A systemic functional analysis of the “ ngp1 + vgp + ngp2 + ngp3 ” construction in English

: The literature to date features two very different approaches to the study of syntax, the formalor structure-based approach and the functional or semantics-based approach, both of which have advantages and disadvantages. The purpose of this paper is to investigate the “ ngp1 + vgp + ngp2 + ngp3 ” construction in English by identifyingthe relationshipbetween theclausalelementsin terms ofclausetypes. The theoretical support is Systemic Functional Linguistics which is semantics-based and which regards form/structure as the realization of meaning/semantics. Speci ﬁ cally, the paper discusses the issue by keeping in mind questions such as “ What kind of process is it? ” , “ How many participants can/must be involved in the process? ” , and “ What roles can/must those participants play? ” . By analyzing three pairs of clauses which share the same “ ngp1 + vgp + ngp2 + ngp3 ” structure, it is found that the clause types SVOO, SOVA, SVOC, and SVO all exhibit the same “ ngp1 + vgp + ngp2 + ngp3 ” structure. Theidenti ﬁ cationoftheclausetypes is conductedby examiningtheProcess in the Transitivity system of the clause. The implication of the present study is that a functional-syntactic analysis should start from meaning and consider how the meaning is realized and that although syntactic analysis at the level of form is necessary, the focus of a good functional-syntactic analysis should be based on semantic analysis at the level of meaning. the to become that like the way I make extremely


Introduction
In the description of clauses in English, one important step is to distinguish between central and peripheral elements of the clause. Quirk et al. (1985: 49) 2 Theoretical preliminaries Traditionally, English clauses are grouped into different clause types, which are regarded as canonical clauses. As Huddleston andPullum (2008 [2005]: 24) point out, canonical clauses are declarative clauses (e.g. She is patient.) while noncanonical clauses are interrogative clauses (e.g. Is she patient?) and imperative clauses (e.g. Be patient.). SFL scholars do not usually talk about clause types by reference to formal criteria, and they instead identify different clause types by examining their process types and the participant(s) expected by the processes. SFL distinguishes between major clause and minor clause, the former of which has the Mood structure (i.e. Subject + Finite). According to Halliday (1994: 43), "every independent clause selects for mood" and "an independent major clause is either indicative or imperative in mood". Due to the focus of this paper, the main examples to be discussed are independent major indicative-declarative clauses.
In SFL terms, "Subject" is a functional label when it is used with "Finite" to form the Mood of the clause, because "Mood plays a special role in carrying out the interpersonal functions of the clause" (Thompson 2014: 53). A certain structural element in a clause (e.g. the girl in The girl is patient.) becomes a functional element in different metafunctional situations: 1) Subject in terms of interpersonal meaning, 2) Actor in terms of experiential meaning, and 3) Theme in terms of textual meaning.
The present paper is an exercise in functional syntax (Fawcett 2000a), and in this framework it is important to distinguish between structural labels and functional labels on the one hand and "what something is" and "what something does" (Thompson 2014) on the other. The "ngp" (nominal group) and the "vgp" (verbal group) in the "ngp1 + vgp + ngp2 + ngp3" construction are structural labels because they tell us what something is, and focus is on "word class" (or part of speech, or group class). By contrast in terms of clause functions, in The boy has given the girl a book, the first nominal group (ngp1) (the boy) serves as Subject of the clause, and the focus is on what function is played by the first nominal group in the clause. For example, as Thompson (2014: 54) observes, in traditional terms, "the Subject is the entity of which something is predicated in the rest of the clause". In terms of expression of the interpersonal meaning, the Subject and the Finite enable the speaker to negotiate with the listener, by either giving information (as in He is away.) or demanding information (as in Is he away?) or proposing action (as in Go away!). The discussion here shows that items such as "ngp" and "vgp" are structural labels and terms such as "Subject" and "Object" are functional labels. When we come across something like the boy, we can say that it is a nominal group, which is a structural label; when it appears in a clause like The boy gets up very early, we can say in terms of its function that it is the Subject of the clause. In this case, "what something is" refers to the group class (or word clause) and "what something does" refers to the function it plays in the clause.
In the "ngp1 + vgp + ngp2 + ngp3" construction there are four slots, one of which (the vgp) is to be filled by a verbal element or verbal group (e.g. give, have given, gave) and the other three are to be filled by a nominal element (e.g. the boy, he). Therefore, we can say that there are a number of slots in any clause and that the choice of the filler of each slot depends on what is expected (or in Huddleston andPullum's 2008 [2005] term "licensed") by the process (verb) and what function it will play in the clause.
