so that it functions in the structure of a unit of the same or lower rank, not the next highest rank. Some of the major cases of rankshift in English are illustrated in the following examples:

 

[9]       i         Why she married him is a mystery.                                                [clause as Subject]

ii         I regretted that I had told them.                                            [clause as Complement]

iii         I've mislaid [the book you lent me].                       [clause as Qualifier in noun group]

iv         I need [another copy of the report].              [prep phrase as Qualifier in noun group]

v         We live [near the station].                      [noun group as Complement in prep phrase]

vi         He stole [an old woman's handbag].         [noun group as Determiner in noun group]

 

The underlined expressions in [i-ii] are clauses rankshifted to function in the structure of another clause, as Subject and Complement respectively.[3] In [iii] we have a clause rankshifted to function in the structure of a unit of a lower rank, a group (a noun group). And in [iv-vi] we have a unit of group/phrase rank rankshifted to function in the structure of a larger unit of the same rank (group and phrase being different terms for the same rank: see Section 4.4). Constructions [iii-iv] are discussed in Part 2 (2: 53L), but without explanation of the term ‘rankshift’ or acknowledgement that it represents a departure from the account of rank given in Part 1.[4]

The failure to mention rankshift in Part 1 leads to a seriously deficient account of the structure of phrases, or word groups (1: 40L). This section says that ‘Word groups comprise a head word and, optionally, one or more dependent words that elaborate the meaning expressed in the head word’. But this is clearly false, failing to account for numerous constructions like those in [9iii-iv, vi] where dependents have the form of clauses, preposition phrases, and noun groups rather than words.

There is indeed an instance of rankshift in the example CF uses in its initial illustration of the rank scale, but it is not recognised as such. In Untried men competed against him (1: 35L) the phrase against him is analysed directly as a sequence of two words, but in fact it consists of a word + a group, since that is the analysis given later for preposition phrases (1: 40L). Thus him is a noun group, like the station in [9iv]: it happens to consist of only one word, but there is nothing exceptional about that. And because it is a group functioning in the structure of a unit of the same rank it is rankshifted. Since there has been no mention of rankshift, the example used should of course have been one that didn't contain any rankshift.


In relatively recent work the term ‘down-ranked’ is used as a variant of ‘rankshifted’, and there is one instance of this term in Part 1 (1: 41L) - but it is in fact misused. In Playing together yesterday, they tore their clothes, the underlined non-finite clause is said to be down-ranked from the full or finite clause when they played together yesterday, but the relation between the non-finite and finite versions is not a matter of down-ranking or rankshift: they are units of the same rank (clause) and have the same function in the larger construction (which FG treats as a clause complex).

 

4.2 Complexes

CF's account of the clause complex is internally inconsistent and misrepresents the nature of this category in FG. It is presented as the highest unit in the rank scale in display [8] above, and the next sentence explicitly says that it is the largest level of structure - but two sentences later we are told that the clause is the largest structure.

CF remarks that ‘in some grammars, a clause complex is called a sentence’ (1: 41L), and indeed in Halliday's early work the highest unit on the rank scale was the sentence. But the change from ‘sentence’ to ‘clause complex’ was not a mere change of terminology: it was part of a fundamental revision of the concept of rank. FG now uses ‘sentence’ for a unit of writing rather than grammar. Within the grammar, there are just four ranks (IFG: 9), for the clause complex is not a rank, not a ‘ranking unit’. Rather, for each rank there is a corresponding complex.

Compare, for example (omitting the highly restricted morpheme complex):

 

[10]     i         He ate his lunch before he played with his friend.                             [clause complex]

ii         He [was trying to start] the car.                                                          [group complex]

iii         He wrote a [very aggressive] letter.                                                     [word complex]

 

In [i] (from CF1: 41L), we have, according to FG, two clauses combined without rankshift, i.e. without one being part of the other - more specifically, a main clause followed by a subordinate clause. Similarly, the bracketed expression in [ii] is a group complex, consisting of a combination of two verbal groups. And in [iii] the bracketed expression is a word complex consisting of two words, an adverb modifying the adjective aggressive. CF doesn't mention these smaller complexes, and doesn't deal with the constructions that FG describes in terms of group or word complexes: I return to this issue in Section 10 below.

 

4.3 Morphemes, morphology and inflection

FG has the morpheme as the lowest unit on the rank scale, but says hardly anything about it: it provides virtually no guidance on this area of grammar.

