THE UNIT"GROUP"
Home Up MICROCONCORD English verb in 18th-19th grammars MODO Y MODALIDAD ENGLISH VP SISTEMICA SUBJUNTIVO EN BROWN J.R. FIRTH SUBJUNTIVO INGLES THE UNIT"GROUP" R.J. DI PIETRO LINGUISTICA Y DIDACTICA HENRY SWEET

 

 

 

THE UNIT "GROUP" IN HALLIDAY'S AN INTRODUCTION TO FUNCTIONAL GRAMMAR

Luis Quereda

Universidad de Granada

 

    It is at the rank of phrase[1] that there is most confusion - because there are here the greatest difficulties - in the description of English. (Halliday, 1961: 252-253).

 

1. INTRODUCTION   

The initial purpose of this article was to discuss all the aspects of the unit group (phrase) in systemic linguistics in general, but this proved to be not an easy task, basically for two reasons. The first is that one can never be sure that one has collected all the information, since as Halliday (Halliday & Martin, 1981:13) himself recognizes much of this literature has never been published and that which has is scattered in a variety of different locations. The second reason, and perhaps the most important, is that there is a great diversity of opinions among systemic linguists on the topic - Hudson and Fawcett, for example, disagree with Halliday in quite a few respects and therefore have different analyses. To try to make a comprehensive contrast of these analyses would be beyond the possibilities of an article like this. Therefore, even though we will try to be as comprehensive as possible, we are going to concentrate on Halliday's view and on the opinion of those linguists, such as Berry, Muir, Scott et al., Sinclair, etc., who have openly recognized that they have followed Halliday's model.

   All the problems are not solved with this restriction of coverage, however. In its early model - Scale and Category - Halliday (1961) established a theoretical framework (a set of four categories: unit, structure, class and system) which could be used to describe the grammatical patterns of all languages. But it was soon proved that the model did not work so nicely. Halliday, however, did not try to solve the possible shortcomings that the model had since, as Butler (1985:39) suggests, "there was a considerable shift in the emphasis of Halliday's work after about 1965, and because of this the problems inherent in Scale and Category linguistics have never been satisfactorily worked out". Since 1965, Halliday has given increasing prominence to semantic phenomena.

   It was not until his publication of An introduction to functional grammar (1985) that Halliday took up these matters again. ("The chapters that follow take as their domain the traditional realm of syntax, the terrain from the sentence to the word" (1985:xxi)). But Halliday's main objective with this grammar is to explain the functional elements of the English clause, and, therefore, some details of the structural description are either implied or simply not stated. To find the clue to his attitude in relation with certain aspects of his organization of the rank group we will have to take into account his previous work and that of his followers. Thus our interest is going to be centred in these two poles: Halliday's Scale and Category model and Halliday's An introduction to functional grammar. This will give us the opportunity not only of analyzing everything in relation to this unit, but also of studying the links and the gaps that there are between these two important stages in systemic linguistics.

   Halliday organizes the rank group in his An introduction to functional grammar in the following way:

   - He distinguishes two different units within the same rank: phrases (a reduced strain of clause) and groups (an enlarged strain of word).

   - Within phrases, he only includes one type of structure: prepositional phrases.

   - Within groups, he distinguishes three basic structures: the verbal group, the nominal group and the adverbial group.

   - Besides, he also introduces two more groups: the preposition group and the conjunction group, although their position within the group rank is not really clarified.

 

   This organization of the group rank presents some problems when one tries to apply it to the description of English. It has inconsistencies and leaves some questions unanswered. The basic problems which it presents can be summarized in the following way:

   (a) The distinction between group and phrase is not stated in a clear way.

   (b) Halliday does not talk of the adjective group, even though this structure has been traditionally recognized by most linguists as a group or phrase. This causes some problems when we are faced with the analysis of the adjective and its modifiers.

   (c) The relation between the three basic groups (verbal, nominal and adverbial) and the other two (preposition and conjunction) is not stated, and since they are such different structures from the functional point of view, it is difficult to grasp what the real entity of the unit group is.

   (d) There are some blank spots in this description since there are many words operating within clause structure which cannot be classified under any of these groups or phrases (especially linkers and binders) and then it is difficult to know whether, in the structure of the clause, we can only find groups and within groups words, or whether words can enter into the structure of clause directly.

   Although these problems are really interconnected, and it will not always be easy to deal with one without making reference to the other, we are going to deal with them under two separate headings: (a) the distinction between group and phrase and (b) phrase: a multi-layered rank, the latter being subdivided into two subheadings: phrases as constituents of clause structure and phrases not operating at clause structure

 

2. THE DISTINCTION BETWEEN GROUP AND PHRASE 

   After Halliday introduced the term group in an article devoted to the analysis of Chinese, systemic linguists, with some exceptions, like that of Hudson, have systematically used the term group instead of the more common term of phrase, in contrast to what is normal in English linguistics (see, for example, Lyons 1968, Allerton 1979, Matthews 1981, Huddleston 1984, Quirk et al. 1985, etc.). But in contrast to all systemic linguists, Halliday proposes to keep both terms, group and phrase, for the description of English. Thus, while Halliday distinguishes between phrase and group, other systemic linguists reject the term phrase and use only that of group.

   None of these linguists, however, justifies his decision for adopting the term group and rejecting that of phrase. Only Young and Sinclair give some explanation. They reject the term phrase because "traditional theory identifies as phrases many things that we shall not regard as groups" (Young, 1980:12) and because it "has so many meanings already in grammars ... [that] ... <phrase= is already used to mean many different parts of a clause, and of a sentence" (Sinclair, 1972:132).

   This is not a very convincing explanation, however. It is true that the term phrase has traditionally been applied to a lot of different things which have nothing to do with the idea that we now have of phrases. But the same can be said of the term group, which has traditionally been associated with combinations of more than one word presenting a certain cohesion[2]. In a quick look at a traditional grammar of English chosen at random (Zandvoort 1969), we can see that the term group has been, to say the least, as inconsistently used as that of phrase. As a matter of fact, Zandvoort uses the term group as a substitute of phrase twice, to avoid repetition[3]. While Zandvoort uses the term phrase only 20 times, he uses the term group about 60 times and under this heading he refers to combinations of words with quite different structure and function[4].

   Although it is not clearly stated by anyone, another possible reason for the substitution of group for phrase is the way transformational grammars use the term phrase. Transformationalists, in contrast to systemicists, use what Hudson (1972:23) calls the <few-ICs-approach=, which means that the verb phrase, for example, not only includes the verb element but also its complements. Systemicists, who use the <many-ICs-approach=, consider the verb and its complements as different constituents, and therefore they prefer the different terms verb group and nominal group to make a clear distinction between these two constituents and the noun and verb phrase of TG grammar[5]. This problem could have been solved in an easier way by giving two different names to the verb phrase when used in its wider sense (cf. Huddleston's extended verb phrase or Allerton's predicate phrase) and leaving the terms verb and noun phrase, as has been the normal practice, for those units whose head is a verb or a noun.

   So, if there are no other reasons, there seems to be no necessity for this change of terminology. But, as we said above, Halliday has never rejected the term phrase. On the contrary, he thinks that both terms, group and phrase, are necessary for the description of English. Halliday, in a seminal paper which is considered the starting point of systemic linguistics, states that "it is useful to have two names for this unit in order to be able to talk about it: I propose to call it the <group= but to make a <class= distinction within it between <group= and <phrase=" (1961:253). Halliday, however, did not really explain in this article the advantages of his decision. In An introduction to functional grammar, Halliday maintains the distinction: "We have to recognize other structural units that are intermediate between [sentences and words] - groups and phrases ... these are in origin just mutations of one or the other ... functionally, the two come together in the middle" (1985:xxi) ... "A group is similar to a phrase ...; we shall use both these terms, with <group= as the more general one" (1985:25).

   Besides, he now gives us some explanation for the distinction: "A phrase - in the sense in which the term is used here - is a reduced strain of clause, while a group is an enlarged strain of word" (1985:xxi) ... "A phrase is different from a group in that, whereas a group is an expansion of a word, a phrase is a contraction of a clause. Starting from opposite ends, the two achieve roughly the same status on the rank scale, as units that lie somewhere intermediate between the rank of a clause and that of a word" (1985:159), or in other words, "although group and phrase are both of intermediate rank as constituents, they have arrived there from different ends: a group is a bloated word, whereas a phrase is a shrunken clause" (1985:192).

   This explanation, although it sounds interesting enough, does not solve the problems involved in a distinction like this. There are two basic problems which need clarification. The first one is to decide which sort of structures fall under the category of phrase and which under that of group; the second will be to see if these two categories, although they belong to the same rank, operate at the same level in clause structure.

   In relation with the first problem, Halliday only includes one structure under the term phrase: the prepositional phrase. All other structures operating at this rank are called groups: the nominal group, the verbal group, the adverbial group, etc. Halliday thinks that prepositional phrases are from the functional point of view minor processes which are more related to clauses than to words: "prepositional phrases are phrases, not groups; they have no logical structure as Head and Modifier, and cannot be reduced to a single element. In this respect, they are clause-like rather than group-like; hence when we interpret the preposition as <minor Predicator= and <minor Process= we are interpreting the prepositional phrase as a kind of <minor clause= - which is what it is" (1985:190) ... "[the] relationship ... [expressed by the preposition] is not unlike a second, minor process which is incidental to the major one for which we have used the term <Process= as label. The preposition itself has some of the qualities of a verb" (1985:142). As Halliday himself points out, the preposition is sometimes similar in meaning to a non-finite verb, with which it is interchangeable. Halliday's examples to prove this are:

          - (He cleaned the floor) with a mop | using a mop.

          - (I'm preparing this) as my prize-winning entry | to be my prize-winning entry.

          - (The police arrested him) without evidence | not having evidence.

          - (She came) for a cup of tea | wanting a cup of tea.

          - (Grass grows) after the rain | following the rain.

   In all this there is only one thing which is absolutely true: prepositional phrases, in contrast to all other groups/phrases cannot be reduced to a single element[6]. Whereas all other groups/phrases are endocentric, prepositional phrases are exocentric. This could be, perhaps, a good reason for making a sharp distinction between prepositional phrases and all the other groups/phrases. But we will discuss this later on in the third part of the article, when we try to establish the different structures operating at the rank group/phrase.

   What I would like to discuss now is the relationship between prepositional phrases and verbal clausal processes, since according to Halliday this seems to be one of the basic reasons for the distinction between phrase and group. Even though this thesis, as it is presented to us, sounds quite interesting, unfortunately it is not built on a very firm basis. And Halliday knows that, since he is aware of the existence of many counterexamples: "not all prepositions have verbal equivalents in this way" (1985:142).