For Halliday (1994: xiv), "[a] language is interpreted as a system of meanings, accompanied by forms through which the meanings can be realized". And according to Halliday, forms of a language are important but are put in a different perspective: forms are "means to an end, rather than as an end in themselves" (Halliday 1994: xiv). As Halliday (1994: 106) says, "[l]anguage enables human beings to build a mental picture of reality, to make sense of what goes on around them and inside them", and "the clause plays a central role, because it embodies a general principle for modeling experiencenamely the principle that reality is made up of PROCESSES" (Halliday 1994: 106). To actualize the meanings expressed by the process, SFL emphasizes the important realizational relationship between form and meaning. In grammatical terms, the stratum of semantics is realized by the central stratum of languagethe lexicogrammatical stratum.
To translate "reality is made up of PROCESSES" into functional-syntactic analysis, we say that if there is a process in a situation at the level of meaning there is a verb in the clause at the level of form; therefore, if there is a verb, there is a clause (see Fawcett 2008: 48). In terms of grammatical analysis, the "ngp1 + vgp + ngp2 + ngp3" construction is a clause, and the combination of clauses will yield a clause complex, as in He gave her a book and she gave him a pen in return, which expresses two Processes realized in two clauses with two verbs.
In their comprehensive grammatical description of the English language, Quirk et al. (1985: 53) argue that in a clause "of the obligatory elements the main verb is the one that wholly or largely determines what form the rest of the structure will take". They identify seven clause types in English, which are presented below with examples: (i) Type SV: Someone was laughing.
(ii) Type SVO: My mother enjoys parties.
(iii) Type SVC: The country became totally independent.
(iv) Type SVA: I have been in the garden.
(v) Type SVOO: Mary gave the visitor a glass of milk.
(vi) Type SVOC: Most people consider these books rather expensive.
(vii) Type SVOA: You must put all the toys upstairs.
This kind of description of clause types is widely accepted in the study of the English language, and the general assumption is that the main verb plays a decisive role in terms of clause elements in the whole structure because it is this element that determines whether an Object and/or Adverbial are needed. Therefore, according to Quirk et al. (1985: 53-54), verbs are grouped into different classes: (i) intransitive verbs, which are not followed by obligatory elements and which occur in Type SV, (ii) transitive verbs, which are followed by an Object and which occur in Types SVO, SVOO, SVOC, and SVOA respectively, (iii) copular verbs, which are followed by a Subject Complement or an Adverbial and which occur in Types SVC and SVA. There are two points concerning the classification of clause types presented in Quirk et al. (1985) which are worth discussing. The first point is that the "V" representing "verb" (in contrast with noun, adjective, etc.) is a formal label while the "S" (Subject, in contrast with Object, Complement, etc.) is a functional label (Thompson 2014: 18-20), and clearly the combination of "S/O/C/A" with "V" ignores the distinction between the formal and functional ways of classification, since the two different categories are mixed. The second point is that it is the main verb that determines the element that follows the verb (or "what form the rest of the structure will take", Quirk et al. 1985: 53). The idea in the second point is in clear contrast with the SFL notion of Transitivity within the ideational metafunction, which refers to a system for describing the whole clause, rather than just the main verb and the obligatory element(s) following it.
In talking about the process type in SFL, there are three basic questions that can be asked about any process and the clause of which it forms the nucleus (Thompson 1996: 79): (i) What kind of process is it? (ii) How many participants can/must be involved in the process? (iii) What roles can/must those participants play?
With Clause (1a) below, it belongs to Type SVOO, with Mary as the indirect Object and a book as the direct Object, according to the analysis in Quirk et al. (1985).
(1) a. John gave Mary a book.
However, in systemic functional terms, in analyzing Clause (1a), we first need to determine the kind of process it is (Is it a Material, Mental, Relational, Verbal, Behavioral, or Existential Process?). Then once we know that it is a Material Process we will see how many participants are involved or expected in the process, in which case the answer is three. Finally, we must see what roles those participants play in the clause: the Actor (John), the Goal (a book), and the Beneficiary (Mary). In terms of systemic functional syntax, Clause (1a) above is made up of "Subject + Predicate + Complement + Complement" (with Finite conflated with the Predicator in the Predicate) in terms of clause structure, which is in contrast with the description of the Process and the associated Participant Roles (PRs) (Actor + Process: Material + Beneficiary + Goal) at the level of meaning.
Since the focus of this paper is the Transitivity analysis of the construction, we will follow Halliday's (Halliday 1994;Halliday and Matthiessen 2014) classification of process types, and the process types involved in the following discussion are the Material Process and the Relational Process.

Three pairs of clauses
This section will analyze three pairs of clauses, which share the same "ngp1 + vgp + ngp2 + ngp3" construction, by using the SFL assumptions and practices. The clause pairs are listed below: (1) a. John gave Mary a book. b. John gave a book to Mary.
(2) a. She made him a good husband. b. She made him a good wife.
(3) a. He gave her an iPhoneX. b. He gave her a kiss.