CF does not operate with the distinction between syntax and morphology explained in AG, Section 4.1: the tacit assumption is that words can be analysed into morphemes in essentially the same way as groups are analysed into words. But this is not a sound assumption: the motivation for the syntax-morphology division was precisely that the grammatical operations involved are by no means entirely the same, and traditional grammar makes no use of the concept of morpheme. CF has a fairly concrete view of morphemes: like other units on the rank scale they are said to be ‘visible in written texts or audible in spoken texts’ (1: 35L). As far as inflectional morphology is concerned, morphemic analysis then works well enough for regular forms (plural dogs can be analysed into dog + s, past tense talked into talk + ed) but it can't cope with irregular forms like those we have in plural men or sheep, past tense took or bent, and so on: these can't be divided into two successive units in a comparable way.

Inflection, CF says, involves morphemes ‘used to express grammatical contrasts such as singular/plural’, etc. (1: 36L), but it doesn't offer a morphemic analysis of any irregular forms like those I have mentioned. Indeed, in Untried men competed against him (the example cited above that is used to illustrate the rank scale), plural men and accusative him are treated as single morphemes.


A more satisfactory way of handling inflection is in terms of lexemes and inflectional forms, as discussed in AG, Section 4.2. Men will then be analysed not as a sequence of two morphemes but into two components of different types: a lexeme man and an inflectional property ‘plural’. Similarly, took will be analysed as the combination of the lexeme take and the inflectional property ‘past tense’, and so on. The rules of inflectional morphology will then specify how plurals, past tenses, etc. are formed: by adding an affix, changing the vowel, and so on.

I should add that there are a number of inflectional systems that are not recognised in CF:

 

o       It says that pronouns don't inflect (1: 38R), but notes that personal and relative pronouns have case, which is a classic inflectional category (and is treated as such in the section on nouns, 1: 36R).

o       CF likewise says that adverbs and determiners don't inflect (1: 38L, 39L), but a number of short adverbs inflect for grade (soon/sooner/soonest) and demonstrative determiners inflect for number (this/these, that/those).

 

Furthermore, although CF explains inflection as a matter of word properties (1: 36L), the discussion of adjective inflection covers not only such morphological formations as larger and largest, but also syntactic constructions like more swollen and most swollen. In this context, the statement that adverbs don't inflect is even more obviously in conflict with what is said about grade in adjectives, for there are innumerable comparative and superlative expressions headed by adverbs, such as more carefully and most carefully.

 

4.4 Groups and phrases

There is just one rank intermediate between clause and word on the rank scale, but CF, like FG, distinguishes between groups and phrases. The main group classes are noun group, verb group and adverb group, and the only phrase class is the preposition phrase. Three problems arise here.

 

o       CF doesn't explain why it is ‘more helpful’ (1: 42L) to make this distinction, restricting ‘phrase’ to preposition phrases instead of using it for all the above classes, as in other modern approaches. It says (1: 40L) that ‘phrases have no logical structure and no head word’, but readers will not understand what is meant by logical structure in this context unless they already have a reasonable knowledge of FG, and the statement that preposition phrases have no head word contradicts what was said on the previous page (1: 39R): ‘prepositions appear ... as head word in prepositional phrases’.

o       CF does not in fact apply the distinction consistently. Thus ‘adverb phrase’ is sometimes mistakenly used instead of ‘adverb group’ (e.g. 1: 38L - adverbs ‘appear as head words in adverb phrases’) and similarly adjectives are said to appear as ‘head words in phrases ... [such as] very hungry’ (1: 36R).

o       In addition, the terms ‘adjectival phrase’ and ‘adverbial phrase’ are sometimes used in the sense these terms have in traditional grammar, which involves a quite different concept of phrase. That is, they are used for prepositional phrases with a function similar respectively to that of an adjective and an adverb, as explained in AG, Section 2.4, where I argued against adopting this usage (1: 38L, 40L, 43R, 44L). What makes the terminological confusion worse is that adverbial phrases in this sense are also in places subsumed under ‘adverbial groups’ together with groups headed by an adverb (e.g. 2: 50L) - and at one point we even find the preposition phrase by himself classified as an adverb (1: 38L).

 

4.5 Group classes

In FG five of the eight parts of speech occur as Head in corresponding group classes. The three main ones are noun, verb and adverb found respectively in nominal groups, verbal groups and adverbial groups; the minor ones are conjunction and preposition, found in conjunction groups and preposition groups. CF (1: 40L-R) gives a very brief presentation of the five group classes.

The most striking feature of the FG scheme is the absence of a class of adjectival groups: in It was very sad, for example, very sad is a nominal group. I will return to this issue in Part 2, for I take this to be a serious weakness of FG. The point to be made here, however, is that CF makes no mention of it. Readers couldn't possibly be expected to know that expressions like very sad are classified together with Sue, a woman, etc. as noun groups, and so will surely be unable to account for the absence of a class of adjective groups.

