Skip to content
Jul 18 / Steve Runge

The trouble with clines 2, SFL-III

Disclaimer: The post below refers the implementation of SFL in NT studies, which may or may not necessarily reflect Halliday’s intentions. My goal is to address the methodological and theoretical flaws of the approach to SFL popularized by Stanley Porter. There is reason to believe that it represents a departure from Hallidayan linguistics (10/2/09).

The last post on clines described one of the benefits they provide in description. However, like all things, there is a limitation to their utility. The more elements one tries to represent in the cline, the less precise the cline tends to be. As I mentioned in my first post in this series on why I have not adopted Systemic Functional Linguistics (SFL) as a methodology, I noted that certain descriptive strategies like trees have inherent limitations, based on their binary nature.

In as much as clines are helpful, certain applications of them within SFL have caused a significant amount of confusion, based on the claims made. This comes about when general principles are used to make specific, ostensibly meaningful claims. How does this happen? Here is the process.

Porter and his students have applied the concept of clines to the development of statistically-determined hierarchies of markedness for various aspects of Koine grammar. They are using a symmetrical view of markedness, not to be confused with asymmetrical. If you do not know what I am talking about, stop now and go read the posts on markedness assigned in my last post.

It all begins by taking statistical frequency of certain elements within discourse (e.g., tense forms, voice, case, etc.), and inversely correlating it to prominence. The more frequently a certain form occurs compared to the other options available in the system of language, the less prominent or significant it is to the discourse. The corollary to this principle is that the less frequently occurring options stand out more from the standpoint of prominence. Linguistically, the “road less travelled” is the more prominent one, and should thus be assigned more significance, or so the story goes. Generally speaking, this principle and corollary will hold true, but one must interpret the results, and not jump directly to prominence claims.

There are any number of factors which affect decision-making in language, and these factors impact the statistical results of general clines. Clines of this nature provide general tendencies, but cannot stand on their own without more filtering.

Consider Porter’s correlation of aspect to prominence. He uses a cline-based view as the foundation for his claims regarding the prominence of the Greek tense forms. Since the aorist is the most frequently occurring, it is the least prominent; whereas the perfect and pluperfect are the least-frequently occurring forms, and thus most prominent. He states,

“Modern linguistics has made students of language aware that language production may usefully be discussed in terms of its opposing choices, so that it is seen in terms of a coordinated system. This implies that when one element is selected, other similar elements in the language are not selected. The perfective (aorist) aspect is the least heavily weighted of the Greek verbal aspects, and hence carries the least significant meaning attached to use of the form. In Greek the aorist is what some have called the ‘default’ tense; that is, it is the tense chosen when there is no reason to choose another. The imperfective (present/imperfect) aspect is more heavily weighted, and to use it in opposition to the perfective (aorist) implies greater semantic significance. The stative (perfect/pluperfect) aspect is most heavily weighted, and to use it in opposition to the perfective (aorist) and imperfective (present/imperfect) aspects implies the greatest semantic significance.” (Porter, Idioms of the Greek New Testament, 22.)

His cline and judgments regarding semantic weighting are grounded in statistical frequency. Use of statistics is a high value to Porter, ostensibly because he believes that they provide empirical, quantifiable support to back his claims. Unfortunately, such claims often turn out to be “clouds without rain.” The claims are sweeping without caveat, and turn out only to apply in certain contexts. Unfortuanately, SFL does not provide details about when and where they apply, just the general claim.

While the use of statistics sounds noble, one must not forget that language and language study fall under the humanities, not the hard sciences. Language choices are made by human beings, not computers. There are mitigating factors influencing language choice that statistics alone cannot quantify. This is where attention to pragmatic factors come into play. Consideration of these factors provides the caveats to know when and where principles apply, and when they do not. While I agree that the aorist is the default tense, an asymmetrical view of markedness would seek to describe what each tense for does, i.e. what feature(s) it “marks” as present. Knowing that one form is more prominent than another holds little explanatory value. Was the decision made strictly on the basis of prominence, not semantics or other pragmatic factors? Based on the breadth of the claims, it is hard to arrive at a different conclusion.

Am I saying that statistics are of absolutely no value? μὴ γένοιτο! On the contrary, they are of great value in many respects for determining basic patterns, such as determining which form is most likely the default or most simple. But statistics are only as valuable as their interpretation. Depending on the size of the sample and other mitigating factors (e.g. genre, emotional tone of the content, stylistic idiosyncrasies of the writer), they can lead you to very reach wrong conclusions.

