# How to convert parentheses notation for trees into an actual tree drawing?

Trees are usually drawn as a set of objects connected by edges.

But sometimes one sees a non-graphical, parentheses-based notation, like on the example below. What does the indentation mean in such context? How does this correspond to vertex-edge representation of a tree?

(ROOT
(S
(NP (NNP John))
(VP (VBD hit)
(NP (DT the) (NN ball)))
(. .)))


## 1 Answer

In this parenthesis notation for trees, $(P\;D_1\;D_2\;D_3)$ means that $P,D_1,D_2$, and $D_3$ are nodes, and $D_1,D_2$, and $D_3$ are daughters of the parent node $P$. This is an ordered tree, so when you draw the tree, the daughter nodes $D_1,D_2,D_3$ should appear in that order from left to right.

Now suppose that $D_1$ has daughters $A_1$ and $A_2$; then the subtree with $D_1$ as root is written $(D_1\;A_1\;A_2)$ and replaces $D_1$ in the description of the subtree with root $P$: $$(P\;(D_1\;A_1\;A_2)\;D_2\;D_3)\;.$$ The subtree itself looks like this:

$$\begin{array}{ccc} &&&&P\\ &&&\diagup&\mid&\diagdown\\ &&D_1&&D_2&&D_3\\ &\diagup&\mid\\ A_1&&A_2 \end{array}$$

Your tree could be written

$$\text{(ROOT (S (NP (NNP John)) (VP (VBD hit) (NP (DT the) (NN ball))) (. .)))}\;,$$

without any indentation; the indentation is there to make it easier to see which nodes are on which level. ROOT and S are the only nodes on their respective levels, and S is one level below ROOT; NP and VP are one level below S and NNP, VBD, and NP are one level deeper yet. The indenting is not entirely consistent deeper in the tree; if it were, you’d have something like this:

(ROOT
(S
(NP
(NNP John))
(VP
(VBD hit)
(NP
(DT the)
(NN ball)))
(. .)))


(In all versions the $(.\;.)$ represents possible other daughters of $S$ besides $NP$ and $VP$.)

Expanding it into the usual graphical representation of a tree gives you something like this: 