   But the question is not only that it is quite difficult to find verbal equivalents to many prepositions, but that it is difficult to see in many of them any verbal character at all. The most evident example of non-verbal preposition is of in examples such as the two of us, the killing of innocent civilians, the results of the meeting. Halliday is also aware of it, but he justifies himself negating the character of preposition to of, although this does not prevent him from calling of-phrases prepositional phrases (see, for example, 1985:173). Halliday says: "The exception is prepositional phrases with of ... the reason is that of is not a true preposition, but rather a structure marker in the nominal group (as to in the verbal group). Hence of-phrases do not occur as clause elements, except those with Matter function in the sense of <about=, as in Of Julius Caesar it is said that he was ambitious" (1985:190).

   But the preposition of is not the only exception. There are many other prepositions which behave in a very similar way to of. The following is a brief relation of prepositional phrases to which one could hardly assign any verbal value at all:

   i) Examples with prepositional verbs (verb phrases followed by prepositional phrases), which Halliday (1985:143) also mentions as a kind of exception:

          - He looked at the directory.

          - We cannot cope with the work.

          - This law doesn't apply to other people.

          - They objected to the plan.

          - I didn't believe in their pretensions.

          - He talked to them.

 

   ii) Examples of prepositional phrases acting as intensive Complement:

          - Long skirts are in fashion again.

          - Some of the pictures are of great interest.

          - She is in good health.

          - These strawberries are for today's picnic.

 

   iii) Examples in which nominal groups/phrases alternate with prepositional phrases, which Halliday also deals with, although not in relation with this problem:

          - He stayed three weeks | He stayed for three weeks.

          - He ran many miles | He ran for many miles.

          - Do it this way | Do it in this way.

          - He arrived this morning | He arrived during this morning.

          - He jumped the wall | He jumped over the wall.

   If we think that, for example, for three weeks has verbal value because we could perhaps paraphrase it as "he stayed being three weeks", then we would have to consider that the nominal group three weeks has also verbal value, since both examples are synonymous[7].

   iv) Examples of adjective complementation. Adjectives always require a preposition when they have a noun phrase as complement. These prepositions have no semantic function and therefore they do not have any verbal meaning:

          - All sales are subject to tax.

          - I am ready for everything.

          - He was eager for revenge.

 

   v) Examples in which the prepositional phrase is interchangeable with a genitive construction. Although the most typical ones are those introduced by of, there are other possibilities:

          - A college for women | A women's college

          - A day in summer | A summer's day

          - The letter from the general | The general's letter

          - The best universities in the world | The world's best universities

 

   vi) Examples of ditransitive verbs:

          - The governors presented the chairman with a medal | The governors presented a medal to the chairman.

          - He blamed the management for the dispute | He blamed the dispute on the management.

          - He gave me the book | He gave the book to me.

   When he analyses examples of ditransitive verbs, Halliday considers the same phrases, as Adjuncts if they are introduced by a preposition (with a medal, for the management, to him, etc.), but as complements if they appear without the preposition (a medal, the management, him, etc.). These prepositions have hardly anything to do with any verbal process, because if they had, we would have to accept that the noun phrases also have this verbal value, since the role played by them is just the same[8].

   These examples[9] prove that it is not always easy to analyse prepositional phrases as verbal processes or, as Halliday suggests, as reduced clauses. But, besides this, there are other problems when we are faced with the distinction between group and phrase. The one in which we are interested now is why prepositional phrases are for Halliday the only possible phrases. If the only requisite for a structure to be a phrase is that of resembling verbal clausal processes, then it is not difficult to find units other than prepositional phrases at this level in English which can be clearly associated with this type of processes. Halliday himself (1985:226) gives us examples of what he calls process nominal groups (the building of the bridge, the defacing of statues of national heroes, the departure of the boats, his handling of the situation, etc.). We can also mention the so-called sentence-adverbs, which are called so precisely because they are often equivalent to a sentence (i.e., to a verbal process). As with the process nominal groups, they are not treated as phrases by Halliday, but as groups. But if a group is a bloated word and a phrase is a shrunken clause, these adverbs have no other interpretation than that of shrunken clauses. Compare, for example:

          - Fortunately I had plenty of food with me.

          - It was fortunate that I had plenty of food with me.

          - I had plenty of food with me, which was fortunate.

 

          - Frankly, he hasn't had a chance.

          - To be frank, he hasn't had a chance.

          - If I may be frank, he hasn't had a chance.  

   Halliday et al. (1964:159) criticize traditional grammars used for teaching English such as that of Eckersley or that of Thomson & Martinet because they are full of <unclear categories=. The distinction of phrase and group in the way it is presented to us by Halliday is an example of <unclear categories= and therefore since their differences have not been established properly and we do not have objective criteria to distinguish them, it would be better not to take into account such a distinction. Besides, since there is no particular reason for the change of terminology, we do not see any objection to keeping, for the description of the English clause, the traditional term phrase for the rank which goes between clause and word. This, as we will see later on, does not mean, of course, that the rank phrase is homogeneous and that everything which falls within this rank is at the same level. This only means that, in our opinion, it is better to have only one unit, even though this unit comprises structures somewhat different, than to have two different ones which necessarily have to be as undefined and as vague as the one we are trying to replace.

 

3. PHRASE: A MULTI-LAYERED RANK

   Although there has traditionally been a tendency to present the unit phrase as homogeneous, this is far from being true. The general notion <consists of= introduced by Halliday (1961:251), according to which a clause consists of phrases, which consist of words, is a clear oversimplification of the real situation. If we study the structure of the English clause thoroughly, we will find that it is not always easy to prove that a clause consists only of phrases, and perhaps we will even have to accept that there are phrases which can never be constituents of a clause.

   Our final notion of the term phrase will, however, depend on the way we define this unit. There have traditionally been two different definitions of the term phrase, which unfortunately have very often been mixed up. On the one hand, a definition based on functional criteria. According to this definition, a phrase is that unit which realizes the syntactic elements of the unit clause. A clause consists of phrases and the different elements of clause structure (i.e., S, P, O or A) are realized by phrases (i.e., noun, verb, adverb or prepositional phrases). On the other hand, phrase has also been defined morphologically as a combination of words in which we have a central and basic word, the head, modified by some other words, the dependents. Any combination of words in which there is such a relationship of dependency can be considered a phrase.

   These definitions do not cover the same type of structures and, therefore, we will have to specify which is the preponderant criterion if we want to understand how English functions at this rank. Halliday does not make a sharp distinction between these two criteria, and, because of that, his analysis of this rank is rather confusing. Halliday (1985:187-189), for example, studies under the same heading three different groups: the adverbial, conjunction and preposition groups. Halliday puts them together since the three of them are structurally very similar; all of them are groups with a head (an adverb, a conjunction and a preposition, respectively), which can be modified by certain similar items (intensifiers such as rather, even, almost, right, etc., as for example in rather easily, even if, right behind). But, although these groups may be structurally similar, their position in the rank is quite different. While the adverbial group may realize a constituent of clause structure (basically the Adjunct element), conjunction and preposition groups can never realize elements of clause structure.

   In order to avoid these problems, we will analyse the phrase rank by giving in the first place preponderance to the morphological criterion: we will consider phrases all those combinations of words in which we can find a head-and-dependents relationship. But this does not mean that we are going to ignore the functional criterion. On the contrary, this criterion will be essential to establish at which level the phrase in question operates, since, as Allerton (1979:205f) points out, "the phrase does not constitute a single point on the rank scale". In this respect, we should distinguish, at least, two different types of phrases: (a) those operating as clause constituents and (b) those which do not operate as constituents of clause structure.

3.1. PHRASES AS CONSTITUENTS OF CLAUSE STRUCTURE 

   The phrase structure of an example such as:

          - The boy was very often quite nervous during last winter.

can be analysed in at least two different ways. On the one hand, some linguists consider that we could distinguish here five different phrases, more or less at the same level within the rank, which are the realizations of the different syntactic elements of the clause. Thus, this clause can be analysed syntactically as:

          - The boy  (S)  was  (P)  very often  (A)  quite nervous  (Cs)  during last winter  (A).

and morphologically as:

          - The boy  (NP)  was  (VP)  very often  (AdvP)  quite nervous  (AdjP)  during last winter (PrepP).

 

   Taking into account this sort of analysis, most grammarians have distinguished five different basic types of phrases in English, which are traditionally known as noun phrase, verb phrase, adjective phrase, adverb phrase and prepositional phrase. The basic function of these phrases is that of realizing the different elements of clause structure, namely that of Subject, Predicator, Object, Complement and Adjunct.

   But, on the other hand, there are other linguists who would not be happy with an analysis like this and would prefer a more delicate one such as:

 

        ┌────────────── S  ──────────────┐                                    

  Noun Phrase                                         Predicate Phrase                                     

                  ┌──────────┬─────┴────┬────────────────┐                      

            Verb Phrase       Adv. Phrase  Adj. Phrase                Prep. Phrase

                                                                                                  ┌──┴──┐

  ┌──┴──┐  ┌─┴─┐             ┌──┴──┐       ┌──┴──┐                     Prep         Noun Phrase  

                                                                                                   ┌─┴──┐     

 Det        N    Aux  V            Quant    Adv       Quant   Adj                               Mod       N

the       boy    ed   be           very  often      quite nervous               during     last      winter 

   An analysis like this, in contrast to the former one, will have to recognize at least three different levels in the phrase rank: at a first and higher level we have to place the predicate phrase, because, as Allerton (1979:185) points out, this phrase "will most commonly divide up not directly into words but rather into further phrases"; at a second and intermediate level we have to place the prepositional phrase, since this phrase also divides up not directly into words but into a preposition and another phrase; and finally, at a third and lower level we have to place those phrases which, as expected, usually divide up directly into words. At this level, we find the noun phrase, the verb phrase, the adjective phrase and the adverb phrase.

   If we compare Halliday's with these two analyses, there immediately appear some differences which may be summarised in the following way: 

   (a) Halliday does not recognize the adjective group/phrase at the lowest level.

   (b) Halliday (and systemicists in general) does not consider necessary the recognition of a predicate phrase at the highest level.

   (c) In relation with the prepositional phrase, it is not clear whether Halliday's terminological distinction between nominal, verbal and adjectival groups and prepositional phrases implies a recognition that these structures are placed within the group rank at a different level.