These six clauses are the same in terms of their structural organizations in that they all have the "ngp1 + vgp + ngp2 + ngp3" construction, but they are different in other aspects, which will be examined in turn in the following discussion.

Analysis of John gave Mary a book
According to Quirk et al. (1985: 53), Clause (1a) above, John gave Mary a book, belongs to the SVOO type, which means that there are two Objectsindirect, Mary, and direct, a bookfollowing the main verb. If we change the clause into Clause (1b) John gave a book to Mary, then it becomes the SVOA type. In talking about the relationship between indirect Objects and prepositional phrases, Quirk et al. (1985: 59) state that "SVOO clauses can be converted into SVOA clauses by the substitution of a prepositional phrase following the direct object for the indirect object preceding it" and they give the following examples to illustrate the point: -She sent Jim a card ∼ She sent a card to Jim.
-She left Jim a card ∼ She left a card for Jim.
According to the argument made above, Clause (1a) belongs to the SVOO type and (1b) to the SVOA type, as the indirect Object Mary (i.e. the Beneficiary) in (1a) has become the completive of the prepositional phrase to Mary in (1b).
Taking a systemic functional approach to syntactic analysis, we will first of all determine the Process type of Clause (1a), which is a Material Process expecting three PRs, the Actor, the Goal, and the Beneficiary in terms of Transitivity, at the level of meaning. Then in terms of clause structure we can carry out a syntactic analysis of Clause (1a) and identify the Subject, the Predicate (Finite + Predicator) and the two Complements (see Table 1).
In terms of Transitivity, Clause (1b) has exactly the same Process, number and type of PRs, as is illustrated in Table 2.
Note that the difference between Tables 1 and 2 only lies in the swapping of the places of a book (Goal) and Mary (Beneficiary) in the clause. But the question one may ask is whether a prepositional phrase can serve as a PR which is a Complement. If Clause (1b) is segmented as in Table 3 below, then the question of whether a prepositional phrase can realize a PR will not occur.   As was stated above, terms such as "Subject", "Object", or "Complement" (in contrast with labels such as "noun", "verb", and "adjective") are functional labels for clause elements, as they indicate the roles they play in the clause. However, terms such as "Actor", "Process", and "Complement" are semantic labels in the Transitivity system.
We now return to the question of whether a prepositional phrase can realize a PR in an SFL analysis. Before doing this, let us look at a non-SFL introduction to English grammar (Huddleston andPullum 2008 [2005]: 65) and see how it identifies the clause elements in Lucy gave the key to the landlord: Lucy is identified as There are a number of issues that are worth discussing but here we only want to emphasize that the prepositional phrase to the landlord is regarded as the Complement, filled by a prepositional phrase. (4) Lynn is a nurse.
(6) Lynn is in the room.
Now let us answer the question from an SFL perspective of syntactic analysis. Table 4 provides the analyses of the above examples, one of which (i.e. Clause (6)) has a prepositional phrase that serves as the PR of Attribute and that functions as a Complement.
The four Processes in Clauses (4)-(7) belong to Attributive Relational Process, which predicts two PRs: the Carrier and the Attribute. As Table 4 shows, in the Transitivity analysis Lynn is the Carrier in all the four clauses, and a nurse (a nominal group), very pretty (an adjectival group), in the room (a prepositional phrase), and upstairs (an adverb) are the Attribute in each clause. The Process is realized by be, which has two functions: (i) it is the Finite/Operator and together with the Subject (i.e. Lynn is) is the Mood; (ii) it is the Process of the situation/clause and this Process expects two PRs. Therefore, the term "conflation" is used to describe this phenomenon: the Finite/Operator is conflated with the main verb of the clause. Note that the PR (Attribute) in Clause (5) is realized by an adjectival group, that in (6) by a prepositional group, and that in (7) by an adverb (or adverbial group), which shows that word classes such as adjectival groups, adverbial groups, and prepositional phrases can serve as a PR.
To support the argument made here, let us use Fawcett's (2008) analysis of a similar clause. In analyzing the clause They are talking about climate change, Fawcett (2008: 190) identifies the prepositional phrase about climate change as the Complement of the clause, with they as the Subject, are as the Operator/auxiliary and talking as the main verb. Our analysis in Table 2 above is in line with Fawcett's analysis.
The analysis here shows that the starting point of doing functional-syntactic analysis is based on meaning/semantics rather than form/structure, as forms are "means to an end, rather than as an end in themselves" (Halliday 1994: xiv). The discussion here also indicates that an SFL analysis of Clause (1a) and (1b) shows that, although they seem very different from a formal perspective, both clauses share the same process types with the same number and type of PRs: to say that Clause (1a) belongs to the type SVOO and (1b) to the type SVOA is adopting a formal criterion based on syntax.
