Greek Syntax: Word Groups

[Note: this is one in a series of posts on Greek syntax and Logos Bible Software. See the Greek category for a full listing. The immediately previous post is here.]

As mentioned in a previous post, the syntactic analysis consists of three primary levels of annotation:

  • Base Level Analysis (Word)
  • Word Group Analysis
  • Clause Analysis

This post will introduce you to the Word Group level of analysis. If this sort of stuff floats your boat, then read on.

Before we can describe the benefits of having a text that denotes word groups, we need to go through some definitions. Don’t worry, it won’t be too painful. Note that these definitions are taken from documentation on the website, which you may want to peruse as well.

Word Group
A word group consists of a single head term and any and all of its modifiers, though it will frequently consist of just a single word.

Head Term
Informally, a word that does not depend/modify any other word in its group.

Any word contained in a word group that is not a head term is considered to be a modifier. These modify the head term either directly or by modifying words that modify the head term. Modifier groups may be nested infinitely (Ro 1.1-6 is an example of a long and heavily nested series of modifiers).

So, as mentioned above, a word group consists of, at minimum, a head term. The word group may also contain modifiers, or words that act to modify the head term in some manner. A phrase such as ? ????? ?????? is a single word group. The head term is ????? to which the other words are in a subordinate relationship. The terms ? and ?????? are referred to as modifiers.

In the annotation, there are four different kinds of modification relationships that are accounted for. An additional relationship of connection is described as well. The below definitions are taken from the Word Group Analysis specification, and in some cases are slightly modified.

Specification occurs when a modifier classifies or identifies the word it modifies. Common examples of specifiers are articles, e.g. ? ??????, and prepositions, e.g. ?? ?????. In a prepositional phrase such as ??? ??? ?????, both ??? and ??? are specifiers of ?????.

Definition occurs when a modifier attributes features to or further defines the word it modifies. Common examples of definers are adjectives (both attributive and predicative structure), appositional words or phrases, and adjectival clauses.

Qualification occurs when a modifier in some way limits or constrains the scope of the word it modifies. Common examples of qualifiers are words in the genitive and dative case, and also negative particles functioning at the word group level.

Relation occurs when a word specified by a preposition (i.e. the object of a preposition) modifies another element within the word group. For example, in the word group ?? ???’ ??? ????????, the term ??? is in a relator relationship with the head term ????????. This relationship only applies to prepositional phrases within word groups and not when the prepositional phrase functions as a clause component.

Connection is a relationship between two word groups (e.g. ?????? ??? ????????) or two modifiers in a single word group (e.g. ???????? ?? ??????? ??? ??????? ????).

So, we’ve got our terms down. The above terminology is all that is required to know when working with the word group analysis. The basic vocabulary is small (eight terms) and flexible. It makes syntax approachable — even for the relative neophyte.

But how is this used? Let’s examine a passage to see what the word group analysis looks like. Since we’ve examined 1 Timothy 4.6 on this blog before, let’s use that passage again to see what sorts of things the word group analysis can offer in examining this same passage.

The clause annotation considers 1Ti 4.6 to be a single primary clause (we’ll get to the clause annotation in a future post). Here is the Greek (UBS4) and English (ESV):

Greek (NA27) English (ESV)
????? ???????????? ???? ???????? ????? ??? ???????? ??????? ?????, ???????????? ???? ?????? ??? ??????? ??? ??? ????? ??????????? ? ??????????????· If you put these things before the brothers, you will be a good servant of Christ Jesus, being trained in the words of the faith and of the good doctrine that you have followed.

But that’s just the text. To understand the Word Group annotation, we’ve got to see it. That’s exactly what the below screen capture does. This is 1Ti 4.6, as a directed graph. This is a resource in Logos Bible Software. I can hover over the Greek text and the morphology will pop up (as shown in the screen capture). I can right-click and keylink to a lexicon like BDAG.

You will see abbreviations like df, ql and the like. These represent the modifier type. They are actually references, which means if you hover your cursor over them the meaning will be expanded. Since you can’t do that with the screen capture (click on it to see a larger version), I’ve provided definitions below:

  • Head: Head Term
  • sp: Specifier
  • df: Definer
  • ql: Qualifier
  • rl: Relator
  • cn: Connector
  • Head (Emb): This is the head of an embedded word group. Yes, word groups may be embedded in certain circumstances.