   (d) Halliday's preposition and conjunction groups do not enter into any of these two analyses, since these groups are not constituents of the clause. 

   We are now going to comment on these points in a more detailed way, analysing Halliday's ideas and presenting our own points of view. We place special emphasis on prepositional phrases (3.1.3.), since they are one of the most difficult structures to handle within this level. The problem of preposition and conjunction groups will be analysed in a more wider scope under the heading of phrases not operating at clause structure.

 

3.1.1. The adjective group/phrase 

   If we centre our study on the third level we have already established (the level at which phrases normally divide up directly into words), we can see that in general terms Halliday's analysis differs from other analyses only in one important point: while most linguists, even most systemicists, recognize the adjective group/phrase, Halliday does not mention it.

   Halliday does not recognize the adjective group as a group or phrase by itself. He does not talk either of the adjective group as a sub-group of any other group or phrase, as other systemicists do[10]. Halliday does not think it necessary to recognize the adjective group, since he (see, for example, 1961:262 or 1963) gives preponderance to syntactic over morphological criteria for his consideration of class. This makes Halliday assume that every element of structure is realized by only one class of unit. He only distinguishes four basic clause elements[11]: a Subject and a Complement element, realized by nominal groups, a Predicator element, realized by verbal groups and an Adjunct element, realized by adverbial groups and prepositional phrases[12]. Halliday does not think it necessary to make a primary distinction between Object (direct and indirect Object) and Complement (Subject and Object Complement), as most linguists do. Since Objects and Complements are, at primary delicacy, the same type of element, everything operating at this element should be, at primary delicacy, a nominal group. As Butler (1985:33) states, this classification of elements is quite inconsistent, since "he allows for the conflation, into one primary class, of groupings of items which are very similar but not identical". Besides this, Halliday falls into another contradiction, since he allows for more than one realization in the Adjunct element (adverbial groups and prepositional phrases) while accepting only one for the Subject and Complement elements.

   Halliday's attitude towards class[13] led certain other systemicists to abandon the syntactic definition in favour of the morphological criterion, based on the internal structure of units. Fawcett (1974-76/1981), for example, recognizes, by considering their morphological characteristics, the adjectival group as a class in its own right, distinct from the nominal and adverbial groups. Fawcett (1981), like Sinclair (1972:197), stresses the structural parallelism between adjective-headed and adverb-headed groups. Young (1980:12) also recognizes the five typical groups which are usually recognized in English grammars now. Young recognizes that the verbal group, the nominal group, the adjectival group, the adverbial group and the prepositional group have their own characteristic structure.

   Despite other systemicists' recognition of the adjectival group, Halliday (1985) does not talk about it. Halliday only mentions the adjective at the rank of word, realizing the functions of deictic, epithet and classifier (1985:164), and, as stated above, this creates some problems in his description of English phrase structure. Some of these problems can be summarized in the following points:

   (a) Halliday (1985:115) has to recognize that, not only nouns and pronouns but also adjectives can operate as heads of nominal groups: "The Attribute is realized as a nominal group ... [which] ... has as Head a noun or an adjective, but not a pronoun"[14].

   (b) Halliday has also to analyse an example like very big as a group of words forming a word complex, although this goes against his own definition of complexes[15].

   (c) Halliday (1985:171) resorts to notions like that of <sub-modification= and <sub-head= in order to explain the different structural possibilities of the adjective phrase, although he does not explain, for example, what he specifically means by <sub-head=[16]. The acceptance of a sub-head within a group is, in our opinion, a clear implicit recognition of another group (the adjective) within the group he is analysing (the nominal), since Halliday never questions the fact that in every group there is only one head.

   These inconsistencies clearly speak in favour of the recognition of the adjective phrase as a phrase by itself. We cannot forget, moreover, that there is enough morphosyntactic evidence which makes advisable the recognition of an adjective phrase in English. This evidence can be summarized in the following points:

   (i) The adjective phrase directly operates as an element of clause structure in the Cs and Co complements:

          - [Very keen on sport] though she is, she is hopeless.

          - His car is [similar though subtly different].

   We cannot find any reason to maintain that the phrases in the examples above are nominal groups which have an adjective as their head.

   (ii) There are important differences in the functional potential of phrases headed by an adjective and those headed by a noun, a verb or an adverb. Adjective phrases have a distribution of their own, different from all other phrases. In the same way as we can say that noun phrases usually realize the Subject and Object elements, and adverb phrases the Adjunct element, we can say that adjective phrases usually realize both the Complement element and the Modifier element in noun phrases. Halliday emphasizes that nominal groups realize the Subject and the Complement elements. Since adjective phrases can never realize the Subject element, there is no reason to consider it as a sub-type of the nominal group.

   (iii) There are also differences in the morphological potential of the phrases headed by an adjective and those headed by a noun or a verb. Although there are some similarity between adjective and adverb phrases, adjective phrases differ sharply as to the kinds of dependents they take when compared with those of other phrases.

   (iv) The ambiguity of examples such as my first disastrous marriage can only be explained if we accept the existence of the adjective phrase. If we understand that in this example disastrous modifies marriage and first modifies disastrous marriage (implying that the speaker has married more than once), we could perhaps talk of submodification. But, on the other hand, if we understand that disastrous modifies marriage and first also modifies marriage, then we will have to accept the existence of two different phrases in asyndetic coordination modifying the nominal group, since it would be difficult to establish the relation between first and marriage in a different way. This second interpretation of the example could be easily expanded as:

          - My [very first] and [quite disastrous] marriage

   Examples like this clearly speak in favour of the distinction between noun and adjective phrases, but not only when they realize the Complement element, but also when they modify nouns attributively and are not direct constituents of clause structure.

   (v) Adjective phrases can be replaced by some pro-forms. This, although it does not constitute definite evidence, is another point to consider when trying to recognize the constituent character of the adjective phrase:

          - Many people consider John [AdjP extremely rude], but I've never found him so.

          - Brett's work is not yet [AdjP  consistent in style and quality], but will no doubt become so.

   It is obvious that the adjective phrase should be given the same status as the noun, verb or adverb phrases in English and that Halliday's description of the rank group lacks some coherence just because he does not admit this.

 

3.1.2. The predicate phrase 

   As pointed out above, many linguists (specially TG grammarians), who follow the <few-ICs-approach=, recognize the existence of phrases which most commonly divide up not directly into words but rather into further phrases. In an example such as my brother plays football in the mornings, many linguists recognize the existence of two basic phrases which result from the division of the clause into two parts, the Subject and the rest of the predication. But whereas the Subject (my brother) is, if there is no rankshifting, usually realized by only one phrase (the noun phrase), the predication is usually realized by more than one phrase. We can see that the phrase plays football in the mornings is formed of a verb phrase (plays), a noun phrase (football) and a prepositional phrase (in the mornings). Following Allerton, we are going to refer to this phrase as the predicate phrase. The acceptance of this type of phrases immediately implies the recognition of, at least, two different levels within the phrase rank: a higher one, in which phrases divide up into further phrases and a lower one, in which phrases divide up into words.

   Halliday, like all other systemicists, does not feel the necessity of recognizing such a type of phrase. Systemicists follow the <many-ICs-approach=, which implies the possibility of analysing the clause into more than two elements. The clause does not necessarily divide up into two constituents (the noun phrase and the verb (predicate) phrase), but it can directly divide up into more than two elements, as for example in: noun phrase + verb phrase + (noun phrase) + (adverb phrase)[17].

   TG grammarians have presented evidence to prove the existence of this type of phrases, mostly based on syntactic processes such as ellipsis, substitution and the like. This syntactic evidence is very important not only to understand the relationships between clauses and their elements but also to clarify the different relations between phrases themselves. But, despite this evidence, we think that this type of phrases do not have much in common with the units which we have already accepted as such (the noun, adjective, verb and adverb phrases). Two are the reasons which have led us to such a conclusion. In the first place, it is not always easy to see in these structures a relationship of the type Head-Dependents. Although the verb phrase is typically considered as the head of this predicate phrase, in most cases all the other phrases, especially the noun phrase, are as essential and obligatory as the verb phrase itself and therefore it is difficult to talk of a head, if we consider that the head is the only obligatory element in the phrase. Besides, whereas typical phrases can be reduced to only one word, this unit can be very rarely reduced to only one element.

   In the second place, the evidence presented by TG grammarians does not always lead to the same type of unit. Thus, for example, Radford (1988:66ff) speaks of verb phrases (in the <few-ICs-approach= sense) as a phrasal constituent since according to him:

   (a) only phrasal constituents can be conjoined:

                   - He may [go to London] and [visit his mother].

   (b) only phrasal constituents can be substituted by pro-forms:

                   - John might [go home] and so might Bill.

   (c) only phrasal constituents can serve as sentence-fragments:

                   - What would drunks do? Put off the customers.

   However, if we analyse these examples carefully, we will see that the phrasal constituent Radford is presenting us with here is not a unit formed of several phrases (the verb phrase, the noun phrase, the adverb phrase, etc.) but a unit formed of a part of the verb phrase, without modality and tense, and other phrases. This may, perhaps, be evidence to prove that modality and tense should be considered outside the structure of the verb phrase. However, if we accept that this evidence is enough to establish phrasal constituents, we will have to accept that there are other phrasal constituents which are formed of verb phrases in which tense and modality are included, since:

   (a) They can be conjoined:

          - He [has gone to Paris] and [may be there for several weeks]

   (b) They can be substituted by pro-forms:

          - Bob says he is going to join the Labour Party. It will be interesting to see whether he does.

   (c) They can serve as sentence-fragments:

          - What did drunks do?                           Put off the customers.

   And we will even have to consider phrasal constituents units which are formed of a noun phrase plus tense and modality, since they can serve as sentence-fragments:

          - Who will wash the dishes?          Joan will

or units which do not only exclude tense and modality but aspect too, since they can be conjoined:

          - He may have [played with the children] and [watched TV]

          - He may have been [playing with the children] and [watching TV]

   Thus this syntactic evidence is important to understand the relationships between clauses and their elements and also to clarify the different relations between phrases themselves. But it is not enough to establish phrases as constituents of clauses, and therefore we do not think convenient to talk of phrases at this level, at least, in the sense we are giving to this term here. We cannot ignore that in syntactic processes such as ellipsis or substitution, it is not only the morphosyntactic aspects that are important; very often it is the semantic implications that play a decisive role. Because of that we sometimes find not only two or more phrases together, but also parts of phrases alone or in combination with other phrases operating in the clause as if they were single units. The study of what the syntactic possibilities of English are in this respect is very interesting if we are to understand the possible different relations between clauses and phrases, but it is not essential to understand the phrasal structure of the English clause as a whole.

