SFL was developed in the early sixties by Michael Halliday, a language teacher and linguist. He based many of his initial ideas on the work of his teacher, language philsopher John Firth. Other authors such as Ruqaiya Hassan, JR Martin and CMIM Matthiessen have expanded upon Halliday’s contribution and today a rich body of SFL literature exists. Everyone working within the field of SFL is concerned with exploring the following question: how do people in real life situations use language to communicate ideas and build relationships?
Since SF linguists are interested in actual language use, this approach to linguistics is uncompromisingly data driven. All concepts and findings within the SFL approach are based on data generated by language users operating in a range of contexts, with a strong emphasis on every day, common situations. Given that SLPs are called upon to help their clients (re)learn communication skills that are relevant to life outside of the clinic, a theory of communication that is entirely derived from data gathered in the real world will likely provide a useful framework around which to build assessment and therapy.)
What is ‘functional’ about SFL?: Examining the three functions of clauses
A good place to start when trying to understand SFL is to focus on the individual terms in the name of this theory. We’ll start with the second term, ‘functional’.
Decades of research on spoken language produced by people using language in everyday tasks revealed that the main unit of spoken language is the clause (Eggins, 2004). Clauses can be arranged into longer cohesive segments of language called ‘texts’; a conversation where two people produce multiple turns, a narrative produced by a single speaker or a joke are just some forms of texts that occur in spoken interaction. Whenever language users produce a clause, they carry out three key functions: an ideational function, an interpersonal function, and a textual function. Each of these three functions is carried out simultaneously in clauses using three overlapping ‘metafunctions’ (in SFL terminology, a ‘metafunction’ is a set of resources that are used to make certain kinds of meaning)
The ideational metafunction transforms human experiences into language; it allows us to communicate information about the world and about our thoughts, essentially using language to represent ‘who did what to whom’. According to an SFL, we use the transitivity system to carry out the ideational function and communicate meanings about the state of the world. Within texts, certain words and groups of words denote participants (usually expressed through nouns) engaged in some kinds of processes (usually expressed through verbs) under certain circumstances (typically usually through prepositional and adverbial phrases) (Eggins, 2004; Halliday & Matthiessen, 2014).
In English, there are five major types of processes: material processes (physical actions such as eating or breaking), mental processes (e.g., thinking, imagining, liking, wanting), relational processes (describing or defining what “is”; e.g. verbs such as is, be, have), verbal processes (the action of conveying messages through language; e.g., say, yell, criticized, whispered), and behavioral processes (physiological experiences; e.g., see, cough, smile, hear).
For each type of process selected by the speaker there is a unique group of different types of relevant participants for a speaker to include in the clause. For example, when a speaker selects a material process which involves physical actions, or what we think of as verbs of doing (e.g., jumping, ran, eat), the participants can include the Actor or who did the process and the Goal or who the process was done to or directed at
| Jesse | jumped | onto the ledge |
| Actor | Material process | Goal |
In comparison, if the speakers selects a verbal process by describing a message that was transmitted through language the participants will not include an Actor or a Goal, but instead the participants would be the Sayer and the Receiver (Eggins, 2004; Thompson, 2013).
| The visitor | criticized | the decor |
| Sayer | Verbal process | Receiver |
Other processes are associated with other sets of participants. Table X provides more details on the different kinds of processes, participants and other elements that appear in clauses.
The third component of the transitivity system, circumstance, uses adverbs or prepositional phrases for adding informational details such as aspects of place and time, manner, cause, contingency, accompaniment, role, and angle (Thompson, 2013).
| The customer | quickly | ate | the guacamole |
| Actor | Circumstance | Material process | Goal |
The interpersonal metafunction helps language users to construct a specific kind of relationship between themselves and the message they are conveying and/or with the people they are interacting with. According to SFL, we use the mood and modality system to carry out the interpersonal function and communicate meanings about our relationship with our words and the people we are talking to. In order to analyze mood, we focus on clauses as units that provide a proposition which consist of a Subject, or an entity (person, place, animal, issue, object, topic, etc.) to which the proposition relates, a Finite, which provides information about the tense of the proposition, and a Residue, which provides details about the proposition. In English, Finite take the form of verb morphemes that denote tense. At a minimum, a Residue consists of a verb or a verbal group. The Residue (an element that always contains the verb) completes the proposition contained in a clause by providing info about processes that Subjects carry out.
