Home » Articles » NG49. Nesting relative clauses

NG49. Nesting relative clauses

NestWe continue looking at the sentences:

(179) The claim was withdrawn

(180) The claim that the moon is a pancake was withdrawn

(181) The claim that the claim that the moon is a pancake was withdrawn is inaccurate

(182) * The claim that the claim that the moon is a pancake was withdrawn

(183) * The claim that the claim that the moon is a pancake was withdrawn is inaccurate is preposterous

How is a subject noun phrase linked to the right verb phrase?  How are the resulting clauses linked together?

Joining relative pronouns

NG48 cheated by assuming that the subordinating junction is that__(subject noun of lower clause).  That was convenient because the two words are nearly adjacent and certainly not separated by a further clause.

The that has already acted as dependent in a junction with the preceding subject noun of the higher clause.  Therefore that must act as parent in a junction with a word in the lower clause.  This must be the verb because everything else has already acted as dependent within the clause.

The situation is similar to what we saw in NG44 where the head of a complete noun phrase links back to a verb.  Now we have the head of a complete clause linking back to a relative pronoun.  In both cases the earlier word is parent in the resulting junction.

Junctions short and long

Here’s sentence (181) again with some subscripts and is in place of was to help the following discussion:

(181a) The claim1 that2 the claim2 that3 the moon is3 a pancake is2 withdrawn is1 inaccurate

The junctions that NG must identify include:








These are the ones on which the clause hierarchy depends.  Other junctions need no comment at this stage, and some of those listed are local and straightforward.  The junctions needing an explanation are those spanning several other words and including the repeating words – claim, that, is.

Linking verbs back to subjects

Junctions between a verb and a preceding noun are not difficult to explain.  The lexicon allows most (noun)__(verb) possibilities, including all those where the verb is a be form.  However the noun cannot have already acted as dependent.

Work thorough the sequence in which junctions are processed in (181).  You’ll see that each verb picks up the correct subject because any intervening noun has already acted as dependent and is therefore disallowed.

Linking verbs back to relative pronouns

Linking back to a relative pronoun is not so easily explained.  In this type of junction the verb is dependent and the pronoun is parent.  Generally a word can be parent more than once in a sentence.  In terms of roles-in-junctions, that3__is3 having already occurred shouldn’t prevent that3__is2 – which would be wrong.

The key to solving this problem is something that has appeared in LanguidSlog many times but has never been fully discussed.  In a sentence with Nero gave… an NG analysis would show three propositions being created at gave:


GIVE / THEME / (null) / 0

GIVE / GOAL / (null) / 0

The term valency is often used in linguistics to mean the number of syntactic elements required by a verb, give having a valency of three.  Valents may be optional (as seen in some give sentences in LanguidSlog) or obligatory.

Words other than verbs may have valency.  Relative pronouns may be a good example, that being obligatorily followed by a clause.  So, in the same way that Nero__gave creates propositions in anticipation of upcoming valents, claim__that creates a proposition CLAIM / QUALIF / (null) / 0 in anticipation of a subordinate clause.

Yes, arguably it is the verbal noun claim that has the valency – as would be shown by omitting that in the ‘reduced’ form.  However that forms the crucial junction with the verb in the subordinate clause.  Without the subordinate clause, the junction claim__that wouldn’t satisfy the valency of claim.

In view of this we should really use COMPLEMENTED BY instead of QUALIFIED BY in the CLAIM propositions.  We’ll stick with QUALIF as in LS48 because it might be confusing to change at this point.

By the time that3 in (181) has been processed, two anticipatory propositions will have been created:

CLAIM′ / QUALIF / (null) / 0

CLAIM″ / QUALIF / (null) / 0

Use of prime and double-prime to distinguish the CLAIM concepts is for consistency with LS48.

When is3 is processed, the right-to-left scan of the word chain (see LS41) finds that3 giving a junction that creates:

(null) / QUALIF / IS3 / 6

What does the SIC concept represent?  This is discussed below.  Meantime is3 is clear enough.

The ‘complementary nulls’ principle allows this to combine with one of the anticipatory propositions – the more recent one apparently.  This leads to delivery of:


When is2 is processed, it should logically pick up that2 but that3 is found first.  It doesn’t matter because both have the same effect:

(null) / QUALIF / IS2 / 6

Only the first of the anticipatory propositions is left, so ‘complementary nulls’ delivers:


Claiming the correct sub-clause

We’ve unravelled the long-distance junctions but does a proposition in the form CLAIM / QUALIFIED BY / IS actually mean something?  Not really: its value is that cognition gets a connection from a particular instance of claim to the meaning of an entire subordinate clause via a particular instance of is.

The inference is that, in the same way a noun like claim creates a distinguishable instance from the permanent concept (see LS48), a verb must also create a distinguishable instance.

Ungrammatical variants

Sentence (182) has too few verbs.  The first claim__that junction creates CLAIM′ / QUALIF / (null) / 0 but this is left incomplete at sentence-end.

Sentence (183) has too many verbs.  The final instance of is has no subject noun with which to form a junction.  The meaning of the sentence before that point could be delivered successfully but the final disconnected fragment must be disconcerting to the hearer.

On, on!

I hope the discussion of these sentences hasn’t been too superficial.  Only eight more posts to go and still a lot of ground to cover.