In the previous blog post I wrote regarding 1Ti 4.6, my question was “How do I find which things are good?” Or, better stated, “How do I locate where ????? is functioning attributively?” My approach was to do a search using the graphical query dialog to find where the adjective ????? agreed with a noun (in the general vicinity) in case and number.

But with word groups, along with the encoding of relationships of modification, this search becomes much more simple. Look at the screen capture again. (I’ll wait … ). You can see that here, ????? is marked as a definer, it is in a relationship of definition in regard to the head term. As mentioned above in the definition for definer, definers are typically adjectives or things in apposition. So instead of constructing a graphical query, worrying about agreement with nouns (or whatever), we can simply search for where ????? is a definer.

Doing that, we locate 13 instances (in 11 verses; 1Ti 4.6 and 6.12 have two hits each) in First Timothy. Just searching the Greek text for the lexical form ????? finds 16 instances in 14 verses. Doing the aforementioned search via graphical query locates 15 instances in 13 verses.

Which is correct? If looking for ????? functioning attributively — some of the hits from the graphical query are false positives. 1Ti 1.8 is an example of false positive because ????? functions as a predicate adjective, translated “the law is good”, not “the good law”. The same goes for 1Ti 4.4. The other hits jive.

It is possible that searching for ????? as a definer is too restrictive. One could search for where ????? is either a definer or a qualifier. This particular addition has no effect in the First Timothy, but it would in other books (Titus, for example).

The important thing to note is that word groups and modifiers allow us to bound our searches differently and more precisely. Instead of working through morphological criteria to approximate syntactic relationships, the syntactic relationships themselves can be directly searched. Instead of thinking about a word that “agrees” with something else in certain criteria, and specifying a number of words that may exist between the words; the relationship itself can be specified: Find where ????? acts as definer.

That is, we can start to think and query on a higher level, the level of syntax. Morphology can still inform our queries (the Base annotation, as mentioned earlier, has a full morphology and that information will be available during composition of syntactic queries) but we can be smarter about the portions we search.

But we can also search for different things, even in the realm of word groups. That’s the area the next post in this series will delve into.


  1. Rick … thanks for a helpful and stimulating post.
    How does syntactical analysis allow for alternatives e.g. in Romans 7:15 does “ou” modify “thelo” or “prasso”? Do the creators of the tagging database make that decision for you in advance?
    My question is, at what point are syntactical decisions made? Does the user of the above analysis tool simply have to work with the decisions encoded in the database, even if we should disagree with them?

  2. Hi Richard.
    The syntactic decisions are made by the project editors; so yes, decisions are made “in advance” — much like with current morphological databases and with editions of any text. When you use an English text of the Bible, for example, the translation decisions are made in advance.
    That said, please realize that there are other options. We’re working on an alternate syntax database for the New Testament that I’ll discuss in a later series of posts. This would be one place to look for a second opinion, much like Logos offers alternate Greek morphologies to consult for second (or third) opinions.
    Also, we realize that the evaluation of syntax can be much more subjective than, say, strict form-based morphological parsing. Because of that, we hope to build a feedback loop into the system. In your example of Romans 7.15, this means that you’d be able to, through the software, forward a note to the editors describing alternate readings you may support if you disagree with their reading. The editors will take this into consideration in reviewing and updating later versions of the database.

  3. Rick … thanks for that.
    I wonder if a future development might be a syntactical database system that includes possible variants along with the “main line”, and that any software used to display it could somehow allow branching as part of its display mechanism.
    As someone with intermediate Greek at best, I would go a long way before disagreeing with what the database editors considered to be the “main line”, but since the NIV and ESV committees disagreed about which verb was modified by “ou” in Rom 7:15, perhaps there is always scope for healthy debate.
    Being able to feed back to the database editor would be useful, but if the underlying system still only allows one reading of the text, how would it make room for alternative and equally valid possibilities?
    Nonetheless, this is an exciting development, not only for scholarly work, but also for those of us still working to get our Greek to a good standard, who regularly make exegetical decisions for purposes of teaching and preaching.
    Thanks again