The analyses in this section indicates that in terms of formal analysis, Clause (1a) and Clause (1b) belong to different clause types, that is, SVOO and SVOA in Quirk et al.'s (1985) classification. But an SFL analysis yields a different picture: both clauses have the same process types and the same number and type of PRs, which also shows the results of different analyses with different (formal vs. semantic) criteria. It should be pointed out here that although Clause (1a) and Clause (1b) are the same in terms of the ideational metafunction they are different in terms of the textual or interpersonal metafunction with different information structures and foci.

Analysis of She made him a good wife
Clauses (2a) She made him a good husband and (2b) She made him a good wife look the same at first sight, but a closer scrutiny will show that they differ in significant ways. Before we analyze (2b), we need to look at (2a), although they have the same "ngp1 + vgp + ngp2 + ngp3" construction in terms of syntactic structure. In (2a) the Process is a Material one, which expects three PRs. In order to make things clear, let us look at two simple Processes with two PRs: She made a cake.
In these two examples, the Processes are Material Processes, each predicting two PRs, the Actor (she) and the Goal (a cake, the wall). The actions expressed in the clause cause the recipients to change in one way or another. In other words, before the action of making in (8) happens, there is no cake, and similarly before the action of painting in (9) takes place the color of the wall is different from what it will be after the painting. Therefore, we can see that the actions in both clauses have the meaning of "causation". We can change both clauses into the "ngp1 + vgp + ngp2 + ngp3" construction by inserting an element of structure into the clauses: (8) a. She made the cake delicious.
(9) a. She painted the wall green.
Again both (8a) and (9a) are Material Processes but they differ from (8) and (9) in that they have three PRs expected by the processes (Table 5).
According to Quirk et al. (1985), (8a) and (9a) belong to the SVOC type, and delicious and green serve as Object Complements. But in systemic functional terms, both the term "Object" and "Object Complement" are simply regarded as Complement, although the relationship between the first Complement (i.e. the cake, the wall) and the second Complement (i.e. delicious, green) is a relational one in terms of meaning. One may argue that the name of the PR "Goal" should be changed into what Fawcett (2000aFawcett ( , 2008 calls "Affected-Carrier", which means that in relation to made and painted, the cake and the wall are Affected and in relation to delicious (as in the cake is delicious) and green (as in the wall is green) the cake and the wall are Carrier. The reason for Fawcett to use the double marking of PR is that on the  (2a) and (2b). Obviously, (2a) is more or less the same as (8a) and (9a) in terms of expressing causation, the obvious difference being the word class of the second Complement: in (2a) it is a nominal group, i.e. a good husband, while in both (8a) and (9a) it is an adjectival group. We can conclude that the Processes in examples (2a), (8a) and (9a) are all Material Processes which expect three PRs: the doer/Actor (Agent, Attributor), the affected/done-to (Affected-Carrier), and the state of the affected (Attribute). As the relationship between the two Complements is like a Relational Process (cf. he is a good husband/the cake is delicious/the wall is green), and as the first Complement has a double role, it is advisable to treat it as "Affected-Carrier" (Fawcett 2000a(Fawcett , 2008, rather than simply "Goal" (Halliday 1994).
Having analyzed (2a) we now turn to (2b). The obvious conclusion one can make is that the relationship between the two Complements is not a relational one in terms of meaning, because him and a good wife does not construe a Relational Process (*he is a good wife). Any native speaker can easily tell that a good wife describes the first nominal group (she), not the second (him). If we ask the question of "What kind of Process is it?", the answer is that it is a Relational Process. In other words, the lexical item make in (2b) realizes a Relational Process, not a Material one as in (2a). This meaning of make is illustrated in dictionaries and grammar books. The following definition and example from Collins Cobuild English-Chinese Dictionary can help to show this "relational" meaning of the word make: If you say that someone will make a soldier, writer, etc., you mean that they have the qualities or the talent which they need to become that thing. EG I like the way you write, and I think you might make a journalist. She will make an extremely lively member of parliament. (Sinclair 2002(Sinclair : 1175 But then comes another question, as there are usually two PRs predicted by a Relational Process (i.e. Carrier, Attribute) (as in Clauses (4)-(7) above), what is the PR of the other nominal group? As she and a good wife are the two PRs of the Relational Process of making, him is the element that represents the person who receives the benefit of having a good wife. Therefore, its PR is Beneficiary, like Mary in Clauses (1a) and (1b). Martin et al. (2010: 123) regard examples such as (2b) as Benefactive (Attributive) Relational clauses, and according to their explanation, we can say that (2b) means (2c) below: (2) c. She was a good wife to him.
Tables 6 and 7 will show the similarity of the two clauses more clearly.