 

3.1.3. The prepositional phrase 

   We saw in the first part of this paper that Halliday considers prepositional phrases as units operating at the group rank, and that in this rank he distinguishes between groups (nominal, verbal and adverbial) and prepositional phrases. We also saw that, although he considers prepositional phrases as a kind of <minor processes=, it is not always easy to assign them verbal value.

   We are now interested, however, in trying to see what the structure of prepositional phrases is and what type of relationship there is between prepositional phrases and the other structures operating at this rank (noun, verb, adjective or adverb phrase) is. Neither Halliday nor other systemicists have properly dealt with these points.

   From the structural point of view, all grammarians agree in pointing out that prepositional phrases are different from all other phrases. While all other phrases are endocentric structures, prepositional phrases are exocentric. Most phrases can be reduced to only one of its elements, which is considered its head: the verb phrase may be reduced to a main verb (may have been working  | work), the noun phrase to a noun (The clever students there |  students), the adverb phrase to an adverb (quite fast |  fast), etc. In contrast, prepositional phrases can never be reduced to only one element (with my

 children  |  * with)[18]. Both the preposition and its complement are obligatory elements which cannot be omitted. Because of this, it is not always easy to establish what the real relationship between these two elements is. Some linguists (cf., for example, Huddleston 1984:331) think that the head of a prepositional phrase is the preposition, even though this goes against the traditional definition of head. Other linguists (cf., for example, Quirk et al. 1985:60) think that prepositional phrases are non-headed phrases. Halliday (1985:189f) seems to agree with this second position when he states that "a prepositional phrase consists of a preposition plus a nominal group" and that "prepositional phrases ... have no logical structure as Head and Modifier, and cannot be reduced to a single element".

   This analysis, however, does not clarify what the syntactic relationship between these two elements is. Whereas the relationship between head and dependents has always been explained in a quite detailed way, not many grammarians explain what the relation between the preposition and the noun phrase is, and therefore we do not really know whether their relation is the same as that of head and dependent, or, if it is different, in what way it differs.

   Some linguists treat prepositional phrases as a kind of embedding structure. Young (1970), for example, thinks that, since "the prepositional object is usually a nominal group ... it looks then, as though all prepositional groups necessarily have a rankshifted unit [a group or a clause] functioning as prepositional object". Young illustrates this in the following way:

                                         Group

                  ┌───────────┴──────────┐    

                Prep.                                             Obj.

                                                                   ( Group )

                 

                                                         ┌─────┴─────┐

                under                                 the                     table

   If we agree with this treatment, we should make clear that prepositional phrases operate at a higher level than all the other phrases included in this rank, since whereas in prepositional phrases we obligatory have a-group-within-a-group, in noun, verb, adjective or adverb phrases we normally have words-within-a-group. Perhaps Halliday's distinction between group and phrase may be taken as an attempt to make this explicit.

   Nevertheless, not all systemicists agree with the idea that in prepositional phrases we have a case of rankshift. Sinclair, as Young himself points out (1972:148), considers this analysis unsatisfactory since "rankshift is normally optional, while here it appears to be obligatory". For Sinclair, there are several important differences between the prepositional group and other cases of rankshift. These differences are basically that (a) the nominal group is essential to the structure of a prepositional group, whereas the rankshifted elements in the other cases are optional and that (b) there are no special restrictions on the nominal group in a prepositional group structure. In all the cases of rankshift, there is some restriction[19].

   There is something very important in this reasoning and which has not been sufficiently underlined in most treatments of prepositional phrases: the nominal group/phrase is essential to the structure of a prepositional phrase. If this is so, why don't we relate prepositional phrases to noun phrases? Although there has been quite a lot of discussion on the problems related with noun and prepositional phrases (think, for example, of descriptive dichotomies such as prepositional object/prepositional phrase - He looked after my father vs. He looked under the bed. - or indirect Object/Adjunct - He gave me the book vs. He gave the book to me), prepositional phrases have been traditionally, and systemicists are not an exception, associated with adverb phrases. This has probably been so because functionally both phrases are very similar, since they usually realize the Adjunct element. However, this functional resemblance should not be a decisive point for the classification and analysis of prepositional phrases. We cannot forget that all the elements of clause structure, with the exception of the Predicator which is always realized by a verb phrase, can be realized by different units. Thus, for example, the Adjunct can be realized not only by adverb and prepositional phrases, but also by noun phrases; the Complement not only by noun phrases and adjective phrases, but also by adverb and prepositional phrases; the Subject by noun phrases and, sometimes, by prepositional phrases too, etc.

   Then it is not function which should matter, but form and structure. And prepositional phrases are identical to noun phrases at least in that they both "have a nominal group which is essential to their structure"[20]. Furthermore, as Sinclair points out, there are no special restrictions on the nominal in a prepositional group structure. These facts can make us think that prepositional phrases could be classified as a subgroup of noun phrases. In fact, we think that we should distinguish two types of noun phrases: (a) unmarked noun phrases: those noun phrases without preposition, as in the toys, the bank, etc. and (b) marked noun phrases: those noun phrases with preposition, as in with the toys, in the bank, etc. Following traditional terminology we could go on calling those which are unmarked noun phrases and those which are marked prepositional phrases. But to call one prepositional phrase and the other noun phrase does not mean that they are not both noun phrases. Prepositional phrases are essentially noun phrases; noun phrases which need a preposition to clarify their semantic relationship with the other elements involved in the process[21].

   There are basically two different ways to signal the relationship between phrases in the English clause: word-order (position of phrases in the clause) and structural markers (prepositions). Noun phrases use both of them to make their relations explicit, but the use of one or the other changes neither its character of noun phrase its semantic relation with all the other elements involved in the process. Compare, for example, these two sentences:

          - The men loaded the truck with hay.

          - The men loaded hay in the truck.

   In these two sentences, we find three noun phrases: the men, the truck and hay. Their semantic relationship is the same in both examples: the men is agent, the truck is location and hay is affected. The relation between them is the same whether or not the noun phrases appear marked with the preposition. And it is the same because, in the cases in which the noun phrases appear without preposition it is their specific position which marks this relationship. While in the first sentence, the relation between <the truck= with regard to the process of loading is marked by position, in the second it is marked by the preposition in. On the other hand, the relation between <hay= with regard to the process of loading is marked by the preposition with in the first sentence, but by position in the second, whereas the relation of <the men= with regard to the process is in both sentences marked by position, Thus, word-order and prepositions are complementary devices to clarify the semantic relationships between phrases.

   In every clause we can distinguish different positions, but these positions have different relevance within the clause. In the English clause we can distinguish (a) some central positions, occupied by those elements, usually obligatory, which are more directly connected with the process and (b) some peripheral positions, occupied by those elements, usually optional, which are more indirectly connected with the process.

   The most central position in the English clause is that of the Predicator, which is the element which specifies the process. Directly connected with the Predicator position, and usually attached to it, we find the Subject position (the position immediately before the Predicator) and the Object position (the position immediately after the Predicator). These positions, as expected, are usually occupied by noun phrases without preposition. Since the elements which occupy these central positions are directly connected with the process, their relation with the process itself is quite clear and, therefore, the elements included in these positions do not normally require any other signal (like prepositions) to express their semantic role. Position is enough.

   The peripheral positions, those which are not directly attached to the Predicator, are usually occupied by Adjunct elements. When these peripheral elements are realized by noun phrases, they normally make use of a preposition to clarify their relation in the process. This is why in the example above the truck and hay go without preposition in the Object position, but with preposition at the moment they are away from this position. Uninflected languages need some structural markers to specify the different relations of all the participants involved in a process. Thus the preposition is basically a marker introduced in the noun phrase to make the semantic relationship with all the other participants (phrases) clearer[22]. In our view, then, prepositional phrases are first of all noun phrases and afterwards prepositional phrases.

   Halliday's analysis of prepositional phrases does not seem to follow these lines. Halliday, for example, insists on the identification of noun phrases with participants and of prepositional phrases with circumstances. By doing so, Halliday is, direct or indirectly, making a sharp distinction between noun and prepositional phrases, on the one hand, and emphasizing the parallelism between adverb phrases and prepositional phrases, on the other. Halliday (1985:142) says: "Typically, the prepositional phrase functions as what we have called a circumstantial element in the clause. Apart from circumstances of quality, which are mostly expressed by adverbial groups, the usual realization of a circumstance is a prepositional phrase. A prepositional phrase contains a nominal group; but while a nominal group represents something which is potentially a participant in the process, a nominal group following a preposition is related to the process only indirectly, with the preposition acting as intermediary".

   But the distinction between participants and circumstances is not as clear-cut as some grammarians usually want us to believe[23] and, therefore, the identification of noun phrase with participant and prepositional phrase with circumstance is definitely difficult to maintain. Obviously, it is easy to find examples in which we find this kind of parallelism. Thus, in an example such as:

          - After dinner, the boys walked along the river.

we clearly have a process (represented by the verb phrase walked), a participant (represented by the noun phrase the boys) and two circumstances, one temporal (represented by the prepositional phrase after dinner) and another spatial (represented by the prepositional phrase along the river). Here we have, then, a perfect exemplification of the dichotomy noun phrases = participants and prepositional phrases = circumstances. However, reality can be seen through many different points of view and what is a circumstance in a process may be a participant in another, and the same kind of process could also have been expressed as:

          - Dinner was followed by a walk along the river.

Here the temporal circumstance is now reflected in the verb followed, the process of walking in the prepositional phrase by a walk and what was a temporal circumstance in the other example (the prepositional phrase after dinner) is now a noun phrase (dinner) which now represents what would have to be understood as a "participant"[24]. Thus, it is quite wrong to identify noun phrases with participants and prepositional phrases with circumstances. Circumstances and participants can be realized both by noun and prepositional phrases, as Halliday finally recognizes[25].

   Nevertheless, if we want to understand all the processes involved in the distinction of participants and circumstances, and their relation to noun and prepositional phrases, we will have to analyse the whole problem in a different way. As we have seen, both participants and circumstances can be realized by noun phrases in English. But the reason why these noun phrases appear unmarked without a preposition or marked with it, will basically depend not on whether they are circumstances or participants but on whether they are direct or indirectly related to the process. When participants or circumstances take part of the process directly, and therefore occupy the central positions of Subject and Object, they do not generally need any preposition to make their semantic role explicit. But when participants or circumstances are related to the process only indirectly, they will not be able to occupy the Subject or the Object position and therefore their semantic role will have to be specified by means of prepositions. Thus, for example, the verb spend may have the "circumstance" time as a direct "participant" of its process. In this case the time expression will occupy the Object position and it will not need any preposition to specify its relation with all the other elements of the process:

          - I spent three days looking for a flat.

