I’ve written before about press reports that the Australian language Jingulu has only 3 verbs. A language with only 3 verbs? – Language Miscellany I’ve now found some discussion of that idea in Mark C Baker’s book Lexical categories: verbs, nouns, and adjectives.
In section 2.10 of his book, Baker discusses whether there exist any languages that have no verbs. He says Jingulu is the most serious candidate he has found for a language that has no verbs. Nevertheless, he concludes that Jingulu does have verbs (and not just 3 ‘light’ verbs.) So, he concludes that all languages have verbs.
In this post, I provide a reminder of how verbs work in Jingulu. I then discuss whether coverbal toots in Jingulu are verbal.
Verbs in Jingulu
As a reminder from my earlier post, Jingulu verbal structures have the following 3 components:
coverbal root – inflection – light verb
In example 1, the coverbal root is ngaba (‘hold), the inflection is nga (showing that the subject of the verb is 1st person singular: ‘I’) and the light verb is ju (do, be).
(1) ngaba-nga-ju karnarinymi
hold-1S-do spear
‘I have a spear’
Coverbal roots
Coverbal roots cannot appear by themselves. They can only appear:
- with a light verb;
- with a suffix turning them into a nominal or an adverb; or
- in some subordinate clauses interpreted as having the same verbal features as the main clause.
Baker comments that the coverbal root usually appears just before the inflection, though sometimes another word can come between them.
Inflection and light verbs
The inflection indicates the (grammatical) person and number of the subject and of the object.
Light verbs are verbs that are ‘semantically bleached’, containing little inherent meaning of their own. An English example of a light verb is take (in take a look): in this verbal phrase, the noun look contributes almost all of the meaning.
The light verbs in Jingulu are bound morphemes:
- they cannot appear separately. They always immediately follow the inflection. Phonologically, they behave like suffixes attached to the inflection.
- light verbs are usually accompanied by a coverbal root, but can occur without a coverbal root—for example, if a coverbal root was used recently and the context makes it clear that the light verb still carries the same meaning.
The form of the light verb indicates whether it is past, present or future. For example, the forms of the light verb meaning ‘do’ or ‘be’ are –ju (present), –nu (past) and –yi (future).
Are the coverbal roots verbs?
Like the paper I mentioned in my earlier post, Baker relies heavily on the description of the language in A grammar of Jingulu, an Aboriginal language of the Northern Territory, Rob Pensalfini, 2003, https://openresearch-repository.anu.edu.au/bitstream/1885/146163/1/PL-536.pdf
Pensalfini’s analysis
Baker summarises as follows how Pensalfini analyses the nature of light verbs and coverbal roots in Jingulu:
- The 3 light verbs are the only true verbs in Jingulu. Only they can mark the role (‘thematic role’) that nouns play in a verbal construction. That is why verbal constructions must always contain a light verb—and why nothing else is always required in verbal constructions.
(Nouns playing a role in a verbal construction are often called arguments—or nominal arguments—of the verb.) - The coverbal roots have no distinctively verbal features. They do not inflect and they do not mark the thematic role that nominal arguments play in a verbal construction. They are optionally adjoined to a clause to increase its semantic content without much affecting its syntax.
Why Baker rejected Pensalifini’s analysis
Baker gives the following reasons for not accepting Pensalfini’s analysis:
- some nouns are derived from coverbal roots in ways that imply the coverbal roots have their own nominal arguments
- what determines the number and flavour of the arguments is the coverbal root, not the light verb.
- if these roots are not verbal, they are (in Baker’s theory) adjectives. But they differ in several ways from items that are uncontroversially adjectives.
I review each of these points below.
Coverbal roots have their own arguments
Some nouns derived from coverbal roots refer to a nominal argument of the coverbal root. For instance, the derivational suffix -ajka converts a coverbal root into a noun playing the role of the coverbal root’s ‘patient’ or ‘theme’—the noun undergoing the activity expressed by the verb (example 2).
(2) darr-ajka
eat-NOML (theme)
‘what one eats, food’
Similarly, the suffix -ajkal creates a derived noun playing the role of the agent carrying out the activity expressed by the coverbal root (examples 3 and 4).