In doing transitivity analysis, one should remember that there are two types of PRs: a normal type that is only associated with certain process types, and another type that can occur with more than one type of process. For example, "Carrier" and "Attribute" occur with Attributive Relational Processes while "Identified" and "Identifier" occur with Identifying Relational Processes; similarly, "Senser" and "Phenomenon" are associated with Mental Processes, and "Actor" and "Goal" with Material Processes. A PR such as "Beneficiary" can occur with a number of process types, as the Examples (10) In these clauses (of three different Process types), for her, for her sister and for her son have the same PR: the Beneficiary. The criterion to determine whether a PR is obligatory (as are Carrier, Attribute, Identified, Identifier in Relational Processes) or optional (as is Beneficiary) is to see whether it is expected by the Process. For example, in most cases a Relational Process expects two PRs, the Carrier and Attribute (Attributive Process) or the Identified and Identifier (Identifying Process). Therefore, we can say that him in (2b), to him in (2c), for her in (10), for her sister in (11) and for her son in (12) serve as optional PRs, and they happen to play the role of the PR of Beneficiary.
(2) b'. She made (him) a good wife. c'. She was a good wife (to him).  He built a house (for her).
By contrast, in (1a) and (1b), Mary and to Mary also serve the PR of Beneficiary, but they are obligatory, not optional, because they are expected by the Process; their absence will make the clauses ungrammatical. Compare the following clauses: (1) b". ?John gave a book. c". ?John gave a book.
(2) b". She made a good wife. c". She made a good wife.
However, it must be emphasized (i) that the guiding principle in doing systemic functional analysis is meaning-based, (ii) that the choice of PRs is determined by the Process, and (iii) that the surface syntactic structure is not the decisive factor in identifying the PRs (see Huang 2007).
In certain contexts of communication, there are instances which do not require or expect the PRs to appear in the syntax. The following example is used by Fawcett (1987: 134) to illustrate the case of having no PR at the level of form: Give generously! In terms of semantics and Transitivity analysis, the Process of giving here predicts three PRs: the giver (Actor), the receiver (Beneficiary), and the thing to be given to (Goal). Suppose that this is the situation where a volunteer wearing a badge and holding a special box is asking passers-by on the main street in Cardiff to donate money for Cancer Research. Here Clause (13) is a proposal for action (i.e. an imperative clause), and the Subject/Actor is the listener (donator) in the real situation, the Goal is British pounds, and the Beneficiary is the organization of Cancer Research. In this situation, it is odd to say something like You give British pound sterling to Cancer Research generously. The point to be made is that whether the PRs appear explicitly in the utterance or not, they are there at the level of meaning and are expected by the Process of giving.
It may be of interest to look at another example to illustrate the difference between an obligatory and an optional PR: He went to Edinburgh from Guangzhou via London in 1988.
There may be very different analyses of this example within the SFL community.
Here we would follow Fawcett's (2000aFawcett's ( , 2008 approach to recognizing PRs other than Actor. There are four PRs in Clause (14): the traveler (Actor), the place to go to (Destination), the place of departure (Source), and the route via which the traveling takes place (Path), alongside a Circumstance of Time (Table 8).
The Process of going usually expects two PRs, the goer (Actor) and the place to go to (Destination). Therefore, we treat "Actor" and "Destination" as obligatory PRs and "Source" and "Path" as optional PRs.
We can now return to Clause (2b) and compare it with (2a) in a more straightforward manner. Clause (2a) is an influential Material clause (which is also analyzed as a Causative Relational clause, see Halliday and Matthiessen 2004: 237-238, 299;Martin et al. 2010: 123) and Clause (2b) is a Benefactive Relational clause (Martin et al. 2010: 123). The reason for saying that (2a) is a Material Process rather than a Relational Process is that the Process of making has an obvious causative meaning, the action of which influences the nominal group that follows it (i.e. ngp2), and this is the reason why the PR "Affected-Carrier" (proposed by Fawcett 2000aFawcett , 2008) is preferred to "Goal" or "Carrier" as proposed in Martin et al. (2010: 123).
In (2a) the relationship between the second nominal group (him) and the third nominal group (a good husband) is clear, and that between the first nominal group (she) and the third nominal group (a good wife) in Clause (2b) is also clear, because of the gender indication (i.e. he was a good husband/she was a good wife). That is why traditionally, a good husband in Clause (2a) is regarded as an Object Complement and a good wife in (2b) as a Subject Complement. According to Martin et al. (2010: 123), this clause can mean (i) she caused him to be a good friend, (ii) she was a good friend to him, and (iii) she made (produced) a good friend for him. The ambiguity occurs because a good friend with no gender indication can serve as a Subject Complement or an Object Complement in traditional terms (Quirk et al. 1985). The issue of ambiguity can be solved by considering the grammatical constraints of gender and number; thus, the meanings of the following clauses are clear: ( If we locate this construction in a context of communication, it is unlikely that any of these clauses will cause misunderstanding in most situations. The discussion above indicates that although She made him a good husband and She made him a good wife have the same structural organization of "ngp1 + vgp + ngp2 + ngp3", the two clauses belong to different Process types and the Processes expect different PRs: with the former three PRs are expected and thus him in She made him a good husband is an obligatory element, whereas with She made him a good wife the element him is an optional element, which is not determined or expected by the Relational Process.