          - I spent a week-end in London.

   But if the time expression enters into the process only indirectly, then it will need a preposition to specify its role in the whole process:

          - I spent all my money in three days.

          - I spent all my energies during my week-end in London.[26] 

   It is true, however, that circumstances are normally more indirectly connected with the process than participants, and therefore it is more likely that a circumstance will appear in the sentence marked with a preposition. But this is not the same as identifying prepositional phrases with circumstances and noun phrases with participants. Therefore, the contrast between noun phrases and prepositional phrases is not essentially a question of participant vs. circumstance, but rather a question of direct vs. indirect relation with the process itself. The more distant and less related the participants are, the more necessary the preposition is. This is the reason why the central positions (the Subject and Object positions) are usually realized by unmarked noun phrases, while peripheral positions (the Adjunct positions) are usually realized by marked noun phrases (prepositional phrases).

   There are cases, however, in which we find prepositional phrases in Object and Subject positions and noun phrases in Adjunct positions. This evidence does not, however, go against our argument. The presence, for example, of prepositional phrases in Subject and Object positions is due to the existence of processes in which the relation between the participants is not clear when the noun phrase appears unmarked without a preposition. On many occasions either the verb phrase or the noun phrase is semantically so vague that it is usually difficult to establish what the real relationship between both elements is. The only way to specify their relationship is, of course, by means of a preposition. Compare, for instance, the sentences in these pairs: 

          - He went into the house.                               - He went after the president.

          - He entered the house.                                 - He followed the president.

 

          - He went by the pub.                                  - Our dog went for the postman.

          - He passed the pub on his way home.            - Our dog attacked the postman.

 

          - He went with Mary.                                      - He went over the wall.

          - He accompanied Mary.                                  - He jumped the wall.

 

          - He went across the road.                              - He went up the mountain.

          - He crossed the road.                                    - He climbed the mountain.

 

          - Her dress goes with her eyes.                         - He went to the British Museum.

          - Her dress matches her eyes.                           - He visited the British Museum. 

   The verb go is semantically vaguer than all the other verbs here (enter, follow, pass, attack, accompany, jump, cross, climb, match and visit). Since it is semantically less precise, it needs a preposition to specify the relation between the two "participants". The relation between the participants and the process is quite clear when we use verbs which are semantically less ambiguous and therefore the noun phrase enter directly (without the need of any preposition) into the process. The relation between both elements (the Subject and the Object) is the same in every example, whether the second element is realized by a prepositional or a noun phrase. Here the noun phrases are as "participant" or "circumstance" as the prepositional phrase, and therefore the identification of noun phrase with participant and prepositional phrase with circumstance is definitely wrong.

   The same kind of phenomenon occurs when we occasionally find noun phrases marked with prepositions in the Subject position, as in:

          - Will after the show be enough?

          - Between 2 and 3 will be perfect.

   In these examples, we need noun phrases marked with prepositions, because the relation between the elements in the clause is not clarified with position only. (cf. ?? will the show be enough ?, ?? 2 and 3 will be perfect). However, when the relation between the elements is not ambiguous, as in:

          - On Tuesday will be fine. (Quirk et al., 1985:658)

          - In the lake will be splendid. (Quirk et al., 1985:658)

          - To York is not far. (Quirk et al., 1985:749)

then, the prepositions will usually be omitted:

          - Tuesday will be fine.

          - The lake will be splendid.

          - York is not far.

These latter examples are much more frequent than the former; the prepositions are here redundant, since position already clarifies the relationship between all the elements.

   But, as Quirk et al. (1985:658) point out, even in this position the lack of preposition may,  theoretically, create linguistic ambiguity whenever the temporal or local reference is not clearly expressed in the context or in the verb. Thus, examples like:

          - Tuesday will be fine.

          - The lake will be splendid.

may mean either:

          - Meeting on Tuesday will be fine

          - Bathing in the lake will be splendid.

or:

          - The weather on Tuesday will be fine.

          - The scenery by the lake will be splendid[27].

   The opposite is also true. We may have noun phrases as circumstances without prepositions, no matter what position (peripheral or central) they occupy in the clause, whenever their definition as circumstance is unambiguously expressed, as in:

          - She bought me a tape-recorder last week.

          - She writes an article or a review for the Times every month.

          - Would you like to stay with us in a cottage a few days?

   The temporal connotations are so clearly implied in these noun phrases that there is no need of any preposition to clarify what their role within the processes in which they are involved is.

   Sometimes, as we saw before (5), there are many occasions in which the noun phrases can appear with and without preposition:

          - I'd like to send this parcel (by) air mail.

          - They play the game (in) a different way.

          - She cooks chicken (in) the way I like.

          - A hat (of) this size will be fine.

          - He takes risks (in) every lecture.

The dropping of the preposition is again possible because the relation between the elements implied in all these processes are very clear.

   All this evidence clearly shows that noun and prepositional phrases are essentially the same kind of structure, and it is more for semantic than for syntactic reasons that noun phrases are marked with prepositions. Prepositions are, therefore, elements whose function is that of specifying the different relationships that there are between a noun phrase and other phrases in the clause, whenever this relationship is not clarified by other means.

   Everything we have already said about the relation between noun and prepositional phrases is closely connected with the alternative some modern approaches to English grammar have taken to describe prepositions and which is to regard them as a set of cases for noun phrases. This approach was institutionalized in Fillmore's case grammar (Fillmore, 1968). Whereas inflected languages have a choice of several cases for nominal groups, uninflected ones have to resort to prepositions. Allerton refers to case in more or less the same way as we have done when discussing prepositions. Allerton (1979:246) says, for example, that "case ... refers to a variety of relationships that nouns contract to verbs, prepositions or the sentence as a whole. The case may be determined by the character of the verb or preposition or by syntactic function". This close relationship has been illustrated with examples in which prepositions and case inflections perform very similar semantic functions. But, as we have already suggested, prepositions not only overlap with case markers, but also with word-order. English allows modification by position; when we place a word before another, we are usually stating a semantic relationship between the two words which can be paraphrased as "a is modified by b". And this semantic relation is similar to that specified by prepositions. Radford (1988:204f), for example, points out that "the parallel between postnominal prepositional phrases and prenominal noun phrases is a very close one". Compare these pairs of examples given by Radford: 

          (a) - the ban on pornography           (a) - recruitment of personnel

          (b) - the pornography ban              (b) - personnel recruitment

 

          (a) - the appeal for charity           (a) - relief from famine

          (b) - the charity appeal               (b) - famine relief

 

          (a) - damage to the brain              (a) - the investigations into fraud

          (b) - brain damage                     (b) - the fraud investigations

 

          (a) - the strike in the shipyard       (a) - the bridge over the river

          (b) - the shipyard strike              (b) - the river bridge 

   The only difference, apart from stylistic ones, between examples in (a) and those in (b) is that the prenominal modification is more ambiguous than the postnominal one. Radford (1988:206) points out that the semantic relation between a prenominal noun phrase and the noun it modifies is much vaguer (and has to be inferred from pragmatic clues) than in the case of a postnominal prepositional phrase. In an example like the [proportional representation] campaign, the semantic relation between both noun phrases is ambiguous and it is by means of prepositions that we can specify it either as (a) the campaign for proportional representation or as (b) the campaign against proportional representation.

   Therefore prepositions are the functional elements which express in a clearer way the relations existing between noun phrases and other phrases in English, and, since this is the only function of prepositions, prepositional phrases should be considered as a subclass within noun phrases. As Jackendoff (1977:17) puts it: "it seems evident that noun phrases and prepositional phrases form a natural phrasal supercategory". The preposition does not change the structure of the noun phrase into a new different "prepositional" structure; the preposition does not alter the structure of the noun phrase, which is in every way the same in every case, whether or not the noun phrase is marked with a preposition. It is, therefore, a mistake to consider prepositional phrases and noun phrases as two unrelated units, as has traditionally been done by most linguists[28]

   So far we have been discussing the nature of prepositional phrases and we have suggested that they should be considered as a subtype of noun phrases. But up to now we have said nothing to clarify what the specific relation between the preposition and the noun phrase it goes with is and how we are going to deal with prepositions in our analysis of phrases. The analysis of prepositions we are going to propose now supports our idea that prepositional phrases are essentially noun phrases. We think that the function of the preposition in phrase structure is similar to that of conjunctions in clause structure.

   Many transformational approaches[29] study subordinating conjunctions as prepositions. For them, the only difference between prepositions and conjunctions is that they have different complements. Whereas prepositions have as their object a noun phrase, conjunctions have a clausal complement, as can be seen in the following examples taken from Radford(1988):

          - There has been no trouble since your departure | There has been no trouble since you left.

          - I'd never met her before the party | I'd never met her before I saw her in the party.

          - You can have some chocolate after your dinner | You can have some chocolate after you've eaten your dinner.

          - Matilda was much envied for her talent | Matilda was much envied for she was very talented.

          - I shall wait until your return | I shall wait until you come back.

   Grammarians in favour of this approach claim that, since this is their only difference, there is no point in establishing two different categories. They argue that prepositions, like many verbs, nouns and adjectives, allow as object either a noun phrase or a subordinate clause. Just as know is a verb whether it is complemented with a noun phrase (I don't know the answer) or a clause (I know you are innocent), in the same way a preposition is a preposition whether it is complemented with a noun phrase (before dinner) or a clause (before they've had dinner)[30]. Whether or not they belong to the same category, conjunctions and prepositions have a very similar function. Conjunctions traditionally realize two different syntactic elements: coordinators and subordinators. Coordinators "have no function as Subject, Adjunct or Complement" (Halliday, 1985:51) and are elements which are outside the structure of the clause[31]. They are, in Halliday's words, <floating= elements. Since they are outside the clause structure, they are usually represented in the following way:

          - [It was Christmas day] and [the snow lay thick on the ground].

   However, we cannot always say the same of subordinators, because we have to distinguish two types: those which can function as S, O, C or A, and those which, like coordinators, do not have any function in clause structure. Among the former, the most typical subordinators are the so-called wh-elements:

          - [ [ Whatever (O) I (S) say (P) ] ] (A) don't believe (P) me (O).