 


5 FINITENESS AND THE MOOD BLOCK

CF's account of the category of finiteness draws on both traditional grammar and FG, which again leads to inconsistency and confusion because those two treatments are mutually incompatible. Moreover, the relevant parts of CF are not wholly faithful to either traditional grammar or FG.

 

5.1 Finiteness in traditional grammar

In traditional grammar finiteness applies in the first instance to verb inflection: the forms of a verb are divided into finite and non-finite ones. These categories then apply derivatively to clauses: a finite clause is one with a finite verb. The idea behind the term is that finite verbs are ‘limited’ in their use by the inflectional properties they carry, notably person, number and tense. Brings, for example, is normally limited to use with a 3rd person singular subject and to reference to non-past time.

CF's Glossary entry for ‘finite verb’ (1: 41R) is based on this traditional view: a finite verb is defined as ‘any form of a verb’ inflected for number, person and tense'. The tense inflections, however, are given as past, present and future, contrary to the earlier statement (1: 37R) that verbs inflect for past or present tense. It is a matter of controversy whether English can properly be said to have a future tense at all, but it certainly doesn't have a future tense inflection. The glossary gives the example The teacher will be Ms Jago, saying that will inflects for singular number, third person and future tense, but traditional grammar would say that it is a future tense auxiliary, which is itself a present tense form (contrasting with past tense would). To put it in terms of the concepts I have been using here, it is not the inflectional component of will that marks futurity but the lexeme component will. I would add that although traditional grammar would say that this will is a 3rd person singular form such a statement has no justification: will remains constant whatever the person and number of the Subject. It is then unreasonable to say that it agrees with the Subject, for that implies that as you vary the person/number of the Subject you vary the verb to match.

The Glossary entry for non-finite verb covers only the present and past participles; the traditional infinitive should be included here, but is given an entry of its own. The entry for the infinitive correctly says that it does not inflect for tense, but that should also be said of the participles (as it was in the section on verbs (1: 37R)): the terms ‘present participle’ and ‘past participle’ are misleading, for they do not contrast as present tense vs past tense. As noted in Section 3.2.3 above, it is a mistake to say, as both these entries do, that non-finite forms cannot take a Subject. It is also a mistake to say that the infinitive verb typically consists of to + the lexical form of the verb: the term ‘lexical’ applies to a class of verbs (verb lexemes), not an inflectional form; in earlier discussion the form in question had been called the ‘base form’ (1: 37R).

 

5.2 Finiteness in FG

FG's account of finiteness is very different. ‘Finite’ is used in the following three senses.

 

(a) In the first place it is the name of a function. The Finite is an element in the structure of the verbal group and of the clause (on the interpersonal dimension). In the verbal group it combines with the Event and (optionally) one or more auxiliaries: had been eating, for example, has the analysis shown in [11]:

 

[11]                                          had     |   been      |   eating

Finite   |   Auxiliary  |   Event

 

In the clause it combines with the Subject to form the Mood element: in He had been eating his sandwiches, for example, the Subject he and the Finite had form the Mood, while been eating his sandwiches is the Residue.

 

(b) Secondly, ‘finite’ is used for a verb class: verbs are divided into three classes, finite, auxiliary and lexical. Adding class labels to the function labels shown in [11] thus gives:

 


[12]                                    had             |   been               |   eating

Finite           |   Auxiliary          |   Event

finite verb    |   auxiliary verb   |   lexical verb

 

Finite verbs are those that occur with the function Finite.

 

(c) Then, thirdly, ‘finite’ can apply derivatively to verbal groups and clauses. Thus had been eating is a finite verbal group and He had been eating his sandwiches is a finite clause.

 

The function ‘Finite’ and the class ‘finite’ are sometimes called, more fully, ‘Finite Operator’ and ‘finite verbal operator’ respectively.

CF's account of finite verbs in the section on verbs (1: 37L) is based on FG and hence is inconsistent with the glossary entry discussed above. According to the latter, all verbs (i.e. verb lexemes) have finite forms, as in traditional grammar: ate, for example, is one of the finite forms of the verb eat. But for FG finite verbs form a fairly small class contrasting with auxiliary and lexical verbs: according to FG, ate is a lexical verb, not a finite one.