Analysis of He gave her a kiss
The clause He gave her a kiss has the structure of "ngp1 + vgp + ngp2 + ngp3" and is the same as He gave her an iPhoneX in terms of its structural organization. Clauses (3a) and (3b) are repeated below for convenience: (3) (a) He gave her an iPhoneX. (b) He gave her a kiss.
As was discussed in Section 3.1 above, gave in (3a) is a verb realizing a Material Process which expects three PRs, the Actor (Subject: he), the Beneficiary (Complement: her), and the Goal (Complement: an iPhoneX) ( Table 9).
Like that in Clause (1a), the Beneficiary (Complement: her) in (3a) can function as the completive of the preposition to and the whole prepositional phrase can be put at the end of the clause.
With Clause (3b), the first question to ask is what Process it is. Obviously, it is a Material Process, but the next question is how many PRs this process expects. Example (18) below is the same as (3b) in terms of its Transitivity meaning: (18) He gave her a hug.
Semantically, give a kiss and give a hug are similar to kiss and hug respectively, and we can simply say that he kissed her and he hugged her instead of he gave her a kiss and he gave her a hug. In this case, the verb (gave) and the third nominal group (i.e. a kiss, a hug) together function as a single process in terms of meaning. That is, give a kiss equals kiss, and give a hug equals hug. This is similar to the expression of have/take a shower, have a bath, have a look, or give a laugh, which simply means shower, bath, look, and laugh respectively. The combination of give and a kiss in Clause (3b) expresses the meaning of a single process of kissing. Similarly, the combination of have and a shower/a bath/a look express the process of shower/ bath/look. Therefore, we can say that semantically she had a shower is more or less the same as she showered, both of which are Material Processes with one PR: the Actor (Table 10).
(19) a. She had a shower. b. She showered.  If we accept that he gave her a kiss is semantically the same as he kissed her, then we can say that example (3b) is a Material Process which predicts two PRs: the Actor and the Goal, in which case he is the former and her is the latter. Table 11 looks odd because "Predicate (Finite + Predicator)" is broken into two parts and is separated by the Complement. We can make the table look nice if we simply have three main columns in the table because there are only three main elements in Clause (3b): he (Actor), gave a kiss (Process), and her (Goal).
If we accept the argument made here, then we have to answer the question of what clause element a kiss is within the verbal group gave a kiss.
The analysis so far shows that at the level of meaning, in terms of Transitivity, she had a shower is a Material Process with one PR and he gave her a kiss is a Material Process with two PRs (the Actor and the Goal). Note that in Clause (3a) (He gave her an iPhoneX) the PR realized by her is Beneficiary whereas in Clause (3b) her plays the PR of Goal.
But how can we handle the groups in these clauses in terms of clause elements at the level of form? In other words, if a shower in she had a shower is not a Complement, what clause element is it? Similarly, if a kiss in (3b) (He gave her a kiss) is not a Complement, what is it at the level of form?
In dealing with "phrasal verbs" (Quirk et al. 1985), Fawcett (2008: 183-188) argues that a single Process at the level of meaning can be realized by more than one word at the level of form, and he uses the term "multi-word Process" to refer to such phenomena, such as brought up in the clause He brought up the problem again (Fawcett 2008: 184). In his analysis Fawcett treats brought as the Main Verb (M) in the clause and up as the Main Verb Extension (MEx). Fawcett (2008: 184) defines MEx as follows: "The Main Verb Extension has this name because it functions as an 'extension' of the Main Verb (M), so that the two elements JOINTLY express a Process" (emphasis in original). Fawcett not only regards combinations such as bring up, get away, go in as "M + MEx", but also argues that in Ike has a swim every day, has is a Main Verb and a swim is its extension: Ike [S] has [M] a swim [MEx] every day [A] (Fawcett 2008: 187). He compares the similarities and differences between Ike swims every day and Ike has a swim every day and argues that at the level of meaning they are the same and at the level of form they are different. As systemic functional syntax is a means, not an end, and as the functional analysis of syntax should reflect its functionality, it is essential to analyze the clause according to how its meaning is expressed (Huang 2007). One of Fawcett's (2008: 187) , which is more or less the same as Clause (3b). The functional-syntactic analysis of (3b) is presented in Table 11.
