Home » Articles » NG53. Attaching simple adverbs (3)

NG53. Attaching simple adverbs (3)

DeeplyDifferences in AGENT attitude may be expressed where the relevant nouns are coordinated.  This piece discusses what is possible.

Also yesterday and probably, which behave differently from each other and from manner adverbs, are briefly discussed.

Mixing adverbs and nouns

LS52 finished with:

(214) * Olivia gladly Nero gave to Poppaea

(215) * Gladly Olivia Nero gave to Poppaea

It might be thought that the hearer senses ungrammaticality in (214) and (215) at Nero rather than at gave.  However the first three words in (215) are fine in (216).

(216) Gladly Olivia Nero and Poppaea gave alms to the poor

As for (214), the first three words occur grammatically in (217).

(217) Olivia gladly Nero madly and Poppaea sadly gave…

Sharing an adverb between nouns

Coordination of the three nouns is not a problem as is shown in LS33.  There are some points about attachment of the adverbs that need to be explained.  A grammatical variant of (216) has gladly at the other end of the noun sequence and should deliver the same propositions to cognition.

(218) Olivia Nero and Poppaea gladly gave…

However any variant of (216) with gladly moved to an intermediate position is ungrammatical.

The propositions for (218) are the most straightforward.  At and__Poppaea:

At Poppaea__gladly:

At Poppaea__gave:

The propositions for (216) evolves differently but the effect is the same.  At gladly__Olivia:

At and__Poppaea:

At Poppaea__gave:

At first glance, ungrammatical (219) is problematic because it ought to give the same structure as (216).

(219) * Olivia gladly Nero and Poppaea gave…

However, the problem disappears if Olivia and Nero get their C categories depending on whether or not they have an adverb dependent.  Different Cs would disallow a COORD proposition linking them.  This will seem odd if it is assumed that a junction between Olivia and Nero is processed immediately after Olivia__gladly.  But the analysis for (112) in LS33 shows that a COORD proposition would be the result of Olivia__and and Nero__and.  In (219), Nero__and is processed without Nero having acquired the same ‘qualified by following adverb’ C that Olivia has.

The (null) / AGENT / OLIVIA proposition may look superfluous; but it ensures that the null shared with (null) / OLIVIA / GLAD is correctly replaced by GIVE; and it disappears at sentence-end because of its zero activation.

In (217), three (null) / AGENT / (noun) propositions are formed and each noun acquires a ‘qualified by following adverb’ C.  The propositions are even stranger and too difficult to draw, but still consistent with the assumptions.  The difficulty of drawing this in two dimensions should not cast doubt on validity: in geometry, a solid (tetrahedron) can be formed by joining as few as four triangles.

Adverb per noun

Sentence (220), a variant of (217) with the adverbs before the nouns they qualify, would create the same structure.

(220) Gladly Olivia madly Nero and sadly Poppaea gave…

Empirically a series of coordinated nouns each with a qualifying adverb either has noun-then-adverb throughout or adverb-then-noun throughout, not a mixture.  The first pairing – in (220), gladly__Olivia – must set a pattern that subsequent pairings follow.  This can be done by giving a noun in (manner adverb)__(noun) and in (noun)__(manner adverb) distinct categories, Cx and Cy.  Once the first noun has acquired Cx, it cannot participate in a rule-of-combination that has Cy.  In (220), Olivia__madly is not recognised.  Therefore Nero must also acquire Cx from madly__Nero, and so on.

Sentence (217) has adverbs following nouns and the logic is slightly different.  In many situations, the lefthand word in a junction must use a C already acquired from an earlier junction in which it was the righthand word.  In (217), Nero acquires the same C as Olivia by linking their concepts in a COORD proposition.  That C allows a junction with following madly and all is well.

To allow only grammatical structures, there must also be a C for ‘not qualified by adverb’.  Possibilities for ungrammatical structures are endless and are not explored here.  More important is to counter any impression that NG is going into contortions to account for empirical facts.  A better way to view this is that NG has the ability to hold multiple lexical-entries for the same phonological word and meaning but with different Cs. This allows a wide range of possibilities, such as (221), that comfortably meets all the language user’s needs.

(221) Olivia and Nero gladly and Poppaea and Quintia sadly gave…

Arguably, (222) should have the same meaning but NG cannot decode it.

(222) * Olivia and Nero gladly and sadly Poppaea and Quintia gave…

What would happen is that sadly is coordinated with gladly and then Poppaea cannot be coordinated with the preceding nouns because it has the ‘not qualified’ C.