          - [ [ Wherever (A) I (S) go (P) ] ] (A) I (S)  carry (P) an umbrella (O).

   There are, nevertheless, many subordinators which do not have any function in the clause they are attached to besides that of linking. In examples like:

          - [ I ] (S) [ couldn't see ] (P) [ because it had got dark ] (A).

          - [ You ] (S) [ could fly ] (P) [ if you were a bird ] (A).

the clauses because it got dark and if you were a bird play a function in the structure of the clause complex, since we could consider them as an expansion of:

          - [ I ] (S) [ couldn't see ] (P) [ for this reason ] (A).

          - [ You ] (S) [ could fly ] (P) [ on this condition ] (A).

but if we analyse the embedded clauses, it would be difficult to assign any function to the subordinators because and if besides that of linking, since they do not represent any participant or circumstance within the clause. The only possible analysis of these embedded clauses is the following:

          - because (?) it (S)  had got (P)  dark (Cs).

          - if (?)  you  (S)  were  (P)  a bird (Cs).

   Since these subordinators do not function as S, O, C or A, we should analyse them in the same way as we have analysed coordinators, that is to say, we should consider them as elements which are outside clause structure; they are also <floating= elements. Thus the complete analysis of these examples should be:

          - I (S) couldn't see (P) [ because [ it (S) had got (P) dark (Cs) ] ]  (A).

          - You (S) could fly (P) [ if [ you (S) were (P) a bird (Cs) ] ]  (A).

   The function of these subordinators, as Halliday (1985:289) points out, is that "of setting up the logical relations that characterize clause complexes in the absence of the structural relationships by which such complexes are defined".

   It should be clear by now that coordinators and many subordinators - basically those with no syntactic function in clause structure - behave in a very similar way. Therefore we could establish a clear parallelism between them, since they share the same function: that of specifying what the semantic relationship between the two clauses is. In the same way that coordinators are considered to be outside clause structure, we can consider that these subordinators are also outside clause structure. Whether floating between two clauses or attached to one of them, the conjunction functions as a relator, as a linking element.

   Prepositions function in the same way as conjunctions but at a different level. Whereas the conjunction is a linking element which specifies the different semantic relationships of clauses, the preposition is a linking element which specifies the different semantic relationships of phrases. Or if we agree with the transformationalists' idea of a supercategory of prepositions which includes both prepositions and conjunctions, we could just say that "prepositions" are linking elements which specify the different semantic relationships between two elements, which may be either clauses or phrases.

   If we apply Halliday's words to prepositions, we will clearly see what the real function of prepositions is: that of setting up the logical relations that characterize phrases and phrase complexes in the absence of the structural relationships by which phrases are defined. Thus, our analysis of prepositions must be the same as the one we have done with coordinators or subordinators. Prepositions are outside phrase structure and therefore they do not function either as Heads or as Dependents. When coordinators link phrases they are always analysed in the same way as when they link clauses. In an example like:

          - [You] (S) [could take] (P) [ [the blue pencil] or [the red pen] ] (O).

the phrasal coordinator or does not have any function in phrase structure, and therefore we analyse it as an element outside phrase structure. The analysis at phrase structure level could be:

          - [the (d) blue (d) pencil (h)] or (?) [the (d) red (d) pen (h) ]

   The only function of the phrasal coordinator is to specify the relationship between the two noun phrases, which in this example is optionality between two possibilities. In the same way, when we analyse sentences like:

          - [You] (S) [could take] (P) [the blue pencil of my brother] (O).

          - [You] (S) [could take] (P) [the big picture on the wall] (O).

we can see that the phrasal subordinators, the prepositions of and on, cannot be assigned any function in phrase structure:

          - [You (h)] [could (d) take (h)] [the (d) blue (d) pencil (h) ] of (?)  [my  (d) brother (h)]

          - [You (h)] [could (d) take (h)] [the (d) big (d) picture (h) ] on (?)  [the (d) wall (h)]

   The phrasal subordinators of and on do not function as dependents or as heads of any phrase; just as the coordinator or they are outside phrase structure. Their only function is to specify what the semantic relationship  is between the head noun and the complement noun phrase modifying it. Of expresses that the relation between blue pencil and my brother is that of possession and on expresses that the relation between the big picture and the wall is that of location. Hence, prepositions and conjunctions are words which are outside phrase and clause structure: conjunctions are outside clause structure, since their function is that of relating clauses; prepositions are outside phrase structure, since their function is that of relating phrases.

   If we accept this sort of analysis, we will have to conclude that prepositional phrases do not really constitute phrasal structures, since the preposition, which is their only identifying element, falls outside this phrasal structure. At this level of phrase, then, we have only four different types of structures in English, which are noun phrases, verb phrases, adjective phrases and adverb phrases.

  

3.2. PHRASES NOT OPERATING AT CLAUSE STRUCTURE 

   To finish this revision of phrase, we will have to analyse briefly those phrases which cannot be considered as being constituents of the clause. This type of phrases appear at two different levels: they either form part of phrase structure or appear forming part of elements which are outside phrasal and clausal structure. As we said at the beginning, Halliday mentions two structures which can be included here, the preposition group and the conjunction group (see 3.2.2.), but he does not specify at what level they operate and, in fact, he studies them together with the adverbial group, which, as we saw before, operates at quite a different level.

 

3.2.1. Phrases operating at phrase structure 

   If we analyse in a more detailed way the structure of noun, adjective and adverb phrases, we can see that they are not really formed only out of words in all cases. Within these phrases we find many examples of "groups of words" which can be described as being in a syntactic relation of Head and Dependents. When we analyse an example such as:

          - Rather too many  students have failed the exams.

we have to distinguish three different phrases: a noun phrase (rather too many students), a verb phrase (have failed) and a noun phrase (the exams). But, while the two last phrases divide up into words, the first noun phrase does not. We can see that there is a kind of phrasal structure in the dependent element of this noun phrase, where many can be considered as head and rather too as dependents. We could also talk of phrasal structure in this dependent element, with too as head and rather as dependent. We cannot talk here of embedding or rankshift since these structures can never appear realizing any constituent element in the clause. We have to consider them as phrases-within-phrases, since this the only level at which they can operate.

   Much work on this area has lately been done by transformational grammarians. Jackendoff (1977), for example, talks of quantifier phrases, measure phrases and degree phrases. According to Jackendoff, adjective phrases, for example, can be modified by a quantifier which can itself be modified by a degree element, as in:

          - This fence is so / too much higher than that one.

          - He is so much more an interesting player.

   In the same way, prepositional phrases can be modified by a quantifier which can itself be modified by a degree element, which may be in turn modified by a noun phrase often referred to as measure phrase, as in:

          -  five times that far along the road

   This suggests that degree elements can themselves be expanded into a phrase containing a degree element and either a quantifier or a noun phrase. Muir (1972), on the other hand, analyses the type of words which can modify deictics and give examples such as:

          - very nearly the best house

          - almost all the other boys

In the first example we can see a degree phrase modifying the superlative adjective and in the degree phrase we can see the intensifier very modifying the degree adverb nearly. In the second example, we can see that the predeterminer all is being modified by almost, in what seems a clear phrasal structure. Although predeterminers, determiners and postdeterminers have lately very often been object of study, their phrasal structure will have to be studied more carefully and the connexion between them more accurately clarified.

   It is not easy to decide how many types of these phrases we should distinguish at this level to have an accurate description of English at this level. The discussion of all the problems involved in the description of this level deserves more investigation and more space. Thus, the only thing we can do here is to verify their existence and hope some new research will be carried out in this area.

 

 3.2.2. Phrases outside phrasal or clausal structure 

   Halliday (1985) recognizes the preposition and the conjunction groups. It is difficult to consider these structures as forming constituent elements of the clause. Therefore, as with the phrases above, the only reason to consider them as groups/phrases is because they could be understood as combinations of words with the specific structure of Head + Dependents. If we analyse an example such as:

          - [ He ] [ went ] [ right (to the end) ]

we can distinguish three constituents: the Subject (He), the Predicator (went) and the Adjunct (right to the end). The constituents are realized by a noun phrase, a verb phrase and a prepositional (or marked noun) phrase. If we analyse the Adjunct constituent, we realize that the prepositional phrase is modified by the optional intensifier right, which functions in a similar way to other modifiers in other phrases. Compare, for example:

          - He was very glad.

          - He ran quite fast.

          - It went right into the wall.

   Just as very and quite are considered to be dependent elements modifying the heads glad and fast in the adjective and adverb phrases, Halliday considers right a dependent element modifying the preposition to, forming what he calls the preposition group.

   Similarly, if we analyse an example like:

          - [ I ] [ did ] [ it ] [ just (because they asked me].

we can distinguish four constituents: the Subject (I), the Predicator (did), the Object (it) and the Adjunct (just because they asked me). If we analyse the Adjunct constituent, we can distinguish three constituents in the clause which is realizing it: the Subject they, realized by a noun phrase, the Predicator asked, realized by a verb phrase and the Object me, realized by a noun phrase. Besides we find the linker because, which seems to be modified by the optional element just. According to Halliday, just has a similar function as that of very, quite or right in the above examples, and this is why he talks of conjunction groups.

   We are not going to discuss now if right modifies just the preposition or the Adjunct constituent as a whole, or if just modifies the conjunction or the whole Adjunct constituent[32]. The only thing we are now interested in is in pointing out that, if in these combinations of words there is a Head-and-Dependent relationship, then these phrasal units are outside phrasal structure, since the words which are supposed to act as heads, the preposition and the conjunction, are, as has been shown above, outside phrasal structure themselves. Therefore, if we are to consider them as groups/phrases, like Halliday does, it should be clear that they operate at a very different level to the one at which the four basic structures of noun, verb, adjective and adverb phrases operate.

 

3.2.3. Words at clause structure 

   We do not want to finish this revision of the phrase rank without mentioning another aspect of the way English operates at this rank, but which poses the opposite problem. Just as we have phrases which do not operate at clause rank, there are also words which operate directly at the clause rank, but which hardly fit our concept of phrases. These words, although they are usually considered to be constituents of the clause in which they are, can never be expanded to form groups of more than one word and, therefore, it is quite contradictory to call a word which can never be combined with any other word a phrase. Thus, for instance, it does not seem very logical to refer to words like already or therefore as phrases in examples such as:

          - He had already left.

          - Therefore we did not like it.

when we know that they do not accept any type of modification. To say that therefore and already are the head of an adverb phrase seems a bit artificial and as Huddleston (1984:116) suggests this may be "analysing the language in a way that makes it out to have a more uniform, neat and regular organisation that in fact it has". This, of course, does not mean that we cannot consider as phrases items which consist of a single word, like, for instance, children in the example:

          - Children were playing in the park.