There are a number of reasons why the importance of the clause element "MEx" cannot be ignored. The most important point is that it helps us to analyze the clause in a systematic way. It should also be pointed out here that although we follow Fawcett's analysis in identifying a clause element which is named MEx, we use "Predicator Extension" instead of "Main Verb Extension", since we basically follow Halliday (e.g. 1994) in the identification of clause elements. Perhaps it is helpful to illustrate the different analyses of Clause (20) below and compare the different analyses in Halliday (1994) and in Fawcett (2008), by putting them into a Table 12. (20) He has kissed the girl.
There are a number of differences between Fawcett's (e.g. 2000aFawcett's (e.g. , 2008 and Halliday's (e.g. 1994) functional-syntactic analysis of clauses, and Fawcett even argues against Halliday's notion of verbal group (Fawcett 2000b(Fawcett , 2000c. One of the main differences between the two different analyses is the use of the technical terms. For example, in (20) kissed is a Predicator for Halliday (1994) but is a Main Verb for Fawcett (2008). Fawcett puts forward theoretical and methodological arguments for his choice of terms (see Fawcett 2000a), but this is not the place for us to join that discussion and/or evaluate its arguments. To help solve our problem in analyzing (3b), we find Fawcett's (2008) concept of MEx helpful, and therefore, it is termed as "Predicator Extension" for the sake of easy discussion. With examples such as he has a swim every day, the syntactic analysis at the level of form (clause element) and the semantic analysis at the level of meaning (Transitivity) look like that in Table 13.
He has a swim every day. If we look at the analyses of (1a) (i.e. John gave Mary a book), (2a) (i.e. She made him a good husband), (2b) (i.e. She made him a good wife), or (3a) (i.e. He gave her an iPhoneX), we can see that (3b) is different from all of them both in terms of clause elements (at the level of form) and Transitivity (at the level of meaning). It is clear that the basic or propositional meaning of combinations such as give a kiss/a hug/a laugh and have a shower/a swim/a look/a bath lies in the nominal group rather than the verb, which is not a real lexical verb, but a "light verb", whose meaning is general and vague. The discussion so far suggests that, although Clause (3b) (He gave her a kiss) and Clauses (1a), (1b), (2a), (2b), and (3a) share the same structural organization of "ngp1 + vgp + ngp2 + ngp3", Clause (3b) is significantly different from the other clauses because its "ngp3" is not an independent element of the clause but functions together with the Predicator as a verbal group or as a main verb extension (in Fawcett's [2000aFawcett's [ , 2008 terminology).

Two different ways of syntactic analyses
In Section 2 above, we reviewed the seven clause types identified by Quirk et al. (1985: 53), which have different clause elements. Although Clauses (1a), (1b), (2a), (2b), (3a), and (3b) (repeated below for the sake of easy reference) all have the combination of "ngp1 + vgp + ngp2 + ngp3", not all of them belong to Type SVOO, Type SVOC or Type SVOA, as they may seem to: (1) a. John gave Mary a book. b. John gave a book to Mary.
(2) a. She made him a good husband. b. She made him a good wife.
(3) a. He gave her an iPhoneX. b. He gave her a kiss. According to the criteria proposed by Quirk et al. (1985), Clause (1a) belongs to Type SVOO and (1b) belongs to Type SVOA; both Clause (2a) and Clause (2b) belong to Type SVOC, although the Complement in the former (i.e. a good husband) is an Object Complement and that in the latter (i.e. a good wife) is a Subject Complement; (3a) is the same as (1a), which belongs to Type SVOO, and (3b) also has the structure of SVOO in Quirk et al.'s (1985) classification. As our analyses in Section 3 show, we distinguish between syntactic analysis in terms of clause structure and semantic analysis at the level of meaning. More importantly, we carry out the syntactic analysis on the basis of the meaning, analyzed in terms of Transitivity, highlighting the importance of the SFL principle that form is the realization of meaning (Huang 2007). That is, the clause elements are identified according to the PRs expected by the Process in each clause. Therefore, if we put our six main examples into the category of the seven clause types identified by Quirk et al. (1985: 53), we have a very different picture, apart from the terms used (i.e. Predicate instead of Verb, Complement instead of Object). Compare: There are three obvious differences between our classification and that according to Quirk et al. (1985), as Table 14 illustrates.
Firstly, as was pointed out earlier, in the classification of clause types by Quirk et al. (1985), the difference between structural labels and functional labels is not recognized, as "S" (Subject) is a functional label (i.e. what something does) and "V" (verb) is a structural label (i.e. what something is).
Secondly, according to the formal criterion (as in Quirk et al. 1985: 59), John gave Mary a book and John gave a book to Mary are regarded as belonging to different clause types (SVOO vs. SVOA) while in our analysis both clauses have the same clause elements (SPCC), no matter whether the Complement element realizing the Beneficiary of the Process functions directly as a clause element or as a completive of the prepositional phrase. Thirdly, with Clause (3b), although there are three nominal groups, one of them is not regarded as functioning as a Complement but as an extension of the Predicator, since this nominal group expresses the basic or propositional meaning of the Process, with the Predicator realized by a light verb.