Statistics are a helpful starting point, but they can only take you so far. Let’s take a look as some practical problems with statistics, clines and symmetrical markedness, taken from the proceedings of a 2006 symposium applying SFL to Koine Greek. Here is a description of the basis for creating a cline regarding grammatical person in Koine:

“Prominence comes from the addresser’s choice. He/she uses certain linguistic devices to highlight certain material or part(s) of a text in order to pack the more important information and deliver it to the addressees. One of the most important means for showing prominence is [symmetrical, SER] markedness, whose fundamental principle is binary opposition. This notion of binary opposition of marked and unmarked is well demonstrated in the contrast between the non-third person and the third person. While the third person indicates general and less specific person, the non-third person gives a more specific and focused boundary: i.e., the boundary between you and I. Thus, the third person is unmarked, and the non-third person is marked.”(Jae Hyun Lee, “Toward a Central Part or Peak: A Methodological Consideration and Its Application to Titus as a Test Case,” Linguistic Institute for Ancient and Biblical Greek Working Document (ver. 0.1 Aug 2006), 106-107.)

The writer’s claim sounds helpful, applying an inverse correlation of frequency to prominence. Note also that the oppositions are binary, and thus can only consider one factor at a time, hence third-person vs. non-third person. In reality, this claim about the prominence of person is less helpful than what is appears.

In most cases, judgments about which grammatical person to select are grounded in the semantic requirements of the context, not the writer’s choice. If I were writing a story not involving me or the audience I am addressing, do I really have a viable choice to use anything other than the third person? SFL’s approach ignores a core presupposition that I have ranted about often: Choice implies meaning. Unless there is some viable pragmatic choice involved in making a linguistic decision, it is illegitimate to assign some special meaning to its use. If I have no choice but to use third person in writing a narrative, there is nothing to be claimed about its use.

On the contrary, I can prove the opposite of this cline holds true–is self-evident even–based on the principles of “choice implying meaning” and the “default versus marked” asymmetrical framework of markedness. Consider how the writer of John’s gospel ostensibly refers to himself in contexts that would prototypically expect the use of first-person:

John 18:15-16 (NASB): 15 And Simon Peter was following Jesus, and so was another disciple. Now that disciple was known to the high priest, and entered with Jesus into the court of the high priest, 16 but Peter was standing at the door outside. So the other disciple, who was known to the high priest, went out and spoke to the doorkeeper, and brought in Peter.

John 20:2-4 (NASB): 2 And so she *ran and *came to Simon Peter, and to the other disciple whom Jesus loved, and *said to them, “They have taken away the Lord out of the tomb, and we do not know where they have laid Him.” 3 Peter therefore went forth, and the other disciple, and they were going to the tomb. 4 And the two were running together; and the other disciple ran ahead faster than Peter, and came to the tomb first;

We would be lead to believe from the proposed SFL cline of person prominence that the use of the third person is the least prominent choice, based on its frequency. But I say again what I have said before: frequency alone is insufficient to make judgments regarding markedness. SFL’s clines rarely provide caveats, exposing their general nature. They are presented as though they closely model the writer’s choices, and thus can be widely applied without filtering the data.

Statistics are of great value in determining basic patterns, but they are only as valuable as their interpretation. Depending on the size of the sample and other mitigating factors, they can even lead you in the wrong direction. Markedness cannot be divorced from choice and expected norms. Choosing a form on the basis of semantic requirements is not the same as a pragmatic choice. Only the latter  can be correlated to prominence, at least as far as most of the rest of the linguistic world uses the term. Remember, although expected norms can often be inferred from frequency, it is not an exact correlation.

Let’s take a closer look at the self-reference data from John. Why exactly does the usage of third person for self-reference stand out? The expected norm in most languages is that one refers to oneself in the first person, not the third. This expectation has nothing to do with statistical frequency, but semantic clarity. John’s use of the third person here stands out, not because of the “prominence” of the person, but because of the break with the expected norms. He is breaking the Gricean maxim “Be clear.” Instead the strategy employs obfuscation by avoiding direct reference. Because he “breaks the rules” by avoiding directly identifying himself, the usage here stands out, bringing about certain pragmatic effects.

Prominence cannot be divorced from expected norms, and expected norms cannot be directly derived from statistics. The usage seen above can be better explained on the basis of expected norms than on some inherent prominence value of the grammatical person. Statistics often correlate closely with default expectations, but it takes proper interpretation of the data to yield meaningful claims. Most of the SFL clines that I have seen regarding the markedness of person, voice, case, etc., move straight from frequency to conclusion, without due consideration to semantic and other mitigating factors.