   Whereas therefore or already do not allow expansion, children can be expanded into a larger construction (cf. all the naughty children who hated me were playing in the park).

  

4. CONCLUSIONS 

   All the evidence presented here, some of it in a somewhat precipitate way, should serve to prove at least the following points:

   (a) That the phrase is not an easy unit to delimit.

   (b) That the idea that clauses consist of phrases, which consist of words is only an idealized conception of the way the English clause is organized.

   (c) That the syntactic possibilities of words are wider and of a more diverse nature than it is usually recognized.

   (d) That if the only requirement for phrases is that of a combination of words with a relation of head/dependents, the unit phrase, as Allerton (1979:205) points out, does not constitute a single point on the rank scale. There are, at least, three different levels at which phrases may occur:

      (i) some phrases (noun, verb, adjective and adverb phrases) realize clausal elements. At this level, we should not talk of prepositional phrases. Prepositional phrases do not really constitute phrasal structures, since the preposition, which is their only identifying element, falls outside this phrasal structure.

      (ii) other phrases (quantifier, measure and degree phrases, for example) only realize phrasal elements.

      (iii) other phrases (conjunction and preposition groups) may even be considered to operate outside clausal and phrasal structure.

   (e) That words can enter into clause structure directly.

 

 REFERENCES 

- Allerton, D.J. (1979), Essentials of grammatical theory. London: Routledge & Kegan Paul.

- Berry, M. (1975), Introduction to systemic linguistics: 1, structures and systems. London: Batsford.

- Berry, M. (1977), Introduction to systemic linguistics: 2, levels and links. London: Batsford.

- Bowers, J.S. (1975), "Adjectives and adverbs in English", Foundations of Language 13, 529-562.

- Butler, C.S. (1985), Systemic linguistics: theory and applications. London: Batsford.

- Emonds, J.E. (1976), A transformational approach to English syntax. New York: Academic Press.

- Fawcett, R.P. (1974), "Some proposals for systemic syntax, Part 1", Midlands Association for Linguistic Study Journal 1, 1-15.

- Fawcett, R.P. (1975), "Some proposals for systemic syntax, Part 2", Midlands Association for Linguistic Study Journal 2, 1, 43-68.

- Fawcett, R.P. (1976), "Some proposals for systemic syntax, Part 3", Midlands Association for Linguistic Study Journal 2, 2, 35-68.

- Fawcett, R.P. (1981), Some proposal for systemic syntax: an iconoclastic approach to scale and category grammar. Revised version with light amendments of the three articles appeared in Midlands Association for Linguistic Study Journal. Cardiff: Department of Behavioural and Communication Studies, The Polytechnic of Wales.

- Fillmore, C.J. (1968), "The case for the case", in E. Bach & R.T. Harms (eds.), Universals in linguistic theory, New York: Holt, Rinehart and Winston, 1-88.

- Halliday, M.A.K. (1956), "Grammatical categories in modern Chinese", Transactions of the Philological Society, 177-224.

- Halliday, (1960), "General linguistics and its application to language teaching" in A. McIntosh & M.A.K. Halliday (1966), Patterns of language: papers in general, descriptive and applied linguistics. London: Longman, 1-41.

- Halliday, M.A.K. (1961),"Categories of the theory of grammar", Word, 17, 241-92. Reprinted in part in Kress (1976), 52-72.

- Halliday, M.A.K. (1963), "Class in relation to the axes of chain and choice in language", Linguistics 2, 5-15. Reprinted in part in Kress (1976), 84-87.

- Halliday, M.A.K. (1966a), "General linguistics and its application to language teaching", in McIntosh & Halliday (1966), 1-41.

- Halliday, M.A.K. (1966b), "Some notes on "deep" grammar", Journal of Linguistics 2, 57-67. Reprinted in part in Kress (1967), 88-98.

- Halliday, M.A.K. (1967a), "Notes on transitivity and theme in English, Part 1", Journal of Linguistics 3, 37-81.

- Halliday, M.A.K. (1967b), "Notes on transitivity and theme in English, Part 2", Journal of Linguistics 3, 199-244.

- Halliday, M.A.K. (1968), "Notes on transitivity and theme in English, Part 3", Journal of Linguistics 4, 179-215.

- Halliday, M.A.K. (1973), Explorations in the functions of language. London: Edward Arnold.

- Halliday, M.A.K. (1977), "Text as semantic choice in social contexts", in T.A. van Dijk & J. Petöfi (eds.), Grammars and descriptions. Berlin: De Gruyter, 176-225.

- Halliday, M.A.K. (1978), Language as social semiotic: the social interpretation of language and meaning. London: Edward Arnold.

- Halliday, M.A.K. (1985), An introduction to functional grammar. London: Edward Arnold.

- Halliday, M.A.K. & J.R. Martin (eds.) (1981), Readings in systemic linguistics. London: Batsford.

- Halliday, M.A.K., A. McIntosh & P. Strevens (1964), The linguistic sciences and language teaching. London: Longman.

- Huddleston, R.D. (1965), "Rank and depth", Language, 41, 574-586.

- Huddleston, R.D. (1984), Introduction to the grammar of English. Cambridge: Cambridge University Press.

- Hudson, R.A. (1971), English complex sentences: an introduction to systemic grammar. Amsterdam: North-Holland.

- Jackendoff, R.S. (1977), X syntax: a study of phrase structure. Cambridge, (Mass): The M.I.T. Press.

- Jaworska, E. (1986), "Prepositional phrases as subjects and objects", Journal of Linguistics, 22, 355-374.

- Jespersen, O. (1924), The philosophy of grammar. London: George Allen & Unwin Ltd.

- Kress, G. (ed.) (1976), Halliday: system and function in language. Oxford University Press.

- Lyons, J. (1968), An introduction to theoretical linguistics. Cambridge: Cambridge University Press.

- Lyons, J. (1977), Semantics. 2 vols. Cambridge: Cambridge University Press.

- Matthews, P.H. (1966), "The concept of rank in Neo-Firthian linguistics", Journal of Linguistics 2.1, 101-109.

- Matthews, P.H. (1981), Syntax. Cambridge: Cambridge University Press.

- McIntosh, A. & M.A.K. Halliday (1966), Patterns of language: papers in general, descriptive and applied linguistics. London: Longman.

- Muir, J. (1972), A modern approach to English grammar: an introduction to systemic grammar. London: Batsford.

- Quirk, R. et. al. (1985), A comprehensive grammar of the English language. London: Longman.

- Radford, A. (1988), Transformational grammar : a first course. Cambridge: Cambridge University Press.

- Scott, F.S., C.C. Bowley, C.S. Brockett, J.G. Brown & P.R. Goddard (1968), English grammar: a linguistic study of its classes and structures. London: Heinemman Educational Books.

- Sinclair, J. McH. (1972), A course in spoken English: grammar. Oxford University Press.

- Young, D. (1980), The structure of English clauses. London: Hutchinson.

- Zandvoort, R.W. (1969), A handbook of English grammar. London: Longman (first edition 1957).


     [1]. Along this paper the terms phrase and group are going to be used, more or less indistinctly, to refer to the same kind of structure. Although we do not think it necessary or convenient to keep both terms and we will defend in this paper that the term phrase is the most appropriate, we will have to keep the Hallidayan distinction group/phrase whenever we make reference to any linguist who uses these terms. Thus we will talk of phrase whenever we refer to this type of structure in a more general way or when we present our own points of view, but talk of group whenever we are presenting the point of view of those linguists who defend this term. Sometimes, to avoid any misunderstanding, we refer to this unit with both terms, as group/phrase. 

     [2]. This is indirectly recognized by Halliday (1985:192) when he says: "A group ... could be interpreted as a WORD COMPLEX: that is to say, a Head word together with other words that modify it. This is why the term GROUP came to be used. It meant <group of words=, or <word group=; and it suggests how the group evolved, by expansion outwards from the word". 

     [3]. See, for example, pp 5-6, where Zandvoort refers to expressions like go hang, make believe as phrases, but as "similar groups" to expressions like hear say or hear tell. See also p 159. 

     [4]. Zandvoort refers as groups, structures such as arriving at the station (p 35), to stop (p 46), to be taken advantage of (p 54), after a day or two's (p 91), to get a taxi (p 134), to think fit (p135), now and then (p 170), makes me sleepy (p 204), granting that (p 220), to be or not to be (p 227), not ... but (p 234), good-for-nothing (p 278), up-to-date (p 282), or out of (p 286). If we compare these examples, it is quite difficult to grasp what all these combinations have in common to be included within the same category, besides the fact that they are formed with more than one word. 

     [5]. As is well-known, the first PS-rule in every traditional transformational grammar stated that the sentence was formed of two immediate constituents: the noun phrase and the verb phrase ( S |   NP + VP). The verb phrase here most commonly divides up not directly into words but rather into phrases, of which one is the verb phrase itself, but the other can be a noun phrase, an adjective phrase, a prepositional phrase or a combination of some of these phrases (cf. [ He (NP) ]  [  ( bought (VP) )  ( a book  (NP) )  ( for Mary (PP)) ( there (AdvP) ) ]  (VP)). Trying to avoid such an ambiguity, systemic grammar may have introduced the term group. 

     [6]. There is one case in which the prepositional phrase can be reduced to only one element: when the preposition in the prepositional phrase is a word that can function both as a preposition and as an adverb, as in:

                   - He was in the restaurant  | He was in.

                   - He went through the park  | He went through.

                   - He ran across the river   | He ran across.

                   - He went down the mountain | He went down.

          But there is a great difference between the reduction of these prepositional phrases and that of other phrases. The reduction of any other type of phrase does not involve a change in the status of the structure in question. The reduction of a verb, noun or adverb phrase to a phrase of only one word does never affect the character of the phrase, which will continue being a verb, a noun or an adverbial phrase. In contrast, when we reduce these prepositional phrases, what we get is not another prepositional phrase but an adverb one, whose head is now an adverb. 

     [7]. Halliday (1985:150) refers to these examples in the following way: "Just as those elements which are treated essentially as circumstances can sometimes occur with a preposition, so at least some elements which are treated essentially as circumstances can sometimes occur without one, as in they stayed two days, they left Wednesday".  