Note also that CF misrepresents FG with respect to the subclassification of finite verbs. For FG finite verbs (finite verbal operators) are divided into two subclasses: temporal and modal (IFG: 115-6).[5] CF, however, says: ‘Finite verbs may be auxiliary verbs that signal tense, or modal verbs ...’. This is a mistake: as we saw above, FG has finite verbs as distinct from auxiliary verbs as well as from lexical verbs, and indeed CF itself had earlier given finite, auxiliary and lexical as three types of verb (1: 35R) and does so again in Part 2 (2: 50R).[6]

 

5.3 The Mood block

There are major empirical problems with CF's account of the ‘Mood block’ ( = FG's Mood element), consisting of the Subject and the Finite (2: 52L). It says that ‘the order of the Subject and the Finite verb determines the Mood of the clause’; and then goes on to say, firstly, that ‘Subject + Finite defines the Declarative Mood’. But this is not true, as is evident from examples like:

 

[13]     i       Never had I seen such chaos.                                                                   [declarative]

ii       More important had been the moral objections.                                          [declarative]

iii       And after that you were planning to go where, exactly?                           [interrogative]

iv       What a fool I had made of myself.                                                            [exclamative]

 

In the first place, there are declaratives like [i-ii] which do not have that order. And secondly, there are non-declaratives like [iii-iv] that do. ([iii] is a type of interrogative typically found in contexts of sustained questioning, as in court-room cross-examinations or quiz shows.) The Subject + Finite order doesn't define declarative mood: it is the default order for clause structure, with departures from it triggered sometimes by the choice of non-declarative mood, sometimes by information-packaging factors.

Nor does the absence of Subject and Finite ‘define’ imperative clause. Most imperatives have no Subject, but it is not at all uncommon for an imperative to have a Subject: Nobody move!, You be quiet, You have a nice day, Don't you speak to me like that, and so on.

 

PART 2. PROBLEMS WITH FUNCTIONAL GRAMMAR

 

There are, in my view, numerous places where strong arguments can be given against the analysis presented in IFG. This is not the place for a comprehensive discussion of this matter, and I will confine myself to two topics that relate directly to issues discussed in Part 1: the definition and explanation of the Complement function and the treatment of expressions headed by adjectives.

 

6. COMPLEMENTS

I noted in Section 3.3 that CF's glossary says, following FG, that a Complement ‘has the potential to be [S]ubject of the verb’, but fails to mention that FG recognises a systematic exception to this condition. Attributes like those underlined in Sue is a nanny and Sue had been ill cannot be made Subject in any related clause but are nevertheless Complements.

As far as FG is concerned, the question arises as to why Attributes are analysed as Complements if they don't have the property of being a potential Subject. This problem is handled, in effect, by revising the definition: ‘A nominal group not functioning as a Subject will be a Complement’ (IFG: 123).[7] But this won't do either. Complements contrast with Adjuncts, which on the next page are said to be ‘typically realised by an adverbial group or a prepositional phrase (rather than by a nominal group)’; there is no mention here of the fact that less typical realisations of Adjuncts include nominal groups, as in She arrived this morning, We met several times, I'll do it my way. Indeed, on the following page (125) IFG itself gives an example of this kind, The teapot the duke had given to my aunt last year, with last year explicitly labelled as Adjunct, and though it is not given a class label its status as a nominal group is uncontroversial.

This is clearly in conflict with the revised definition of Complement. Compare, then:

 

[14]                                                             Potential Subject?  Nominal group?    Complement?

  i         Kim hired a nanny.                              Yes                       Yes                  Yes

ii         Sue is a nanny. / Sue is ill                   No                        Yes                  Yes

iii         They got married last year.                  No                        Yes                  No

 

According to FG practice the underlined elements in [i] and [ii] are Complements, while that in [iii] is an Adjunct, but the first definition of Complement admits only [i], and hence is too restrictive, while the revised definition admits all three, and hence is not restrictive enough.

This problem can be resolved if we replace FG's account of the Complement function by that adopted in many other modern approaches, where a Complement is an element licensed by the verb, as explained in AG, Section 3. The underlined elements in [14i-ii] qualify as Complements because they are admissible only with certain verbs - they could not occur, for example, with such verbs as disappear or cease. But last year in [iii] is not restricted in this way as it can occur with any verb: it is therefore not a Complement but an Adjunct.

         On this account, a Complement needn't be a potential Subject and nor need it be a nominal group. In He threw the ball into the pool, the preposition phrase into the pool, no less than the nominal group the ball, has to be licensed by the verb and hence qualifies as a Complement.

As demonstrated in AG, this conception of Complement not only solves the problem that arises with FG's definition, but has the added advantage that it generalises to other constructions. In [His DIVE into the pool] created hardly a ripple, for example, the underlined preposition phrase is a Complement in the bracketed noun phrase, licensed by the noun dive, marked as Head by capitals.