Let’s give them the benefit of the doubt regarding prominence, and narrow the claim to discourses in general. Let’s say that the claim they are making is that, within a given narrative, the places where the first and second person are used are typically more prominent than those places where it is not used. Generally speaking, first and second person are restricted to reported speeches, not narrative proper (unless it is a first-person narrative, such as the “we” portions of Acts). If this is the claim, it would be far more precise to state that, generally speaking, content reported in speeches is more prominent than that in narrative proper. If this is true, then the prominence claim is better associated with genre than with grammatical person. But even this claim would only sometimes be true, not always. Again, statistics often provide a helpful baseline at the start. Making claims about prominence directly from statistics, without caveats, will leave you with only general claims, and without the means of determining when and where they hold true.

Have I comprehensively demonstrated that all SFL clines are inherently flawed? μὴ γένοιτο! I have selected representative examples, the ones that practitioners of SFL themselves have used for the same purpose. Clines and statistics can be a very helpful thing, but they have also been used to assert claims that are at least misleading, if not inaccurate.

Like my MA adviser Dr. Perkins rightly impressed upon me, presuppositions hold the key to understanding someone’s view of the data. Understanding the impact of presuppositions is of the utmost importance, boys and girls. Adopting methods based  flawed presuppositions will paint you into a corner so fast it will make your head spin, all the while sounding empirical and robust. So like the tird knight in Indiana Jones and the Last Crusade said, “Choose wisely!”

It’s 6 o’clock in the morning, do you know where your presuppositions are?


  1. Mike Aubrey / Jul 18 2009

    I’m thinking Kwong would have been a good example here.

  2. Carl Conrad / Jul 18 2009

    What a splendid and useful discussion! Several reactions come to mind:
    (1) Re: statistics. I can recall, while working in a faculty committee on B.A. curricular requirements, arguing over whether History belongs with Social Science on grounds that History increasingly relies upon statistics) or with Humanities, William Gass (the novelist & poet) averred that even Mathematics belongs to the Humanities. Academics are all too prone to imagine that if you can bring measurement into a discipline, what you are saying is more scientific and less subjective. But do they know where their presuppositions are?
    (2) I don’t have the statistics to prove it (of course!), but I’m particularly skeptical about assertions made regarding Perfect and Pluperfect vs. Aorist; I think these “tenses” are in the process of fusing in the Koine, and that some choices between them are not necessarily significant.

  3. Steve Runge / Jul 18 2009

    Good point, Mike. It was beyond long anyhow, but there is no shortage of similar examples. For those not aware of Ivan Kwong’s work, he applied a symmetrical cline of markedness to word order combinations in Luke. My RBL review of his book can be read here.

  4. Benjamin Shaw / Jul 19 2009

    Dear Steve: A very useful post. Off the top of my head, at least three practical considerations are ignored by such a dependence on statistics. The first is that the New Testament itself is a pretty small sample from which to work. The second is that the problem is made worse by the fact that the NT was written by a variety of authors, each with his own style, so that what might statistically be true of Luke probably isn’t true of John. Third, there is the consideration of genre. So that what might be true of John’s style in his Gospel probably isn’t true in the epistles (which are so short as to be statistically meaningless), and certainly isn’t true of Revelation (which I think John wrote, but which to my mind can’t be proven or disproven statistically).

  5. Steve Runge / Jul 19 2009

    I think that your point is valid with regard to proving or disproving something with statistics. They provide a helpful baseline, but little more. Your comments reflect the points I made in the review of Ivan Kwong’s work in the comment above. But at some point, if extended too far, your point breaks down. Though there are different authors with different styles, they are all speaking the same language, even Mark! My point is that a suitable description of a grammatical device should work just as well in high-falutin’ Luke as it does in John or Revelation. This was part of the motivation for making the LDGNT analysis comprehensive of the GNT, to make sure it could really account for everything. Revelation was pretty straight-forward, it was Peter and 2 Corinthians that caused me fits in places. Languages have rules that are followed and are even exploited for pragmatic reasons. Statistics lacks the ability to filter out the pragmatic noise to get to the core principles and rules, in many cases. It also takes a solid typological understanding of how languages work, not just hypothesizing idiosyncrasies of Greek or John. Stephen Levinsohn once adjured me that if I claimed the something occurred in Koine that is not found in any other language, then I was probably wrong. I have done the research for the grammar, and can demonstrate that the devices I am claiming are present and accomplish comparable functions in other languages. To paraphrase a quote from F. Danker, “Linguistics is not for sissies.”