In English, speakers can manipulate the presence or absence of the Subject or Finite, as well as the order in which these elements are presented to produce clauses with different moods. Three basic moods exist in English, namely declaratives, interrogatives, and imperatives.
Declaratives consist of a Subject followed by a Finite and Residue and provide the listener with a statement:
| Jesse | jump | -s |
| Subject; entity to which the proposition relates | Residue; provides details about the proposition related to the subject | Finite;present tense verbal morpheme which provides information about when the proposition took place; |
Interrogatives consist of a Finite followed by a Subject and ask the listener a question:
| Can | Jesse | jump |
| Finite; present tense verbal morpheme which provides information about when the proposition took place | Subject; entity to which the proposition relates | Residue; provides details about the proposition related to the subject |
Imperatives consist of Residues only and function as commands given to the listener:
| Jump! |
| Residue only |
Another aspect of the interpersonal function SFL examines is the modal system. This system allows us to expresses attitudes of various kinds. By using modal verbs (e.g., should, might, could), modal adjectives (e.g., frequent, usual) or modal adverbs (e.g., probably, certainly, typically), we can convey to the listener different meanings about our feelings concerning the proposition we are communicating in the clause.
Each of the following clauses has the same basic underlying ideational meaning (i.e. all of the clauses describe the same participants and processes) but in each case, the speaker constructs a different relationship between themselves, the listener and the proposition contained in the clause. A brief explanation of the implications of the mood and modality configuration for each clause is provided:
Jesse jumps.
A declarative statement; the speaker presents the proposition as a simple fact.
Jesse usually jumps.
A declarative statement; the speaker provides a qualifier in the form of the adverb ‘usually’ to indicate her belief about the usuality of the proposition
Jesse should jump.
A declarative statement featuring a modal verb ‘should’; the speaker believes that it would be a good idea for Jesse to jump
Jesse might jump.
A declarative statement featuring a modal verb ‘might’; the speaker believes that it is possible that Jesse will jump at some point in the future
Jesse might’ve jumped
A declarative statement featuring two modal verbs ‘might’ and ‘have’; the speaker believes that there is the possibility that Jesse jumped at some point in the past
Jesse never jumps
A declarative statement featuring the qualifier ‘never’ attached to the verb ; the speaker believes the proposition contained in the clause has occurred zero times and will occur zero times in the future
Can Jesse jump?
A interrogative; the speaker is enquiring about the possibility that Jesse is able to jump; the question casts the speaker in the role of asker and the listener is the role of someone who is likely to know the answer
Did Jesse jump?
An interrogative; the speaker is enquiring about whether or not the proposition of the clause was true at some point in the past; the question casts the speaker in the role of asker and the listener is the role of someone who is likely to know the answer
Could Jesse have jumped?
An interrogative; the speaker is enquiring about the possibility that the proposition was true at some point in the past; the question casts the speaker in the role of asker and the listener is the role of someone who is likely to know the answer
Jump!
An imperative; the speaker is giving the listener a command to carry out; the command casts the speaker in the role of someone who can give commands and the listener who should carry out commands
The textual metafunction allows speakers to emphasize one part of the clause over others and to organize the flow of information across clauses. According to SFL, we use the theme/rheme system to communicate meanings about which part of a clause should be interpreted as being most important. In English, the first portion of the clause constitutes the theme and the remainder of the clause makes up the rheme. The theme provides the main information of the clause and the rheme provides further information concerning the theme.
Topical themes take the form of processes, participants, or circumstances. Each of the following clauses has the same basic underlying ideational meaning (i.e. all of the clauses describe the same participants and processes) but in each case the speaker has selected a different topical theme in order to highlight different parts of the clause. A brief explanation of the implications of the theme choice in each clause is provided
The woman recently ate the guacamole
The speaker wants to emphasize who carried out the process.
