Opentext.org & Traditional Grammar

It might sound odd to hear that one of the problems of Opentext.org’s database is that it depends on traditional categories to a fault. But then, perhaps “traditional” is an unfair description, since the kind of construction discussed below is noted as distinct in the reference grammars (e.g. BDF §412). Perhaps then the better word is “simplistic” or a dependency upon lower level grammars instead of the advanced reference grammars.

Probably the most evident example of this tendency grammar is its treatment of Greek participles. The common explanation of the substantival participle introductory and intermediate grammars is that it functions in the place of a noun and is commonly recognized by the presence of the article. “[I]f the participle has the article it must be either adjectival (proper) or substantival. Second, if it is articular and is not related in a dependent fashion to any substantive in the sentence, then it is substantival.”[1] Now this is generally a helpful working definition, especially for teaching students how to determine the function of a given participle, but it cannot provide explanation for every instance of a substantival participle. The result is that when we move to the world of structural representation this description fails. That is because there is more going on with substantival participles than simply the addition of the article. A close examination of Opentext.org’s representation of one such participle construction in Ephesians 4:24 is quite helpful in showing this point.

Fig. 1image

Note that a prepositional phase, κατὰ θεὸν (by God), appears between the article and the participle. Opentext recognizes that this prepositional phrase modifies the participle and yet follows the traditional understanding of substantival participles by crossing lines to connect the article directly with the participle. But this example suggests that what we actually have is a substantival participial clause functioning in the attributive position of the Noun Phrase.

This is confirmed by the fact that we have the same phenomenon occurring with Greek substantival infinitives. Opentext.org tags at least 338 substantival participles and infinitives with this gap structure.[2] This sort of crosscategorial evidence suggests that what we have here in Ephesians 4:24 is not a specific use of the participle or the infinitive, but rather a unique feature of the Greek article that allows it to turn a variety of different constituent types into substantives. On this understanding of the article, Ephesians 4:24 should look more like Figure 2 below.[3]

Fig. 2image

In this diagram, it is clear that the article is the specifier for the entire participial clause rather than only the participle itself.

[UPDATE] As a result of dialog in the comments, I’ve been corrected that this isn’t actually a problem with Opentext.org, but really framework specific issue. Apparently, there is a distinction made between the specifier node for the article and the rest of the nodes.

If that is the case, and I have every reason to believe it is, then the majority of my criticism is significantly less valid, since there still is a distinction being made somehow. So what I’m wondering at this point is whether 1) the distinction is something that is simply assumed to exist by practitioners of Systemics and not “physically” represented, or 2) the distinction would normally be easily seen in other Systemics work in some way not is missing from Opentext.org or 3) for some reason I’m simply not seeing something in the diagram that makes it clear that not all nodes are equal to each other (which is possible).


[1] Daniel B. Wallace, Greek Grammar Beyond the Basics: An Exegetical Syntax of the New Testament (Zondervan, 1997), 619.

[2] This is from searching for a [clause [conj – not present][clause component [word group [head term [modifier [word 1 – article]] [GAP] [word 2 – participle OR infinitive]]]]]. Since [conj – not present] disallows postpositives such as δέ from being in the GAP, but also rejects hits with non-post positive conjunctions (e.g. καί), there are likely more than 338 hits .

[3] Figure 2 is an adaption of Opentext.org’s representation using Logos Bible Software’s Sentence Diagramming tool. Nowhere in the Opentext database does such a structure appear. In fact there is only one place in Opentext.org where a modifier of any kind connects directly to a clause in Hebrews 4:10. This analysis could also be considered insufficient, since the article is subordinate to the embedded clause itself. It is more likely that the Article is governing the embedded clause as marking the attributive position to the Noun Phrase it modifies.

19 thoughts on “Opentext.org & Traditional Grammar

  1. I’m not convinced that this “crosscategorial evidence” is persuasive. Your conclusion might be necessary to explain your diagram, but that’s not an adequate proof. It is much simpler to use the (yes, very traditional) explanation that various forms/constructions may fill the modifier slot between article and [noun/participle/etc].