 

7 GROUPS HEADED BY ADJECTIVES

 

Of the four open word classes noun, verb, adverb and adjective,[8] FG has corresponding group classes for the first three:

 

[15]               Word class           Group class                           Examples, with Head underlined

  i         noun                   nominal group                        dogs       some big black dogs

ii         verb                     verbal group                          worked   had been working

iii         adverb                 adverbial group                      slowly     more slowly than ever

 

This pattern does not extend, however, to adjectives. In She was [hungry] and This is [highly unusual] the underlined words are adjectives but the bracketed groups that they head are nominal groups, not adjectival groups.[9] No explanation or justification is given for this analysis. If the bracketed groups were classified as adjectival, this would permit a more general statement where (leaving aside various elliptical and similar constructions) the Head function in a group of a given class was realised by a word of the corresponding word class. Thus the Head in a nominal group would normally be a noun, that in a verbal group a verb, that in an adverbial group an adverb and that in an adjectival group an adjective.

A consequence of FG's failure to recognise a category of adjectival group on a par with nominal, verbal and adverbial group is that there is no systematic description of expressions with an adjective as Head. (It is worth noting in this connection that IFG's index has a single entry for ‘adjective’ - referring to a discussion little more than a page in length.) In particular, there is no satisfactory account of the complementation of adjectives.

Consider, then, such examples as these:

 

[16]     i   a.    his WILLINGNESS to help us                        b.   his HABIT of surprising us

ii   a.    He is WILLING to help us.                             b.   He is CAPABLE of surprising us.

 

For FG, the capitalised words in [i] are nouns functioning as Head of the nominal group, and the underlined expressions are Qualifier. But the examples in [ii] are handled quite differently (cf. IFG: 514, 518): the capitalised words are adjectives, but they are functioning in the structure of a verbal group, is willing and is capable, and the underlined expressions are likewise verbal groups, subordinate to the preceding ones. A more transparent and systematic analysis would treat the adjectives like the nouns:

 

[17]     i   a.    [his WILLINGNESS to help us]                      b.   [his HABIT of surprising us]

ii   a.    He is [WILLING to help us].                           b.   He is [CAPABLE of surprising us].

 

The bracketed sequences are noun phrases in [i], adjective phrases in [ii]; they are headed by nouns and adjectives respectively, and the Head is followed by a Complement.

 

PART 3. SIMPLIFYING FUNCTIONAL GRAMMAR

 

FG's description of English is extremely complicated and naturally requires massive simplification in any presentation oriented towards schoolteaching. That, however, is a very difficult undertaking, and I have to say that CF has not gone about it in the right way. For the only kind of simplification we find here is simplification by omission.

A great deal must of course be omitted, but omission itself is not enough. Consider again the issue of expressions headed by adjectives, as in:


[18]     i         a very annoying delay                            [word complex as Modifier in nominal group]

ii         The delay was very annoying.                    [word complex as Head in nominal group]

 

FG treats very annoying as a word complex, with the adjective annoying as Head and the adverb very as Modifier: as noted in Section 4.2, the concept of complex applies not just to the combination of clauses but also to the combination of groups/phrases and of words. In [i] this word complex is Modifier to the Head noun delay, whereas in [ii] it functions as head of a nominal group containing no other elements. Now CF omits this, but offers no alternative analysis in its place. This means that it says nothing about how to analyse expressions headed by adjectives. That won't do, because these are basic and very frequent constructions. You have to make a choice between presenting FG's analysis and finding an alternative analysis: you can't just ignore the construction altogether. In Section 7 I argued for an alternative analysis, one where expressions headed by adjectives are adjective phrases, but the point I'm wanting to emphasise here is that if you decide, on grounds of simplicity, not to retain some aspect of FG's description you need to be prepared to present some other description if you are dealing with some core area of the grammar.

In the following sections I will discuss four issues where CF could be improved by simplifying.

 

8. WORD ‘SUPERCLASSES’

Early in Part 1 (1: 35R), CF lists the open and closed word classes, the parts of speech, and then adds that they fall into three groups (which I'll refer to as ‘superclasses’):

 

[19]               Superclass         Parts of speech included in the superclass

  i         nominals              nouns, adjectives, determiner, numerals, pronouns

ii         verbials                verbs, prepositions

iii         adverbials            adverbs, conjunctions

 

This is something that can be simplified just by omission. The three superclass terms are not used again, and the classification does not even determine the organisation of the following discussion of word classes, which is based on the distinction between open and closed classes. A lot of terminology is inevitably introduced in CF, but there's no point in including in the list terms that are of no use.

 

9 GROUPS AND PHRASES

I noted in Section 4.4 that CF follows FG in making a terminological distinction between phrases, of which there is just one class, preposition phrase, and groups, of which there are several classes, noun group, verb group, and so on. This is an unnecessary distinction. There is only one rank between clause and word: we don't divide clauses into one or more phrases and then each phrase into one or more groups. The difference between noun groups and preposition phrases is a matter of their structure, not of their rank. And if there's only one rank, it is simpler to have only one term, and the natural one to use is ‘phrase’. Making this simplification would have no significant effect on anything else in the grammar.