The guacamole was recently eaten by the woman
The speaker wants to emphasize the goal of the process, perhaps in order to clarify which food was eaten by the woman, or in answer to a question about which foods the man ate.
Recently, the woman ate the guacamole
The speaker wants to emphasize information about when the process took place, perhaps as part of a discussion about when certain recent events occurred.
Textual themes indicate how the ideas in a current clause are linked to previous clauses. Conjunctions (e.g., and, but, while, because) tell the listener about how the present clause is related to previous clauses. For example, the theme because in the clause Because she was hungry tells the listener that the rheme she was hungry should be interpreted as the reason for why a process in a previous clause occurred. Similarly, conjunctive adjuncts (e.g., in other words, as a result, actually), and continuity adjuncts (e.g., yes, no, well, oh) create cohesion between previous and current clauses. Clauses may be cohered to previous clauses produced by one speaker. For example, in the clause complex I asked for more guacamole so I could give you some ‘so’ acts as theme of the second clause and tells the listener about the relationship between the first and the second clauses). Alternatively clauses may be cohered to clauses produced by other speakers. For example, if Speaker A asks Speaker B Did you buy enough ripe avocadoes to make guacamole?, Speaker B might answer Well, I tried but there weren’t any at the store. The well in the second clause functions as the theme and lets Speaker A know how Speaker B wants their clause to relate to the antecedent clause; in this case well lets Speaker A know that Speaker B interprets the answer they are about to give as bad news that Speaker A will likely not welcome.
Interpersonal themes are used to indicate opinions and feelings, about what follows. In each of the following clauses, the interpersonal theme in bold indicates some aspect about the speaker’s view of the rheme:
Apparently, it’s not okay to eat all the guacamole.
Obviously, it’s not okay to eat all the guacamole.
Perhaps, it’s not okay to eat all the guacamole.
As mentioned about, the textual metafunction is concerned with creating longer pieces of coherent language (several turns between speakers in a conversation, multiple clauses produced by a single speaker, speeches, narratives, etc). Assessing how speakers handle themes and rhemes could reveal something about their functional ability to construct and maintaining the macro-structure of any interaction. This may be especially valuable for clients who have difficulty introducing new topics, maintaining the current topic across an extended interaction, indicating to interlocuotrs that they wish to change topics and demonstrating how the ideas they convey are related to previous ideas discussed in a conversation.
As mentioned at the beginning of this section, each clause communicates ideational, interpersonal and textual meanings at the same time. Below, we demonstrate how a single clause can be analyzed in terms of the three types of meaning.
| I | should’ve | bought | more | avocadoes | yesterday | |
| Ideational function | Actor | Process | Goal | Circumstance | ||
| Interpersonal function | Subject | Finite | Residue | |||
| Textual function | Theme | Rheme |
What is systemic about SFL?: Making choices in context
The first term in the name of SFL is ‘systemic’. One of the central insights Halliday and other writers offer us is that people choose from a ‘system’ of choices when developing clauses and longer pieces of language. For example, a speaker engaged in conversation can choose from a range of mood options when formulating their turn; they could produce a declarative, a wh-interrogative (‘Wh’-word question), a polar interrogative (‘yes/no’ question), an exclusive imperative (command directed at another person), or an inclusive imperative (a command directed at the speaker and listener; ‘Let’s grab a bite to eat’). Similarly, speakers choose from a range options when creating ideational meanings (who does what to whom) and textual meanings (determining which part of the message should be emphasized).
How do speakers make choices about the clauses they build? SF linguists argue that our choices are strongly influenced by the context in which an interaction takes place. According to SFL, language users consider three primary issues when constructing clauses and texts that are appropriate for a given context. Each of these issues influences the choices that speakers make when selecting options from each metafunction.
The field of a context concerns the topic or the subject matter that a text is about; in a conversation, the topic of the discussion acts as the field. The topic of conversation will play a central role in influencing how a speaker uses components of the ideational metafunction to talk about the world, determining what processes, participants and circumstances speakers will encode in clauses.