  2. I had thought of your suggestion previously, but that still doesn’t explain why Opentext crosses lines here since they don’t do that with regular article-modifier-noun constructions.

    If we take that approach, I would ask, does the κατα θεον still function within the participle clause? Or does it modifier the participial clause? As I see, it κατα θεον is the only argument phrase in this participial clause, so it seems to function better as being within the clause than in the attributive position.

    And actually, the fact that they don’t do that suggests that they recognize it as distinct from article-modifier-noun constructions.

  3. It might also help to look at a bit larger chunk of text as well. Note that this is part of a larger phrase in second attributive position: τὸν καινὸν ἄνθρωπον τὸν κατὰ θεὸν κτισθέντα ἐν δικαιοσύνῃ καὶ ὁσιότητι. That is: [article noun article modifier] as the basic construction, and each of those two elements also have embedded modifiers: [article modifier noun / article modifier noun]. And the basic phrase also has additional prepositional modifiers following (ἐν + 2 objs). An alternative description would see it as one nominal phrase with a (first attributive) modifier followed by a participial phrase in aposition with its own modifiers. But in either case, the initial articles both govern their own substantives. I’d prefer the first analysis of 2d attributive position which explains more obviously why it’s structured as it is.

  4. I would take κατα θεον as an instance of preverbal focus in the participle clause (Levinsohn, 2000, 37).

    Anyway, I knew that of the three posts I had about flaws in Opentext, I knew this one would be the most controversial – though your analysis of the prepositional phrase in the attributive position would also not have the gap that we see in Opentext here.

  5. If you think of the predicator κτισθἐντα as head term of the clause, you can see that you have only moved the article connection higher up in the same κτισθἐντα-NP as you intended. But Opentext.org separates the word group nodes from the clause “nodes” for a real reason. A participle is a hybrid: a declensional verb. κτισθἐντα relates to some dependents like DO or PP as a verb. It relates to other dependents like the article as a substantive. So they have two node slots for one verb. That is the reason that Opentext.org hooked the PP(relator) above and the article below and crossed the branches. Your rearrangement would be valid if all nodes were equal. It is typical of Constituent Structure diagramming, but it violates Opentext.org’s and Halliday’s basic organization. You are unraveling their whole ball of yarn and want them to start over on a different basis. Bonne chance mon ami.

  6. Dennis, you’re close when you write:

    You are unraveling their whole ball of yarn and want them to start over on a different basis. Bonne chance mon ami.

    That’s not quite it though. I don’t want them to start over on a different basis I want them to start over because they don’t seem to have done the grammatical analysis before the annotation.

    The amount of work that went into this database should have had a massive grammar volume published with it. Instead we have an inconsistent database and no grammatical explanation whatsoever for why it looks the way it does. There’s no excuse for that. None.

    So when you write,

    A participle is a hybrid: a declensional verb. κτισθἐντα relates to some dependents like DO or PP as a verb. It relates to other dependents like the article as a substantive.

    my first inclination is to think that perhaps that’s the case, but where is the evidence for such a claim? I don’t want them to do a generative analysis, I want them to do a rigorous analysis. As fa as I can tell Opentext isn’t (see my other posts, especially the one on relative clauses, which are a disaster). I don’t see any evidence that a Greek participle is a hybrid. Now, I’m more than willing to be convinced of the fact, but it needs to be proven (something that writing a grammar with the analysis could have done). That’s what doing syntax is about, isn’t it? At present, I don’t think we have a κτισθἐντα-NP. I think we have a verb in a complement clause. But my main gripe with Halliday’s Systemics is that’s just as much limited to English as Chomsky is – just from different theoretical directions. The syntactic analysis has always seemed to me more of a taxonomy than an analysis – a suspicion that only grew when no grammar came out at the same time as the completion of the database.

    Even still, my frustration with the analysis doesn’t change if I try to view things from your description. If not “all nodes are created equal,” then shouldn’t that be clear in the representation? Because as things stand, it isn’t. There is nothing in the diagram that tells me the article’s node isn’t equal to the others. So regardless of the crossing lines issue (I work within two very different frameworks, one of which does cross lines), I still consider the database to be flawed.