No doubt it would be fairly easy to teach students to call expressions like the dog and with the dog a noun group and a preposition phrase respectively, but what is the point of doing this if they don't know why the distinction is being made? And CF certainly doesn't provide any such explanation. Dropping the distinction is a simplication that can be made at no cost.

 

10 COMPLEXES AND EMBEDDING

A more radical suggestion I would make is to drop the concept of a clause complex. I mentioned in Section 4.2 that CF fails to mention that the concept of complex applies not only to clauses but also to groups (was trying to start) and words (very aggressive): this is another case where simplification by omission won't do, for we need to have some account of the analysis of such high frequency constructions. A second omission is that CF doesn't deal with FG's distinction between a complex and a construction involving rankshift, or embedding. This distinction is illustrated in:

 


[20]     i         I thought that I had informed his wife.                                              [clause complex]

ii         I regretted that I had informed his wife.                                                       [rankshift]

 

In [i] we have two successive clauses (separately underlined) forming a clause complex, whereas in [ii] the underlined clause is rankshifted to function as Complement within the larger clause that is co-extensive with the whole sentence (cf. IFG: 476). The basis for the different structures is that in [ii] but not [i] we understand it to be a fact that I had informed his wife. Whatever the merits of this account the distinction is surely too subtle for inclusion in a grammar oriented towards schoolteaching: we need to simplify and provide an analysis that focusses on the evident grammatical similarity between the examples and handles them in the same way.

By failing to mention any construction where one clause is rankshifted to function as an element within the structure of another clause, CF tacitly invites the reader to assume that examples like [20ii] will be clause complexes, like [i]. But a better way of simplifying FG's analysis is to drop the clause complex analysis and treat that I had informed his wife as a Complement in both examples.

There are several reasons why this is the better solution.

 

(a) In the first place, the distinction between Complements and Adjuncts that we have drawn in clause structure applies to the subordinate clauses of FG's clause complexes. In [20i], for example, that I had informed his wife has to be licensed by an appropriate verb (you can't say *They inquired that I had informed his wife, and the like), but in [10i] before he played with his friend is not restricted to occurrence with a particular subclass of verb. On a rankshift analysis the two subordinate clauses will thus be Complement and Adjunct respectively.

 

(b) We won't need the other kinds of complex either. Thus in [10ii] ws trying to start will not be a unit: rather, to start the car will be a rankshifted clause functioning as Complement, licensed by the verb try. And in [10iii] very aggressive will be a rankshifted adjective phrase functioning as Modifier within the noun phrase a very aggressive letter.

 

(c) This means that there will be no need for a concept of a complex distinct from a rank. The constructions that FG treats as complexes can be handled in terms of categories that are independently needed elsewhere in the grammar.[10]

 

11. THE MOOD FUNCTION AND MULTI-DIMENSIONAL STRUCTURE

It is a well-known property of FG that it regards clauses as being simultaneously structured on a number of different dimensions, representing its properties in respect of the textual, interpersonal and ideational ‘metafunctions’. In [21] I show the interpersonal and ideational structures for the clause Kim had seen the intruder clearly:

 

[21]     i         Interpersonal analysis

            Kim        had     |  seen          |  the intruder     |  clearly.

            Subject  |  Finite   |  Predicator  |  Complement    |  Adjunct

            Mood                   |  Residue

 

ii         Ideational analysis

            Kim        had seen                  the intruder     clearly.

Actor |   Process                    |  Goal                |  Circumstance

 

On the interpersonal dimension there are two levels of structure. On one we have five elements, on the other just two: the Subject and the Finite combine to form the Mood function (‘Mood block’ in CF's terminology), while the other elements form the Residue.


I believe there is considerable scope for simplification here. I take up first the status of the Mood function and then turn to the question of co-existing structures.

 

11.1 The Mood function

The first thing to note about the Mood function is that it is anomalous in three respects relative to the most straightforward type of function.

 

o       Its realisation is not a unit of any kind: the sequence Kim had is not a clause, not a group or phrase, not a word.

o       In the simplest type of clause, such as Kim saw the intruder, the Finite is not distinct from the Predicator: they are said to be ‘fused’. The verb saw is thus the realisation of Finite and Predicator combined.

o       The Finite function is realised by finite verbs, but these do not form a class of verb lexemes nor a class of inflectional forms. As we observed in Section 5.2, finite verbs contrast with auxiliary verbs and lexical verbs. In Kim may have seen the intruder the have is an auxiliary verb but it is a form of the same verb lexeme, have, as the finite verb had of [21i], so it follows that finite verbs are not a class of lexemes. On the other hand, saw is the past tense form of see just as had is the past tense form of have, yet while past tense had is a finite verb, past tense saw is not, so finite forms are not a class of inflectional forms (as they are in traditional grammar).