The tenor of a context concerns the relationships that exist between participants; in a conversation, interactants may be equals in terms of status or knowledge, or one participant might know much more about a given topic than others. Other relationship factors such as whether participants have known each other for a long time, and whether their relationship is professional or personal, and the level of intimacy between individuals in a given situation form part of the tenor of a conversation. All of these and other aspects of the inter-participant relationship will influence how a speaker makes interpersonal metafunction choices, determining the mood and modality choices that speakers make.
The mode of a context concerns the medium of communication and norms for constructing larger segments of language. Face-to-face conversation usually takes place in a spoken or signed medium. Different genres of conversation are structured differently; the sequence of turns that speakers produce when engaged in casual chitchat differs from the sequence found in interviews which in turn differs from the sequence noted in retail service encounters. These mode aspects influence how speakers make textual metafunction choices, determining how speakers emphasize portions of clauses and how speakers cohere clauses into longer texts.
Taken together, the field, tenor and mode of clauses and texts used in a given situation make up what SF linguists call the ‘register’. Different registers, with different field, tenor and mode configurations, may be typical in different contexts; for example, the register variables (and therefore the types of clauses and texts) we encounter in casual conversations between friends are different from the register variables (and therefore the types of clauses and texts) we encounter when we observe a boss and employee talking over lunch. The cultural environment in which a conversation takes place also shapes the field, tenor and mode of clauses and texts. For example, many white Anglo-Saxon Americans regard talking about personal wealth to be taboo, so certain fields and ideational components (process and participants) tend not to be used in conversations between people from this background.
Applying SFL to casual conversation: Discourse Structure Analysis
Up to this point, our discussion has focused primarily on how speakers produce clauses. We can now zoom out and think about the role that groups of clauses (called ‘clause complexes’) play when two people are participate in a dialogue. According to SFL, conversation is a ‘give and take’ exchange-based activity. In goal focused encounters (for example, when purchasing something in a store), speakers use resources from each of the three systems (ideational, interpersonal and textual) to construct clauses of strings of clauses that help them either demand (‘command’ in Halliday’s terminology) or offer good-and-services. Similarly, in less goal focused encounters (for example, when two friends meet to chat without a specific, clear agenda in mind) speakers harness linguistic resources to request information via questions or provide information via statements. Halliday (Halliday & Matthiessen, 2004) argued that these four actions (command, offer, question, statement) represent the basic speech roles or speech functions assumed by speakers during conversation.
In the early 2000s two linguists, Suzanne Eggins and Diane Slade, applied the principles and concepts developed in previous SF studies to casual conversation. They expanded Halliday’s original four speech functions in order to more precisely describe and account for the processes they encountered when analyzing conversational data. Their efforts lead to the development of Discourse Structure Analysis, an analysis framework which helps us to understand how conversations are built over time by at least two interactants. For more information about Discourse Structure Analysis, the reader is advised to consult Eggins and Slade (2004) and Armstrong and Mortensen (2006).
Eggins and Slade’s found that when analyzing casual conversation, the easiest unit to identify is the turn which includes all of the talk produced by one speaker before another speaker produces talk (Eggins & Slade, 1997). However, a turn can include multiple clauses that have an impact on the direction a developing conversation moves in. Thus, Eggins and Slade break down turns into ‘moves’ and use this unit as the lowest level of analysis. A move can be identified within conversation as a unit after which speakership could change and the transfer would occur without interruption of the original speaker. Moves may be composed of clauses (groups of linked clauses), or a clause and all its embedded clauses. Table XXX provides details on how the boundaries of these three units (turns, moves and clauses) compare.
| They | don’t | like | cilantro | because | they | say | it | tastes | soapy | ||||||||||||
| Turn | They | don’t | like | cilantro | because | they | say | it | tastes | soapy | |||||||||||
| Clause | They don’t like cilantro | because they say | it tastes soapy | ||||||||||||||||||
| Moves | They don’t like cilantro | because they say it tastes soapy | |||||||||||||||||||
While the move was originally defined by Halliday, Eggins and Slade developed guidelines for identifying moves within casual conversation (Eggins & Slade, 1997). These guidelines are based on the function that the move has within the conversation. Here we provide an overview of the most important moves.