Assignment and name of PRs
In our Transitivity analysis of Clause (3b), we assigned the PR of "Goal" to the Complement filled by her, not the PR of "Beneficiary", which is in contrast with her in Clause (3a). Compare: ( This analysis shows that Clause (3a) and Clause (3b) belong to different clause types although they have the same number of structural groups at the level of form.
The her in Clause (3a) is regarded as the Beneficiary, as the Process predicts three PRs (i.e. Actor, Goal, Beneficiary). This is because in Clause (3a) the object/ thing to be given is an iPhoneX and her is the receiver of the thing, which is naturally the Beneficiary. By contrast, the Process in Clause (3b) only predicts two PRs, the Actor (the doer) and the Goal (the done-to). If her in (3b) were to be given the role of "Beneficiary" there would be no "Goal" in this Process, which means that there would be no receiver (or Affected, in Fawcett's [2008] term) of the Process of kissing.
In Clause (3b), the Complement is not a Beneficiary for two reasons: (i) in a Material Process if the Actor (Subject) initiates an action that affects the Complement, the PR of the Complement is the Goal (or Affected, as in Fawcett [2008]), as in John kissed/hugged/touched/beat Mary and Mary opened/closed/broke the door. Therefore, as He gave her a kiss/a hug is more or less the same as He kissed her/ hugged her at the level of meaning, her in (3b) is obviously the Goal, like her in He kissed her.
There is another point concerning the name of the PR "Beneficiary". In its literal sense, "Beneficiary" is positive, as in Clause (1a) and (3a). But as a technical term, "Beneficiary" can be assigned to the participant which/who does not receive something (the Goal) which is beneficial from the receiver's point of view. This is clear with the enemy in (22) and that guy in (23) below: We gave the enemy a good beating.

(23)
We should teach that guy a lesson.
Although we have the same give in (1a), (3a), and (22), the participants (receivers) which the actions reach or affect are different. In both (1a) and (3a) the actions are beneficial to the participants in the literal sense, but in (22), the Process of beating the enemy harshly is not beneficial to the enemy in the literal sense; it is "Sufferer" rather than "Beneficiary".
Similarly, in examples such as I taught her English, the receiver of the action of teaching has benefits and is a natural Beneficiary in the literal sense, but in (23), that guy does not get benefit from the action of teaching him a lesson, although one may argue that the Beneficiary would benefit after this event as he may actually learn from the "lesson".
In terms of the Process types, I taught him English is similar to I told her the news, which may lead to the assumption that both Processes are basically Verbal Processes, in which case the Process expects three PRs: the Sayer (I, I), the Receiver (him, her), and the Verbiage (English, the news). By contrast, the meaning of teaching somebody a lesson is idiomatic in that it does not express the normal meaning of teaching (and learning). Therefore, arguably, teaching in (23) can be regarded as a Material Process or a Verbal Process.

Conclusions
This paper is an academic exercise of functional-syntactic analysis, trying to analyze clauses with the structural organization "ngp1 + vgp + ngp2 + ngp3". The theoretical basis is SFL (e.g. Fawcett 2000aFawcett , 2008Halliday 1994;Halliday and Matthiessen 2014;Martin et al. 2010) and the guiding principle is the assumption that form is the realization of meaning (Huang 2007).
The discussion indicates that the identification of seven clause types (i.e. SV, SVO, SVC, SVA, SVOO, SVOC, SVOA) by Quirk et al. (1985: 53) has two obvious weaknesses. One is that they mix a functional label, e.g. Subject, with a structural label, i.e. Verb, the former being used to answer the question of what something does in the clause while the latter being used to refer to what something is in terms of word clause. The other weakness is that a formal analysis of the differences between John gave Mary a book and John gave a book to Mary does not help much in a meaning-oriented approach to language, especially in terms of ideational meaning.
The analysis in this paper highlights the importance of doing a functional analysis of syntax, and in so doing questions such as "What kind of process is it?", "How many participants can/must be involved in the process?", and "What roles can/must those participants play?" should be kept in mind.
The paper suggests that a good syntactic analysis should start from meaning and consider how the meaning is realized. Although syntactic analysis at the level of form is necessary, the focus of a good syntactic analysis should be based on semantic analysis at the level of meaning. The purpose of syntactic analysis should enable one to see how the meaning is expressed rather than how the forms are different. To summarize, our analysis shows 1) that although John gave Mary a book and John gave a book to Mary are structurally different, their experiential meanings are the same, and 2) that She made him a good husband and She made him a good wife are two different process types and have very different meanings, and 3) that He gave her an iPhoneX and He gave her a kiss are entirely different in terms of their experiential meaning.