 

This cluster of exceptional properties provides grounds for asking whether we really need Mood and Finite as functional elements. And the answer is no. The main work they do is to enable us to state the difference between declarative and interrogative clauses in structural terms. Compare, then:

[22]     i         Your father has arrived.                                                [declarative: Subject + Finite]

ii         Has your father arrived?                                  [yes/no interrogative: Finite + Subject]

 

In the declarative [i] the Subject precedes the Finite, but the Finite comes before the Subject in the interrogative counterpart [ii] (more specifically, it is what FG calls a ‘yes/no interrogative’ since it expresses a question to which the answers are ‘yes’ and ‘no’). But we don't need this complicated apparatus to describe the difference between [i] and [ii]. We can simply say that the interrogative is formed by inverting the Subject and the auxiliary verb - where ‘auxiliary verb’ now denotes a class of verb lexemes contrasting with lexical verbs. If the declarative doesn't contain an auxiliary verb, we need to insert a form of do so that the rule of Subject-auxiliary inversion can apply. Thus the interrogative counterpart of Your father arrived yesterday is Did your father arrive yesterday? (not *Arrived your father yesterday?).

This traditional idea of Subject-auxiliary inversion is surely very easy for students to grasp - a good deal easier, I would suggest, than the concepts of Mood and Finite elements that commonly fuse with Residue and Predicator respectively.

 

11.2 Do we need separate interpersonal and ideational structures?

If we drop the Mood and Finite functions the structure in [21i] can be simplified to:

 

[23]               Kim        had seen       |   the intruder     |   clearly.

            Subject  |  Predicator      |   Complement   |   Adjunct

 

This divides the clause into the same units as the ideational structure [21ii]: we no longer have two structures differing in how they break down the clause into its parts.

We can now simplify further. As the Predicator is realised here by the whole verbal group, there is no need for a distinction between Predicator and Process. And Circumstances are simply a subtype of Adjunct, as stated in IFG (p. 125). As for ideational functions such as Actor, Goal (and a fair number of others: Behaver, Senser, Phenomenon, etc.), these can be treated as semantic roles associated with Subjects and Complements.


One advantage of this is that it deals with the problem of prepositions mentioned above in connection with CF's example [7iii]. Compare, then:

 

[24]     i         He gave the parcel to his brother at home. (=[7iii]).

ii         He gave his brother the parcel at home.

 

In both it was his brother who received the parcel, so the Recipient role is most appropriately associated with the noun phrase his brother. The underlined phrase in [i] is Complement, but the to serves only as a marker of the Recipient role, not as part of the description of the person who filled that role. So when we describe the semantic roles associated with Complements realised by preposition phrases it is the noun phrase within the preposition phrase that expresses the Participant.

Nor do we need a separate structure for the textual metafunction. For [23] we can simply say that the Theme role is associated with the Subject. And for The intruder Kim had seen clearly, which has the same four units in a different order, we can say that the Theme role is associated with the Complement.[11]

I am suggesting, therefore, that a valuable simplification would be to dispense with separate clause structures for the three metafunctions, where by ‘separate structures’ I mean divisions of the clause into a distinct sets of units. This might at first seem a heretical suggestion, but that would be a superficial judgement. The important thing about FG's metafunctions is that they involve different kinds of meaning. Compare, for example, such pairs as the following:

 

[25]     i   a.    Kim saw the intruder.                               b.   The intruder saw Kim.

ii   a.    Kim saw the intruder.                               b.   Did Kim see the intruder?

iii   a.    Kim saw the intruder.                               b.   The intruder Kim saw.

 

The examples in [i] differ with respect to what in other frameworks is generally called ‘propositional meaning’: with this type of meaning we are concerned with what conditions would have to be met for the clauses to be used to say something that is true. But those in [ii] don't differ in this way: [iia] would characteristically be used to make a statement and [iib] to ask a question, and while statements can be true or false, questions cannot. And in [iii] the difference has nothing to do with truth: it's not possible for [iiia] to be true when [iiib] is false, or vice versa. The difference is one of information packaging: the intruder is assigned a kind of prominence in [b] that it lacks in [a].

There are thus different kinds of meaning difference between the pairs in [25], but it doesn't follow that the different kinds of meaning are expressed in FG's three different kinds of grammatical structure.

All linguists agree on the need to distinguish various different kinds of meaning, but only a fairly small minority go along with the thesis that they correlate with different grammatical structures. There are certainly grounds for questioning that thesis. Note, for example, that what I have called Subject-auxiliary inversion not only serves to distinguish certain kinds of interrogative clause from declaratives, but may also be triggered by the placement of various kinds of unit in initial position, as in Only then did Kim see the intruder, where such initial placement has to do with information packaging, not interpersonal meaning. Nor is there any correlation between aspects of interpersonal meaning and the grammatical distinction between Complements and Adjuncts. I am arguing, therefore, that at least in the interests of simplification we could dispense with the separate clause structures thesis without losing anything essential: it looks to me like a very worthwhile simplification.