     [8]. Halliday is aware of this when he states: "Semantically, therefore, Agent, Beneficiary and Range have some features of participants and some of circumstances: they are mixed. And this is reflected in the fact that grammatically also they are mixed: they may enter into the clause either directly as nominal groups (Participant-like) or indirectly as prepositional phrases (Circumstance-like)" (1985:149). 

     [9]. Other examples of more difficult classification taken from Halliday (1985) and which cannot be given verbal value easily are:

                   - I want to talk to them face to face rather than on telephone.

                   - Why can't they arrive on time instead of two hours late.

                   - It's upstairs to the left of the landing in the main bedroom against the far wall in the top drawer at the back right hand corner.

                   - Alice in Wonderland.

                   - Nibling first at one and then at the other.

                   - We are aware of our responsability to our critics.

                   - What have the elephants done to the pier ? 

     [10]. Scott et al., Butler, Berry, and Sinclair consider, like Halliday, that the four basic groups in English clause structure are the nominal, verbal, adverbial and prepositional groups. Despite that, they, in contrast to Halliday, study the adjective group as a kind of sub-group. Their analyses are, nevertheless, different. Whereas Sinclair includes the adjectival group as a sub-group of the adverbial group, Scott et al., Butler, and Berry include it as a sub-group of the nominal. 

     [11]. See Halliday (1960:10), Halliday (1961:256-257) or Halliday (1963:86). 

     [12]. Halliday considered prepositional phrases as a subgroup of the adverbial group in the 1961 model, but later on (see 1977:177 or 1978:129), he recognized a class of prepositional phrase, different from the adverbial group. In 1985 Halliday has made the distinction clearer, and adverbial groups and prepositional phrases are considered as two different units. 

     [13]. Paradoxically, in the commentary provided on the papers in section 1 of Halliday & Martin (1981:19) we find the following comment: "Another factor involved here relates to Halliday's attempt in 1961 to define classes in terms of the potential occurrence of items at certain elements of structure in immediately higher ranking units. Throughout the sixties classes came increasingly to be formulated in terms of their distribution and constitution, and in terms of meaning, enabling a more elaborate formalization of classes into systems and systems into networks". Despite this commentary, Halliday keeps on classifying the adjective in the nominal group not only when it functions attributively (The old man) but also when it functions predicatively (the man is old). 

     [14].  Halliday (1967a:41) argues that "it may seem a little unusual to consider happy in she looked happy and the clothes in she washed the clothes as realizing the same primary element of structure, the Complement. But there are certain generalizations to be made concerning both, for example the proportion she washed the clothes : the clothes she washed (but not the sheets)  ::  she looked happy : happy she looked (but not excited); moreover ... certain elements usually considered as noun predicates (intensive Complement) may be better regarded as <direct Object= (extensive Complement). It is noteworthy that she  in she washed the clothes and she in she looked happy likewise express different semantic notions, and yet there seems no adequate reason for denying that they both realize the element "Subject". Halliday, of course, says nothing of examples such as she went to the station : to the station she went (but not to the airport). 

     [15]. Halliday (1985:252) says that "group and phrases form complexes in the same way that clauses do, by parataxis or hypotaxis. Only elements having the same function can be linked in this way. Typically this will mean members of the same class: verbal group with verbal group, nominal group with nominal group and so on. But it also includes other combinations, especially: adverbial group with prepositional phrase, since these share many of the same circumstantial functions in the clause; and nominal group with prepositional phrase, as Attribute (e.g. plain or with cream)". It is difficult to think that very and big can, in any way, be considered members of the same class. 

     [16]. Sinclair (1972:160-61) not only talks of sub-modification, but also of sub-qualification. We think it would be much easier to accept the existence of the adjective group/phrase. 

     [17]. See Hudson (1972:21ff) for an explanation of the basic reasons of systemic linguistics to reject the <few-ICs-approach= of transformational grammar in favour of the <many-ICs-approach=. 

     [18]. See, nevertheless, note 6 above. 

     [19]. Although we think that the reasons Sinclair gives to reject the idea of rankshift are really convincing, we do not agree with his analysis of prepositional phrases. Sinclair (1972:148) thinks "that a prepositional group is a compounding of an adverbial and a nominal group without either of them being rankshifted". For Sinclair (1972:209), the adverbial group has the structure (m) h, and the headword is called a preposition. The nominal group has unrestricted structure, and is usually called an object, to echo the term object in clause structure. It is useful to think of prepositions as <transitive adverbs=, requiring a nominal to complete the structure". However, Sinclair does not explain what he means by "a compounding of groups" and therefore we do not know what the relationship between the adverbial and the nominal group is. He does not explain either in what way prepositions (apart from their possibility of being modified) resemble adverbial groups to be classified as such. 

     [20]. For prepositional phrases with the structure <preposition= + <adjective or adverb phrase=, see note 28 below. 

     [21].  It should be clear from the very beginning that, by presenting this classification, we are not attempting to get rid of prepositional phrases as a structure in the English clause. It would be very difficult to understand a description of English which does not recognize the existence of such a construction (at surface structure at least). Prepositional phrases do not only differ from noun phrases in their basic functions, but they also differ in their structure. Prepositional phrases, for example, may be modified by degree elements such as right (cf. It was right at the bottom) and noun phrases cannot. We are putting prepositional and noun phrases together because we think that by doing so, we will improve our understanding of the relationship which exists between the preposition and the noun phrase to which it is attached. Thus, our only attempt is to try to interconnect both structures, since in our opinion prepositional phrases are closer to noun phrases than to any other type of phrases in clause structure. We definitely do not agree with the treatment of prepositional phrases as a subtype of adverb phrases or as a structure of its own, unrelated to any of the other phrases in the clause. 

     [22]. To say that prepositions are phrase markers does not mean that prepositions do not have <lexical= meaning. On the contrary, since the function of prepositions is to make the relationship between the noun phrase and the other phrases more explicit, they need to have meaning to fulfill their function successfully. To clarify, for example, the relation between the book and the table, the prepositions on, down, near, behind, etc. need to have meaning, that which specifies the position of the book in relation to the table. 

     [23]. Halliday, for example, presents the configuration of the English clause as one in which we find a clear parallelism between verbal groups and processes, nominal groups and participants, and adverbial groups and circumstances: "The processes of the external world ... are encoded in English ... into three types of structural element: the process itself, the participants in the process (animate and inanimate), and the attendant circumstances. This is the basis of the very widespread three-way distinction of constituents of the clause into verbals, nominals and the rest (adverbials, in the broader sense)" (Halliday, "On functional grammars", in Kress, 1976:20-21). 

     [24]. Perhaps it is not out of place to mention here the example of Jespersen in which participants, processes and circumstances are indistinctly expressed by verb, noun, adjective, adverb or prepositional phrases: He moved astonishingly fast, He moved with astonishing rapidity,  His movements were astonishingly rapid, His rapid movements astonished us, the rapidity of his movements was astonishing, etc. (Jespersen, 1924:91). 

     [25]. In spite of his defence of the general principle by which  noun phrases realize participants and prepositional phrases circumstances, Halliday has to accept that not only participants can be realized by prepositional phrases (see quotation in note 8, above), but also that circumstances are sometimes realized by noun phrases. Halliday (1985:150) states: "Just as those elements which are treated essentially as participants can sometimes occur with a preposition, so at least some elements which are treated essentially as circumstances can sometimes occur without one, as in they stayed two days, they left Wednesday".

          Along Halliday's An introduction to functional grammar we have found many examples of participants expressed by prepositional phrases. Some of them are:

                   - The bridge was built by the army.

                   - Speak roughly to your little boy.

                   - She baked a pie for the children. 

     [26]. Cf. also other examples such as: he went that way, which direction did she run ?, they run two miles, come this way please, we climbed a further thousand feet. 

     [27]. We cannot forget that prepositional phrases may even be marked with another preposition. This will happen when we want to specify the relationship already marked with the first preposition more clearly, as in:

                   - He picked up the gun from under the table.

                   - We didn't meet until after the show.

                   - Food has been scarce since before the war. 

     [28].  Perhaps someone may argue that prepositional phrases exceptionally allow other types of complements. Adjective phrases and adverb phrases may occasionally function as complements of a preposition, as can be seen in the following examples, some of them taken from Huddleston (1984:345) and Quirk et al. (1985:658):

    (a) Adjective phrases: [ He regarded it] as inefficient, [He took her] for dead, [It went] from bad to worse, [He was dressed] in black, for better or for worse, in brief, etc.

          (b) Adverb phrases: from here, since then, at once, before long, until now, in there, by far, etc.

          We think, however, that this is not against our treatment of prepositional phrases as a subtype of noun phrases. On the contrary, the fact that there are adjective and adverb phrases specified by prepositions may only mean that we should have to accept the existence of marked adjective phrases as well as marked adverb phrases. This, nevertheless, perhaps needs further investigation. 

     [29]. See, for example, Emonds (1976:172-176), or Radford (1988). The examples here have been taken from Radford (1988). 

     [30]. It is true that not all "prepositions" accept clausal complements in the same way that many "conjunctions" do not allow noun phrase complements. But this is not enough evidence to destroy the argument, since we consider verbs items like hope and dive, even though the former does only accept clausal complements and the latter only nominal complements. 

     [31]. In his Scale-and-Category model, Halliday did not accept this. But, after some criticism (see Mathews 1966), he changed his mind. Mathews criticized the principle of <total accountability= in Halliday's rank-based theory since according to that a word like and would have to be considered a sentence consisting of a single clause, which itself consists of a single group, this in turn consisting of a single word, which itself consists of a single morpheme. Mathews' criticism provoked a change in Halliday's view, since in order not to have to accept that and is a clause, he (1966) had to concede that items such as and and or can <float= between ranks, and so might be best regarded as not having constituent status. Thus certain elements needed to be considered outside the rank-based hierarchy of constituents.

          Young, however, keeps on analysing coordinators as Adjuncts, although he does not give any special reason for his analysis:

                   - You (S) were (P) wrong (Cattr) but (A) nevertheless (A) you (S) made (P) the right decision (Co). (Young, 1970:223)

                   - He (S) opened (P) the door (Co) but (A) not (Aneg) the window (Co). (Young, 1970:223) 

     [32]. We are inclined to favour this second analysis, since we cannot have examples like:

                   * He went right to.

                   * I did it just because.

but we can have examples such as:

                   - He went right there.

                   - I did it just [for that reason].

          If this is so, we will have to distinguish between modifiers of words, modifiers of phrases and modifiers of clauses. Moreover, we will also have to investigate if the relation between modifier and phrase and between modifier and clause can be really defined as a head-and-dependent relationship.