I chose the second one.
All sentences in all languages can be represented by constituent struc¬ture trees, and all have syntactic rules that determine the linear order of words and their hierarchical structure.
The first one is not suitable. Such divisions sound most unnatural to English native speakers according to their linguistic knowledge. Intuitively, English speakers divide these sentences in the way that obviously follows a particular hierarchical order, as is shown in the second one.
This type of segmentation truthfully reveals the hierarchical nature of sentence structure: sentences are organized with words of the same syntactic category, such as noun phrase (NP) or verb phrase (VP), grouped together.