I recently saw something… odd on YouTube. However, something good came out of this. See, back in Syntax class (and French Syntax, and even Semantics), when making a basic Principles and Parameters style tree of a sentence, based on X-bar theory, we were always told that everything needed to be formed in binary trees… except conjunctions for some reason. This always struck me as odd.
Now, this may have been because my professors wanted to wean me off of the PnP system entirely, thus negating the need to go into conjunctions at all (although the ultimate goal was to get me into Optimality Theory, more on why I think that’s a stupid idea at some later point), but it’s not like it’d be that hard to learn a binary method. I even came up with one on my own and used it on my syntax final:
Now for why the video is important. You see, the second line of this story is “The monkey and the crab were walking,” (Saru to kani ga aruite mashita.) The sentence is broken up by actions, so that it comes out like this: “Saru to… kani ga… aruite mashita.” (“The monkey and… the crab were… walking.”) Notice that the pauses are such that to (“and”) goes with the first word in the list, saru (“monkey”), and both are separated from the next word in the list, kani (“crab”) by a pause. This is because there is a prosodic boundary between to and kani. It is a minor prosodic boundary to be sure, but a more influential one than that between saru and to (since technically there’s some sort of prosodic boundary between any two different nodes on a PnP tree).
Why is this important? Well, try saying that same short list in English, with a pause where it seems natural. You just said something like “the monkey… and the crab,” didn’t you? See, in Japanese, the conjunction came before the pause, but in English it came after. That’s important because… Well, I guess I’ll have to get into some more detail on what that tree means for those of you who didn’t suffer through a syntax class.
You’ll notice three types of nodes in that tree. Those that end in a capital P are phrases, sets of words meant to interact with a specific type of word. A ConjP is a conjunction phrase, DP is a determiner phrase (think of determiners as articles (the, a, an) for now, although that’s not quite accurate), and NP is a noun phrase. The things with an apostrophe (actually a “bar”) are bar-level nodes, something that can have child nodes, but technically acts like a head. The base form is the head, a single word that must be in the associated phrase (Conj for ConjP, D for DP, etc.), although in some cases it can be an unexpressed word. Each phrase has four potential parts: a specifier, which is another phrase immediately below the parent phrase (like the DP on the left is for the ConjP, commonly necessary, but not always), the head (without which there cannot be a phrase), the adjunct (something which is both parented and next to a bar level, completely optional, and not shown here), and a complement (a phrase that is a sister to the head (child of a bar level if there is a specifier, child of the phrase if not) and very often necessary).
Now, X-bar theory isn’t perfect (hence the need for more flexible (Projection Principle) or theoretical (Optimality Theory) frameworks), but it does tend to fit most basic linguistic situations, and it’s easy for a beginner to analyse. Watch as I do so.
You may have noticed in the above tree that the head comes before the specifier. This is because, if you ascribe to the head directionality parameter (which you kind of have to if you’re working with X-bar theory, though again it’s imperfect) English is a head-initial language. Japanese, on the other hand, is a head-final language, which is why X no Y is translated as “Y of X,” among other things. (I personally believe that it’s related to their surname-first practice as well, but that’s something I haven’t really worked on in detail… yet.) This also means that the tree for this sentence would be built in the opposite direction, which shows that the prosodic boundary would be weaker between the conjunction and the first noun phrase, rather than the second. This predicts that, if a pause is warranted, it will be after the conjunction in Japanese, but before it in English. The following image shows a tree of the phrase in both English and Japanese, with the location of the relevant prosodic boundary indicated by the broken red line. Notice that in neither case does it bifurcate a branch until it reaches the ConjP level:
Please ignore the arrows; they’re part of transformational gramar that is irrelevant to this discussion.
Now, of course, this isn’t proof that this method of diagramming a conjunction phrase is the best (in fact, there’s growing proof that X-bar theory isn’t the best way at all), but it’s an indication of the predictive power of the hypothesis, and moreover it is no more difficult to diagram than any other X-bar level, as you can see from the fact that the ConjP is structured in essentially the same manner as the DP. So, if I can come up with this, I’m certain that my professors and the experts that wrote my textbooks must have as well. I’m smart, but I’m not that smart. So, if it’s not a difficult method to use, and it’s not a difficult method to come up with… why didn’t anyone teach me this in college?