  6. Paul Gibbs / Nov 9 2009

    “Choice implies meaning. Unless there is some viable pragmatic choice involved in making a linguistic decision, it is illegitimate to assign some special meaning to its use. If I have no choice but to use third person in writing a narrative, there is nothing to be claimed about its use.”

    I suppose the SFL approach would say that your choice of third person provides the meaning needed for the listener to tell who you’re talking about. You chose “her”: not explicitly yourself, and not explictly me–so you “mean” someone else. The selection of person provides important interpersonal data within the context of communication. And, conveniently, “interpersonal” here is actually one of the three main SFL “metafunctions”–the various types of meaning embedded in linguistic events.

  7. Steve Runge / Nov 9 2009

    If I am semantically required to do something to “provide the meaning needed for the listener to tell who I’m talking about,” then there is no pragmatic decision involved. If I have to do something, can it be considered a choice? The utterance has meaning, but there are no pragmatic decisions regarding the use of a specially marked form compared to the normal, unmarked one. See the posts introducing markedness for more of an explanation.

  8. Anthony Smith / Nov 23 2010

    ” If I have to do something, can it be considered a choice?”

    Sounds like a theological discussion to me! jk. Dr. Runge, I love your work. I saw you at ETS and bought your book that day. I’ve nearly finished it and it is fantastic. I appreciate your desire to disseminate discourse studies into the hands of the church. As a preacher, this is exactly what I am trying to do!

    I’m glad that someone has said something about the obsession with SFL in NT studies. There are more (and possibly better) models out there. Either way, linguistic models must be understood for what they are-paradigms for understanding certain aspects of (a) language. They are not holy writ and must be explored and prodded, even thrown away if they don’t work.

  9. Andrew Rozalowsky / Aug 12 2013

    Hi Steve,

    Sorry I’m coming to this several years late! Since the discussion is still ongoing, perhaps you’ll permit it.

    I’m struck by a few things here. One, early in the post you talk about Porter’s reliance on frequency and I thought you were about to show it but quote only his conclusions about the grounding of the aspects from *Idioms* p. 22, giving me the impression that they somehow show his argumentation. Those conclusions, however, don’t show his reliance upon frequency but merely state the conclusions reached by applying several criteria for markedness, frequency/distribution being one of them. Distribution doesn’t show up at all in those conclusions and I would more quickly assume some sort of semantic criterion. His *Verbal Aspect* monograph, as I’m sure you’re aware, gives much more argumentation as does now his 2009 “Prominence: An Overview” essay, with more work coming down the lane.That said, there is reliance upon distribution (as you admit can generally be helpful).

    Two, I appreciate that this post attempts to deal with whether the approach is helpful/unhelpful. The other posts in your SFL/markedness series appear to me to say (not exclusively I admit) that if Porter is not a mainstream Hallidayan (in the current conception of Halliday? His views have changed since the 50s/60s and there are indeed different developments of SFL, biblical scholars or not) there is a problem. But Porter, himself a linguist (who can discuss any linguistic model at length on command), is interested in developing the most helpful model (agree with the conclusions or not) for Hellenistic Greek. It’s a fair statement to wish he were more explicit in where he differs but that he is choosing to deviate is, in my understanding, not a problem to him.

    Third, you’ve given me lots of things to chew on as I consider what may be helpful/unhelpful about SFL, so thanks for your work.

    So much more to talk about, but I look forward to the conversation continuing.

  10. Steve Runge / Aug 12 2013

    Hi Andrew,

    This post is quite dated, initial findings from my paper on the historical present. There is a fuller version of the critique presented at ETS 2010. These provide more detail and interact with the Stan’s dissertation and prominence article. This critique has been revised to focus solely on the misuse of contrastive substitution, going into substantially more detail about how contrastive examples should be selected. I spoke with Stan about these issues at SBL in San Francisco. It became clear I had properly understood his positions, and that we had some fundamental disagreements. However, he prefers to wait until things make it into print to respond to them. For this reason I submitted the article to NovT. It should be out early next year. You’d need to wait for that article to have a more up to date understanding of my position. My SBL paper in Baltimore will tackle some of the remaining issues.

    I would encourage you to read the primary literature for yourself rather than accepting Stan’s take on things. His understanding of markedness is fundamentally at odds with the claims of Comrie, Zwicky and others he cites as support. None use a symmetrical view of markedness, nor is this view synonymous or reconcilable with an asymmetrical approach.

  11. Andrew Rozalowsky / Aug 12 2013

    Thanks for your response, Steve.

    I look forward to reading your up to date findings.

    Re: the primary literature. Absolutely. I’ve been doing that all summer and hope to go where the evidence leads me.

Comments are closed.