    But again, as I said, I knew this would be the most controversial post of the criticisms I have against Opentext.org. I’d suggest looking at the others I’ve written.

  7. Mike,
    I too would do the whole thing differently.
    And articulate, and perform QC, and attach analysis with evidence, too, if only I had a little more time.

    As Fitzgerald said,
    Ah, Love! could you and I with Him conspire
    To grasp this sorry Scheme of Things Entire,
    Would not we shatter it to bits–and then
    Re-mould it nearer to the Heart’s desire!
    XCIX THE RUBÁIYÁT OF OMAR KHAYYÁM

    But what Omar, like all lost romantics who do not acknowledge the fall of Adam, did not acknowledge was that his friend, although agreeing with him on the shattering bit, would not have agreed with him in the re-moulding phase. And neither would He.

    Not being a defender of Opentext.org, I only wanted to make the tiny point that it does have a coherent perspective that leads to a different structure that cannot be mixed with your traditional constituent structure norms. Hooking the article at the EC “node” is like boring a new shoelace hole in your church shoes. The fact that you redesigned a different shoe does not invalidate the design you bought.

    The hybrid nature of the participle is a morphological fact. The visible tense sign θε identifies the word as a denizen of the conjugational realm, specifically belonging to the sixth principal part, a pacifist principality, which also confines tense. But the ντα-morpheme classifies the word as a denizen of the declensional realm, probably a masculine accusative singular, specifically one of the third declension paradigms. Evidence for the normal substantivizer function of the article and the adverbial (to-the-verb) function of the PP is plentiful. Opentext.org is different, and its quality control is frontier level, but its organization is not completely loopy and can often be equivalenced to something closer to our heart’s desire. We are unlikely find an unassailable syntax model here below, so in trying to pick a better one we want to weigh as many of the competing arguments as we can understand.

    Sorry about the inappropriate use of the term NP, which is often supposed to refer to the lexical class noun of the head term. I was using it loosely as a nominal phrase, in the common manner of referring to the phrase function, here as an appositive because the second position attributive, which your example isolates, is structured as an appositive is. It would be better called a second position attributive adjectival phrase, functionally, and lexically a participial phrase (VP or clause in the modern English sense).

    Anyway, please keep up your good work.

  8. Hi Dennis, thanks for the response.

    I want you to know that when I originally wrote these criticism, they were part of a 20 page paper that gave much more explanation about my view of Opentext. I really do appreciate the work they put into it. The greatest fault of Opentext isn’t in the analysis or anything like that, their greatest fault is simply that they were the first and being the first will always extremely difficult. My hope is that we can build on their strengths and their weaknesses. I consider the vast majority of their work to be accurate and correct and as I’ve been working out my own perspectives on Greek syntax, I’ve been highly dependent upon Opentext. They’ve made it possible to do distributional searches that wouldn’t have been possible previously. So in spite of my critical words, I greatly appreciate having the database available.

    In retrospect, I wish I should have expanded footnote 3 into a actual paragraph rather than a footnote. I tried to note that placing the article on the EC was also inadequate. I actually would have preferred to have the article as a sister to the EC rather than a daughter, but the amount of work that would have taken to represent using what I was given just wasn’t worth the effort at the time.

    The rest of this comment is just ramblings now, feel free to move on if you’d like.

    As the question of the participle’s nature again, its just as likely in my mind that the -ντα morpheme is an agreement marker. The whether adverbial or adjectival, this morpheme always agrees with its referent – that’s why we have dative and accusative adverbial participles sometimes and also why we sometimes have genitive participles that agree with a noun and other ones where there is no noun to agree with (i.e. genitive absolute). Now whether I could get anyone else to accept that analysis is a major question, but it does work quite nicely. And its not completely unusual to have agreement morphemes marking things other than person and number. Russian past indicative verbs agree with their subject in gender and use the exact same gender morphemes that adjectives use for agreement.

    Anyway, that’s just something to think about. Its a perspective that I believe to work quite well.