Behaviour of temporal adverbs

It has been shown that gladly is a dependent, before the verb, of the subject noun and, after the verb, of an object noun.  Exceptionally, gladly is the dependent of the verb if the two are adjacent.  Whilst gladly__(noun) cannot apply after the verb because (verb)__gladly must take precedence, (noun)__gladly takes precedence over (verb)__gladly.

What about yesterday?  That the behaviour of temporals is fundamentally like that of manner adverbs is shown by their occurrence with coordinated nouns.

(223) Yesterday Nero and today Poppaea gave…

(224) Nero gave Olivia yesterday and Renata today…

But manner adverbs are more restricted because of their attitude-of-AGENT semantics.  (205), (206) and (212) in LS52 show that yesterday can form dependencies with be forms, whether auxiliary or copular, where gladly cannot.  And while (214) and (215) are ungrammatical, variants with yesterday are not.

(225) Olivia yesterday Nero gave to Poppaea

(226) Yesterday Olivia Nero gave to Poppaea

In (225) and (226), only (null) / OLIVIA / YESTERDAY is created, not (null) / AGENT / OLIVIA, and there is no problem when Nero__gave is processed.  This contrasts with (214) and (215), which create both (null) / OLIVIA / GLADLY and (null) / AGENT / OLIVIA.

Qualification-by-adverb of coordinated nouns poses another question.  Why is the hearer uncomfortable with mixing temporal and manner adverbs?

(227) ? Yesterday Nero and gladly Poppaea gave…

An NG analysis can easily show this to be ungrammatical by proposing a finer distinction of C categories.  Putting a ‘temporal before’ C on Nero would require the same C on Poppaea, disallowing ‘manner before’, ‘temporal after’ etc.  Alternatively, (227) may simply be disconcerting to cognition.  But the meaning is not impossible, so rejection in automatic sentence-processing seems more likely.

Behaviour of probably

Another sort of behaviour is shown by probably.

(228) Probably Nero gave Olivia to Poppaea

(229) Nero probably gave Olivia to Poppaea

(230) Nero gave Olivia probably to Poppaea

(231) Probably Olivia was given to Poppaea by Nero

(232) Olivia was probably given to Poppaea by Nero

(233) Olivia was given probably to Poppaea by Nero

(234) Olivia was given to Poppaea probably by Nero

(235) Probably Poppaea was given Olivia by Nero

(236) Poppaea was probably given Olivia by Nero

(237) Poppaea was given probably Olivia by Nero

(238) Poppaea was given Olivia probably by Nero

These examples show that the position of probably focuses the slight uncertainty it conveys on AGENT or THEME or GOAL (the person may have been someone else) or on the verb itself (the act was possibly selling, not giving) rather than on to the entire sentence.  A simple rule seems to apply: probably is the dependent of what immediately follows, associating the idea of a certainty-deficit with that parent.

This rule is quite different from what was developed above for gladly.  A consequence of allowing probably__(noun) is that it should be possible in places where gladly__(noun) is impossible.

(239) Nero gave probably Olivia to Poppaea

(240) Nero gave probably Poppaea Olivia

(241) Nero gave Poppaea probably Olivia

(242) Poppaea was given probably Olivia by Nero

These sentences are unusual but, because probably has a stronger semantic association with a following noun than with a preceding verb or remote AGENT, they do seem better than their equivalents with gladly.  To analyse them as ungrammatical would have probably__(noun) behaving differently in pre- and post-verb positions.  That would be impossible because, in either position, the junction between this noun and the verb has not yet been processed.  To let us pursue other matters they are assumed without further discussion to be grammatical.

With sentence-final probably, the certainty-deficit could apply to the whole sentence (perhaps with some irony) or to the last noun.

(243) Nero gave Olivia to Poppaea probably

So far, probably has always preceded its parent.  A gave__probably or (noun)__probably junction would upset most of the analyses suggested above.  It is however possible for probably to qualify the preceding word but this requires the adverb to be distinguished prosodically.  For example, in Nero probably gave, the adverb is comprehended as qualifying Nero if there is some separation between probably and gave.  In (243), sentence-end after probably must have the same effect, making it qualify Poppaea.  With some separation between Poppaea and probably, as well as sentence-end, there is a third pattern which can distinguish probably so as to select the gave__probably junction.

Similarly, separation after sentence-initial probably should give a whole-sentence interpretation (without irony) to (228), (233) and (239).


This is the last of three over-long pieces on simple adverbs.  Next we’ll look at attachment of preposition phrases, as adjuncts and as valents.