Home » Articles » NG38. Coordination after the verb (2)

NG38. Coordination after the verb (2)

 This post picks up where NG34 left off.  The essential point is that sentences only need to be treated as multi-clause where subjects or verbs or direct object/indirect object pairings are coordinated.

Back to basics

Simple noun-plus-noun coordination, after the verb, might be thought of as sharing subject, verb and one or other object. That sort of multiple-clause approach is unwieldy for sentences where the coordinated nouns come at the end and a complete ‘first clause’ arguably exists:

(145) John gave apples to Mary Nancy and Odette

(146) John gave Mary apples bananas and cherries

Worse, the approach is nonsensical if the coordinated nouns immediately follow the verb:

(147) John gave apples bananas and cherries to Mary

(148) John gave Mary Nancy and Odette apples

In sentences (111), (112) and (113) from LS33, incremental processing provides few clues about the syntactic roles of nouns before the verb.  Coordination therefore depends crucially on junctions between and (or another appropriate conjunction) and a noun.  These junctions generate propositions that coexist to deliver complete coordinated argument.

In contrast, a noun immediately following a ditransitive verb is unmarked if direct-object or marked if indirect-object.  If direct-object, any following noun before the dative preposition must be coordinated with it.  If indirect-object, any following noun is coordinated with it if also marked or is direct-object if unmarked.  This allows a simpler approach to coordination after the verb.

Coordinated direct object

The analysis for (147) shows pairs of nouns delivering COORD propositions directly.

(147) John gave apples bananas and cherries to Mary

The string of unmarked nouns following the verb in (147) has to be the direct object and end-of-coordination is marked by to or by sentence-end.

Coordinated indirect object

What if the nouns are marked?

(148) John gave Mary Nancy and Odette apples

An Odette__apples junction would mix marked and unmarked and is therefore not available.

LS35 noted in the context of coordinated pairings of indirect-object plus direct-object after the verb that and could be optional.  This also applies to individual objects coordinated after the verb.  John gave apples bananas cherries to Mary and John gave Mary Nancy Odette apples both seem perfectly good.

An analysis for John gave Mary Nancy and Odette apples bananas and cherries should not need detailing.  Indeed it should be clear that the sentence will work with either or both of the ands omitted.

Into the workshop

Dative-only verbs do not require semantically-distinctive indirect objects.

(149) John fitted washers on rivets nuts and splitpins on bolts and screws …

In (149) it appears at first glance that the role of nuts and splitpins cannot be determined until the second on is encountered, while screws may be coordinated with indirect object bolts or may be a further direct object.  In double-object constructions, complexity from coordination can be managed incrementally using the marked/unmarked distinction.  In constructions with dative-only verbs no such lexical marking is available and it must be assumed that there is phonological marking – using prosody or punctuation – to identify syntactic roles.

Essentially the scheme is:

  • the first noun following the verb is a direct-object (DO)
  • a noun following a DO is a DO
  • a noun following a preposition is an indirect-object (IO)
  • a noun following an IO not phonologically marked is an IO
  • a noun following a phonologically-marked IO is a DO
  • DO and DO, IO and IO, IO and DO are all possible but DO and IO is not possible
  • and is not needed for distinguishing DO from IO
  • and signifies the type of coordination.

The only new idea necessary for the analysis of a sentence like (149) is the omission of any (noun1)__(noun2) junction where noun1 is phonologically marked.  In (149), rivets is a phonologically-marked IO.  If bolts is also marked, screws is a DO; but otherwise screws should be a marked (or sentence-final) IO.  Assuming screws is indeed sentence final:

(149) John fitted washers on rivets nuts and splitpins on bolts and screws

As with (122), it is assumed that the material forming the first complement is dropped when the activation-less frame is carried forward.  This leaves John fitted nuts and splitpins on bolts and screws, enabling fitted__nuts and avoiding rivets__nuts.

A change of scene

That’s six Network Grammar pieces dedicated to coordination – and there are still some loose-ends to tie up.  A big issue largely avoided is backtracking, the need for which would invalidate NG.  However a possible need for backtracking will be seen again when we look at noun phrases with qualifiers and determiners – NPs much more complicated than plain Nero.

But author and readers deserve a break from that sort of thing so next time let’s take an NG perspective on language acquisition.