#syntactic tree

Today, we want to introduce you our two cacti-lovers...it s Gary and Samantha. Both of them have a dream to be someday syntactic cacti trees, but first they have  to stop fighting with each other...

The ‘tree’ is a fantastic way to represent hierarchical, branching structures. We find them in family trees, phylogenetic trees, glossogenetic trees, and they are used by many linguists to represent syllable and sentence structure.

In representational models of syntactic structure, we effectively have a tree-frame with branches occupied by the elements of the sentence we are representing. Representational models are very useful for descriptions but many syntacticians take the view that representational ‘trees’ are not so good for explanations. Within generative grammar, X-bar Theory started life as a theory about the representation of sentences. It identified important structural relationships obtaining between elements, for example, c-command (based on dominance and sisterhood) which generally says that A c-commands B iff A is the sister of B or B is dominated by the sister of A. By analogy with family trees, it would be like saying, my mother (A) c-commands my maternal aunt and all my aunt’s descendants. C-command relationships are most often invoked in binding (e.g. in determining the reference of reflexives such as ‘himself’ and ‘myself’). This is oversimplifying, but the point is that only certain relationships seemed to be syntactically significant. The c-command relationship picks out one of many possible relationships that can be traced on a tree diagram. Why then is this relationship so important in syntax compared to others and how can we capture this fact in our theories?

Derivational models of syntactic structure seek to build (or derive) the syntactic tree. Now, instead of having a structural frame and hooking elements onto the ends of branches, we start with the syntactic elements and aim to combine them, effectively building the structural frame as we go. How we know which elements to combine and when is another matter and one which I touched on in the entry below under the heading ‘Linguistics: Selfish Syntax’ (http://linguismstics.tumblr.com/post/16299533223/linguistics-selfish-syntax). What is important for this entry is the difference between a representational tree and a derivational tree.

In natural human language there are many instances of elements appearing in one position but being interpreted in another. For example, wh-movement in English.

What did you buy?

The element ‘what’ appears at the beginning of the sentence but all English speakers will interpret this as being the object of the verb ‘buy’ (English objects normally appear after the verb).

What did you buy (what)?

In many syntactic theories, and certainly in Chomskyan syntax, this is accounted for by saying that ‘what’ has undergone movement. Evidence for the bracketed ‘(what)’ being the original position is the idea that this is where it is interpreted and that English does permit ‘what’ to appear in this position in certain circumstances, e.g. “You bought WHAT?!”

However, what does it mean for an element to ‘move’ in a derivational model of syntax? In the Minimalist Program, elements undergo Merge (a (if not the only) structure building operation in syntax) to build hierarchical structure. To say something ‘moves’ is tantamount to saying that both of the elements that are undergoing Merge are already in the structure built so far, so movement is essentially an instance of Internal Merge (as opposed to External Merge when one of the elements involved is not already part of the structure). In earlier theories, movement involved a copy (or trace) being left behind in an element’s original position. In the wh-movement example above, the lower copy of ‘what’ is not phonologically realised but it is interpreted. However, in a model where movement is Internal Merge, ‘copy’ becomes a way of saying that one element is merged to multiple parts of the syntactic structure (as opposed to there being multiple copies of this one element, each copy occupying a unique place in the structure).

What we have now is a syntactic tree that is no longer nice and upright (and intuitively tree-like) but one which builds and bends, twists and turns, and whose separate branches sometimes join up again in vast conceptual loops as we establish structure via Merge between individual elements.