 

***


12. CONCLUSION

My primary aim in this paper has been to warn readers of the inconsistency, error and confusion in CF. Regrettably, the scale and seriousness of its defects make it unusable as a teachers' resource: I have recommended that some kind of revised edition of Parts 1 and 2 should be posted on the ETAQ website and that members should be warned to disregard the original printed versions. It would be a matter of great concern if teachers were actually to present that material to their students - to tell them that won't and capable of are adverbs, that Sam's in Sam's folder is simultaneously a possessive pronoun and a determiner, that the Theme of a clause represents known information, that imperative and non-finite clauses can't have a Subject, and so on.

I have also wanted, secondarily, to question certain aspects of FG's description of English and to suggest some ways in which it might profitably be simplified for use in schools. I am aware that many teachers are deeply committed to FG, and I was warned that readers might be alienated by criticism of it. I hope that that will not happen. It is difficult to believe that the features of the model that teachers find appealing are dependent on the particular ways in which the Complement function is defined and adjectives analysed. It is important that teachers should be aware that while FG may be a great improvement over traditional grammar it is not the only alternative to the latter. Within the academic discipline of linguistics it occupies very much a minority position.

There is, moreover, one aspect of FG that underlies some of the problems I have noted with CF. FG pays less attention than most other contemporary approaches to the issue of providing grammatical evidence and criteria for its analyses. Quite often, semantic rather than grammatical factors seem to determine the grammatical analysis. I do not deny that our major task is to show how sentences convey meaning, but this is best done by relating grammatical form to meaning, not by allowing semantic considerations to distort the description of grammatical form. There are a number of quite basic errors in CF, such as the classification of won't and capable of as adverbs, that must be due to this feature of the underlying model: I don't see how anyone whose training had been within the framework of any other modern approach could have come up with an analysis like this.[12]

 



[1] I shall be happy to enter into email correspondence concerning this paper: my address is <rhuddleston@aapt.net.au>. I wish to thank Geoff Pullum and Anne Horan for helpful comments on earlier drafts.

[2] This has appeared in three editions, published by Edward Arnold: the first two (1985, 1994) were written by M.A.K. Halliday alone, while the last (2004) was a revision by C.M.M. Matthiessen.

[3] FG in fact has them functioning as Head in nominal group structure, but it is simpler to have them functioning directly in clause structure.

[4] The discussion, moreover, gives an incorrect account of examples like The adventurer climbed [to the peak with a snow-covered cap]. The underlined preposition phrase is said to be embedded (i.e. rankshifted) into the unit I have enclosed in brackets, which is said to be an adverbial group. It is not an adverbial group but a preposition phrase, consisting of the preposition to + a noun group, and the rankshift here then is into the Qualifier position in a noun group, just like that in [9iv].

[5] The temporal operators include did/does, had/has, was/is, etc., while the modal operators include can, may, could, might, must, etc.

[6] The confusion is made worse by the fact that in the example given to illustrate a modal finite, The small boy could have eaten his lunch, the underlining highlights not just modal could but also auxiliary have. And in Table 4 (2: 57) ‘modal’ is incorrectly contrasted with ‘finite’ in the heading for the first element in verb group structure.

[7] Recall that, as noted in Section 4.4, FG treats Attributes with an adjective as Head, like ill, as nominal groups, not just those with a noun as Head, like a nanny.

[8] An ‘open’ word class is a large class that readily accommodates the addition of new members as the vocabulary develops.

[9] There is a parenthetical remark in IFG (p. 331) saying that this type of nominal group is ‘sometimes referred to distinctively as ‘adjectival group’’, but this is potentially misleading and confusing: FG has no adjectival group that is distinct from the nominal group and comparable to the adjective phrase of other theories.

[10]    Like CF, I have confined my attention here to constructions involving subordination, as opposed to coordination.

[11]    One of CF's examples has the Mood block as Theme: Is the thief going to give evidence? (2: 53L). This is a highly questionable analysis, however, one which would be better omitted from a school-level resource, especially as CF offers no commentary, no explanation as to why we should take the underlined sequence as Theme.

[12]    As it happens, the opposite kind of error is found in a recent FG textbook for primary teachers, where the adverb probably is classified as an auxiliary verb: L. Droga & S. Humphrey, Grammar and Meaning: An introduction for primary teachers, Berry: Target Texts (2003: 38). Again, one couldn't imagine a non-FG textbook making this gross error.