  9. Mike,
    Just a pickup on your seminal observations.

    Abstract ideas blow with the wind. But the best are grounded. The noun and the verb are the suns; and adjectives, articles, pronouns, participles and adverbs are satellites that float around but are attracted to one of the leaders, explicitly or elliptically. Sure, the case morphemes are agreement markers, but which sun do they agree with? In German, Latin, Sanskrit, whatever, they agree with the noun predominantly, the declensional sun, not primarily with the verb, and in Greek not at all. And what kind of agreement? Verbs and nouns both have number. Verbs may occasionally have gender as in Hebrew sometimes. But only nouns and their followers have case. The morphemes are visible, not abstract. So the language has drawn some kinds of lines in the sand. If we blur them more than the original is blurred, our models get blurry. Even the so-called adverbial participle is directly aligned with the nominative subject, or some other declensional substantive. The Greek participle cannot shed its declensional limits.

    As for phrase nodes in models and diagrams, the multiplication of them to make various grammatical points is a plague of clutter. Sometimes it can degenerate into artwork. I personally have abandoned all models involving multiple nodes per word, like constituent structure, x-bar and Opentext.org, and have gone to the word grammar model of the minimalist, one-to-one correspondence between phrase nodes and words. It has more discipline and ruggedness. But I do not reject Reed-Kellogg or other diagramming tools.

  10. Its not that seminal of an idea, particularly since its not mine originally, at least not entirely. Either way, when I look at participles I see a verb with its arguments as an entire clause functioning as an adjective – not merely the participle. That’s hardly abstract.

    If you’re still reading this comment thread, I’d be interested if you share more about word grammar. That is, if you’re willing. Specifically, I’ve bee perusing Richard Hudson’s website (http://www.phon.ucl.ac.uk/home/dick/wg.htm)
    but for the life of me cannot find any discussion of how word grammar deals with discourse in general and information structure in particular.

  11. Mike,
    I would like to respond to both threads: the participle and the word grammar.

    Sometimes an exchange drifts. Here is a logical recap of the participle thread.
    1. You moved a declensional modifier in an Opentext.org (OTO) diagram up to modify the participial clause the way constituent structure (CS) diagrammers do, violating the opentext.org norms.
    2. I said that you can do that in the CS model, but OTO has a basis within the language for creating a model that divides its “nodes” into two separate kinds, the clause and word group, and for drawing a boundary line between them that cannot be crossed.
    3. You said you do not see a basis for that boundary, denying the legitimacy of the OTO divided model.
    4. I said that the participle, the verb from your example, has two ways of relating to its dependents: as a verb and as a substantive. The verb way goes with the clause components. The substantive way goes with the modifiers of the word group.
    5. You, abstracter and obliterator of distinctions, said that you did not see the two natures of the hybrid participle, denying the legitimacy of OTO’s division.
    6. I said that besides being attested functionally, the two natures are a morphological fact. the participle has a verb morpheme and a declensional morpheme.
    7. You abstracted and said you do not see the declensional morpheme as declensional. For you, it is just an agreement marker, still denying the legitimacy of OTO’s division. This kind of abstracting of one characteristic to deny another is not logical. It is like saying that a blue house is not really blue, it is essentially just a house.
    8. I said that even though case is an agreement marker, it is still declensional. Case endings pattern on nouns. Tense signs pattern on verbs. Different. Two, not one same, deep properties of the language.
    9. You responded without addressing this chain of claims. You have not agreed that case is declensional. You have not agreed that the participle is a hybrid. You have not agreed that it has two ways of relating to its dependents. You have not agreed that OTO has a legitimate basis for dividing the “nodes” for the participle into clauses and word groups.
    10. It is not a quesion of appreciation for OTO’s general value. I am not saying you are too critical. And I am not asking you to adopt OTO’s division in a model. I would not adopt it. I vote with you for the equality of nodes. I am only asking you to acknowledge that there are real verb-declensional boundaries in the language that provide a real legitimacy for the OTO discriminating model itself. You can hit on OTO. I would. OTO is like President Bush. He protected us from another 9-11 for 7 years. The criticism that he did not use the bully pulpit to communicate is legitimate, but the criticism that his protection measures were bad is not self evident just because we do not see it. Your statements that OTO should have volumes of documentation, justification and theory to back up its individual and global decisions is right on. Then we could engage. But the criticism that the divided model is not well grounded is not self evident just because you do not see it. You need to criticize OTO for the right reasons and only the right reasons.

