Home » Articles » LS13. NG’s tabula rasa

LS13. NG’s tabula rasa


This week a format is developed for analysing sentences using NG.  Processing is shown step by step, highlighting where meaning is delivered.  The approach is far better than traditional static representations of ‘structure’.  The only tree a syntactician really needs is one with fairy lights.

The format is applied to some variants of Nero is giving Olivia to Poppaea with the same predicate/argument relations.


The format must show the sequence of junctions in a sentence and allow each junction to create as many as three propositions:


Each proposition must contain three concepts.  One of these is the relation.  Let’s give this the general name REL.

The other two must be in a specific sequence. Remember, AAA / RELX / BBB is not the same as BBB / RELX / AAA.  They need to be unambiguously named.  But how?

Other writers borrow terminology from logic or mathematics.  For example, Richard Hudson uses ‘argument’ and ‘value’.  But such labels can be misleading.  As shown in LS9, the concepts need not be in any systematic relationship.

I’ve given them three-letter names in a clear alphabetic sequence – QUO, REL, SIC.  Pure whimsy.  QUO and SIC are arbitrary and intended to mean nothing.  If you can’t resist thinking of them as Latin … well, that’s nice but don’t think too hard.


And ACT?  That’s simply the activation on the proposition.

Sentence (5) tabulated

table 1

The reasoning behind these steps can be found in LanguidSlog 11 and 12.  Three new things here.  One is the shading: propositions are unshaded when delivered to cognition, i.e. when they have gathered a full set of concepts and full activation.  Another is using columns 1, 2 and 3 to build the agent, theme and goal propositions respectively: this is merely for clarity.


The other new thing is zero activation from some of the junctions.  For (5) the three propositions shown above only need 3xA in total.  If one occurrence of a phonological word brings activation once, it is handy to show that happening when the word makes its one appearance as a dependent.  But that would total 6xA – or perhaps 5xA as is is never a dependent.

It can be assumed that a verb does bring activation but it’s used for delivering a proposition (not shown here or in the analyses in the next several LanguidSlog posts) reflecting tense, aspect etc.

As for a preposition phrase, I arbitrarily assume it’s the preposition rather than its complement that brings A.  (Perhaps the noun also brings activation but it is superfluous.)  In (5) this neatly puts consolidation of THEME / OLIVIA on to a separate line in the table.

Missing argument

Sentence (7) lacks an argument to deliver GOAL:

(7) Nero is giving Olivia

At sentence-end (as in the fourth row in the table above) two incomplete propositions share the activation brought by Olivia.  But GIVE / THEME / OLIVIA must somehow inherit it all and be delivered.

This behaviour seems to be common to ditransitive verbs that allow both the to-dative form as in (5) and the double-object form, Nero is giving Poppaea Olivia.  Nonetheless there must be a mechanism to make it happen.

My guess is that it’s done by an asymmetric split of activation between the two incomplete propositions.  Then if not otherwise resolved at sentence-end, the activation on the weaker devolves on to the stronger.  Let’s assume A = six units of activation:

table 2

A hostage to fortune?  Yes: at some point I have to explain the acquisition of language knowledge for junctions like giving__Olivia.

Reversed arguments

In sentence (8) the sequence of the post-verb material in (5) is reversed – as would more typically occur when THEME is from a heavy phrase (such as …a partridge in a pear tree) following an active verb.

(8) Nero is giving to Poppaea Olivia

table 3

Arguably it is the already-complete GIVE / GOAL proposition, rather than sentence-end, that forces completion of GIVE / THEME / OLIVIA.  However the analyses for (5), (7) and (8) are consistent in their treatment of giving__Olivia.

Fronted to-phrase

Try using the method to analyse sentence (9).  Your analysis should also work for sentence (10) if the last couple of lines are omitted.

(9) To Poppaea Nero is giving Olivia

(10) To Poppaea Nero is giving

Here’s an empty table to print out (click to open PDF).

I’ll include my solution in LanguidSlog next time.

Hazards ahead

If you think this is all a bit glib, you’re right: the sentences analysed so far are untypically simple.  Fronting the to-phrase hints at what’s coming next: fronting the direct object as in Olivia Nero is giving to Poppaea.  We also need to look at the double-object form, Nero is giving Poppaea Olivia, and at passives, and then at the possibilities for fronting in DO and passive forms.  In many cases, determining a thematic role cannot be done until well after the noun phrase is encountered.

Mr Nice-Guy

One comment