Writers are constantly advised that items in a list (in coordination in running text, in a bulleted list, in each level of an outline) must be parallel in form, in particular must all have the same syntactic structure (some Language Log discussion here). This is a struggle for the typical writer, who is just trying to convey a series of ideas, each in whatever form seems most natural to the writer. So we get flagrantly non-parallel lists, as on this postcard from Boston:
The items in the list:
(1) was settled in 1630… [VP with omitted subject referring to Boston]
(2) there is archaelogical evident that suggests Native American presence dating back… [S in which presence is understood as 'presence in Boston']
(3) used to have a lot more hills… [VP as in (1)]
(4) nicknames include… [S with subject knicknames, understood as 'Boston's nicknames, nicknames for Boston']
(5) is home to more than… [VP as in (1)]
(6) famous people from Boston include… [S with explicit mention of Boston in the PP from Boston]
Each item has a reference, implicit or explicit, to Boston. Each is about Boston — the heading on the postcard is “Fun Facts about Boston”, after all — but some of the items are VPs, some Ss, and the references to Boston are managed in four different ways. The text is coherent, but flagrantly non-parallel.
Nevertheless, much of the advice literature insists on parallelism. Yet, as I noted in the Language Log posting linked to above,
the advice literature on parallelism exhibits all three of these problematic features: a fuzzy notion of parallelism (more generally, a failure to distinguish grammar, usage, and rhetoric), a seat-of-the-pants syntactic theory, and wildly overgeneralized prescriptions.
There seems to be a general feeling that parallelism is a good thing because it makes texts easier to understand. But formal parallelism is a means to creating coherent texts, and I believe that coherence (which can be achieved by means other than formal parallelism) is a crucial factor in language processing.
In some “simplified registers”, non-parallelism is common, even conventionalized. This is the case in handbook descriptions of plants and animals, which are designed to be as informative as possible in as few words as possible. Here, for example, is the beginning of the Walrus entry in the Fieldbook of Natural History (2nd ed., by Palmer & Fowler, 1975), p. 700:
Length, male 10 to 12 ft. Weight of male, average to 2,7000 lb. Female about 2/3 size of male. Almost hairless, with wrinkled skin. Head relatively small. Nose blunt, with course bristles. Practically tailless. Flat nails on five toes on front flippers. Two flat and three pointed nails on hind flippers.
All the sentences are verbless, and some are subjectless as well (though these all refer to the walrus); the subjectless sentences then have the form of AdjPs or NPs. For some of the sentences, a form of the verb be is to be supplied, for some, in particular, the last two, a form of the verb have. In any case, the text is flagrantly non-parallel, but highly coherent and easy to understand.
In this context, economy takes precedence over formal parallelism.