    New thread direction: the phrase nodes in the model.
    I also looked at Richard Hudson on the web, although I have not read his books. He had plenty of examples so that his system of hierarchy was identifiable. In general I was very impressed with the system of constructing the concentric series of phrase set containments out of individual word-word relations. However, specifically, his aversion to tree diagramming is simply bizarre since the word grammar is well adapted to that kind of diagramming and the diagram is very important. Second, it was a typical English order-dependent model, ignoring the inflectional signals for syntax that all the inflected languages use. He indulged in the typical kind of foolishness you see among English grammarians who pretend to UG (universal grammar), for example making the article head of the noun because it was the article phrase initiator. I though he addressed syntax, whose unit is the sentence. I do not know how he relates syntax to the big picture.

    According to Trask, the word grammar is a dependency grammar, first fully developed by Tesniere(1959). It focuses on word-word relations in order to structure grammar. It is a counterpart to constituent structure, which focuses on constructing bigger phrases from littler ones. Also closely connected with dependency grammar in my mind is relational grammar, which takes a similar approach, focusing on hierarchy built from relations. It was started by Perlmutter in the 70’s, but is apparently more specialized. Anyway, when I said word grammar, I was just thinking of it as a representative of the dependency or relational grammar model that builds the structure from ordered word-word relations rather than from the CS join.

  12. I don’t know way you’re trying to make this an argument. I wasn’t arguing. So let me try to make one thing clear, the one thing that seems to be the problem here. In all ten of your points, you say that I’ve denied that OTO’s (<- I like this abbreviation, by the way) model is legitimate.

    Now the problem with that is I haven’t been even thinking about OTO the end of the first half of comment #8. Everything I’ve said about participles since the words, "…I greatly appreciate having the database available," has had NOTHING TO DO WITH OPENTEXT.ORG. Nothing at all. The topic had already changed. I had already headed into a different direction. So when you write ten times that I’m "denying the legitimacy of OTO’s division," I say that I haven’t been talking about it for some time now.

    This isn’t a debate, my posts and my comments are just random, (mostly) unorganized thoughts about Greek syntax. This isn’t a presentation at a conference or a thesis or anything where lengthy argumentation is necessary. Its a blog.

  13. Mike,
    Your rejection of the differentiated participle enables your rejection of OTO’s differentiated “nodes” on the participle. The logical connection is immediate. I do believe you when you say you were not thinking about the first part of the chain when you went on to the supporting part. But we are blessed to be wired for logic, even if it is unconscious. Blogs are part of life now. They are as good a place to learn from each other as anywhere. Again, best wishes and keep up the good work.

  14. That’s valid. I’ll give you that. If you could do me a favor then, would you read the post again. I’ve edited it in light of our discussion. I don’t want to make criticisms against them that are really criticisms of the theory itself. To be honest, I see value in all linguistic frameworks. There is no one framework that get everything right.

  15. That’s good. For the OTO system the predicator, subject, complement and adjunct are clause components to the EC/PC/SC. The sp/ql/df/rl are modifiers to the word group. There is a two tier (major/minor) system and they do not mix. Putting the article down in the wg does not mean that its scope is less than the PP adjunct, the way that it does in constituent structure. The scope is determined by which word the article attaches to. If it attaches to the predicator, its scope is the whole EC, if you make the equivalence that the predicator heads the clause.

  16. Little is the operative word, although they did have the clause and word group division in it. When they first encounter OTO, the older students will come speaking ReedKelloggese (RK) but the younger students will come speaking CSese. It is a little easier for the RKese speakers to translate than for the CSese speakers, because both OTO and RK have the non-hierarchical clause-line (one is horizontal and one is vertical) whereas the CSese speakers see everything in terms of hierarchical nodes. Of course the LITTLE glossary did not include the necessary OTOese-CSese dictionary. You just have to puzzle it out on your own. Maybe we could write one. I suppose there would be customers.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s