Opening moves introduce new subjects or ideas for participants to engage with and initiate a sequence of moves that relate to these new ideas (Eggins & Slade, 1997). In conversation, opening moves may take the form of statements of opinion or fact, or questions about opinions or facts.
Two examples of opening moves are provided below:
Cilantro is disgusting! (statement of opinion)
When will the avocado shipment arrive? (question about a fact)
Once an opening move has been made, participants in conversations are faced with choices about how to engage with the opening move. As illustrated in the example below, the current speaker may choose to make a prolongingmove, holding the floor and expanding on their opening contribution:
Cilantro is disgusting….because it tastes like soap!
Or another speaker might make a reacting move to follow up on the opening move. Two sub-types of reacting moves exist: responding and rejoinding moves. Responding moves ends the sequence initiated by the opening move;. In the three examples below, a question of fact is followed by an answering move, a statement of opinion is followed by an agreeing move, and a statement of fact is followed by a registering move. In all three instances the follow-up brings the interactional ‘business’ related to the opening move to a close
Speaker A: When will the avocado shipment arrive? (question about a fact)
Speaker B: Next Wednesday (responding move which provides the requested info)
Speaker A: Cilantro is disgusting! (Statement of opinion)
Speaker B: Oh my goodness, it totally is! (responding move that agrees with the previously offered opinion)
Speaker A: I added more cilantro than the recipe called for (statement of fact)
Speaker B: Mhmmm (register move; brief turn that demonstrates the speaker is attending to other participants)
Rejoinder moves prolong the sequence initiated by the opening move and allow participants to keep engaging with the original opening move. In the example below, an opening move in which a speaker offers an opinion is followed by a rebound move in which the speaker questions the legitimacy of the previous move:
Speaker A: If you don’t add cilantro, it’s not guacamole (statement of opinion)
Speaker B: Are you kidding me? (rebound move in which speaker questions the legitimacy of the previous move)
Another kind of rejoinder move (a move that will keep the sequence related to the initial opening move going) is the tracking move. Tracking moves occur when a speaker says something ambiguous or difficult for the listener to understand. As can be seen in the example below, tracking moves alert the speaker to the fact that a communication breakdown has occured.
Tracking moves tend to be followed by resolving moves and then relevant responder moves (bascially, the conversation continues once the communication breakdown has been repaired):
Speaker A: Are you a fan of cilantro? (opening move, question concerning a fact)
Speaker B: What’s ‘cilantro’? (tracking move; Speaker B has failed to understand the entirety of Speaker A’s turn)
Speaker A: It’s a leafy green herb people put on guacamole (resolving move designed to help clarify previous move made by Speaker A)
Speaker B: Oh yes! I love that stuff! (answering move relating to original opening move)
Within a conversation, moves are arranged into exchanges. Each exchange begins with an opening and continues until the participants are done negotiating the proposition offered in the opening move; until they have handled all the interactional ‘business’ related to the opening move.
Different types of moves have different kinds and levels of impact on the ongoing conversation. Opening moves intro new material for participants to talk about and thus play an important role in building on and introducing topics. Reacting moves, by comparison, are always made as moves that follow-up on opening moves. These moves thus have less potential for moving the conversation in a direction. Within a conversation, the people who make opening moves have more active control over what is talked about while the people who make reacting moves follow the lead established by opening move makers. If one person makes many more opening moves than others, then the conversation may start to look one-sided, with one person exercising a lot of control over what is talked about.
DSA thus allows researchers and clinicians to form a global assessment of the entire conversation and the role that each participant played there-in. In interactions between people with and without neurogenic communication disorders, it is possible (or perhaps even likely) that the neurotypical partner plays a much larger role in determining what is talked about and the direction conversations move in. Various researchers are aware of this possibility and have used DSA to examine issues like participation and agency in interactions. Several have employed E and S’s framework to identify potential barriers between PWA and their communication partner when engaged in high-level, convergent-seeking conversations and provide avenues in communication partner training for addressing these barriers (Armstrong et al., 2013).