(3) nagany-ajkal-irni
sing-NOML (agent)-FEM
‘singer’
(4) ngirrm-ajkal-a murdika-rna
sing-NOML (agent)-MASC car-DAT
fixer of cars, mechanic’
Baker concludes that those nominal arguments are already present in the coverbal root, rather than being inserted during the derivation. He doesn’t provide any analysis supporting that argument, and I can’t see how he gets there.
The coverbal root sets the argument structure
In Jingulu, what determines the number and thematic nature of a verb’s arguments (the verb’s argument structure) is the coverbal root, not the light verb. For instance, in examples 1 (repeated below), 5 and 6, the coverbal root ngaba (‘hold’) requires an argument referring to the agent theme (patient), in this case karnarinymi (‘spear’). That is the case, whichever of the 3 light verbs accompanies the coverbal root.
(1) ngaba-nga-ju karnarinymi
hold-1S-do spear
‘I have a spear’
(5) ngaba-nga-rriyi karnarinymi
hold-1S-will.go spear
‘I will take a spear’
(6) ngaba-jiyimi karnarinymi
hold-come spear
‘he’s bringing a spear’
Baker suggests that it is best to analyse the 3 inflected verbs as thematically inert verbal auxiliaries, rather than as light verbs. In this respect, they are like English have (when used in forming the present tense), be (when used in forming the passive) and similar auxiliaries in many other Indo-European languages.
The only element of meaning they contribute is about direction: away from the reference point (‘go’), towards the reference point (‘come’) or unmarked for direction (‘do / ‘be’).
Despite Baker’s suggestion, I have kept the terms ‘light verb’ and coverbal root’ in this post, partly because I also used those terms in my earlier post on Jingulu.
Coverbal roots differ from adjectives
Coverbal roots differ in various ways from items that are, uncontroversially, regarded as adjectives. Unlike coverbal roots, adjectives:
- agree in gender with an associated noun;
- can form predicates without combining with a light verb;
- when used as predicates appear after (rather than before) the associated noun;
- can be used attributively, modifying nouns directly without being inserted in a relative clause
How coverbal roots differ from adjectives
Example 7 illustrates the first 3 of these properties.
(7) miring-mi bardakurru-mi
Gum-VEG good-VEG
‘gum is good’
In that example, the adjective bardakurru (‘good’):
- agrees in gender (in this case, ‘vegetable’ gender) with the associated noun (miring, ‘gum’);
- forms a predicate without needing to combine with a copula (like English is); and
- follows the associated noun.
Example 8 illustrates the 4th property mentioned above:
(8) Jami-na diman-a-rni laja-ardu ngamal-u lamb-u
that-MASC horse-MASC-ERG carry-go big-NEUT load-NEUT
‘that horse is carrying a big load’
In that example, the adjective ngamal (‘big’) is used attributively, modifying the associated noun lamb (‘load’) directly. It can do that without being inserted in, for example, a relative clause, such as ‘the load, which is big’.
What are coverbal roots, if not adjectives?
In his book, Baker’s book sketches out a theory that there are only 3 types of lexical word: verbs; nouns; and adjectives. In his theory, all other words are functional—they have little or no lexical meaning.
If coverbal roots are not verbs and, presumably, are not nouns, they can only be adjectives (in his theory). Baker acknowledges that coverbal roots could be analysed as a subset of adjectives. But he argues that this would gain nothing because coverbal roots share almost no properties with items that are uncontroversially adjectives.
As Baker admits, if people do not accept his theory that there are only 3 lexical word categories, they may not agree that coverbal roots must be verbs if they are not adjectives.
Baker’s definitions
Verb: a lexical category (ie a lexical word) that has a specifier.
Noun: a lexical category bearing a referential index.
(The index is an ordered pair of integers. One integer denotes that instance of the noun. The other integer denotes another identical noun.)
Adjective: a lexical category that is not a verb (has no specifier) and is not a noun (bears no referential index)
Summary
Baker concludes that all natural human languages have verbs (and, indeed, nouns and adjectives, though I haven’t discussed that in this post.)
In reaching that conclusion, he rejects claims that the only verbs in Jingulu are the 3 light verbs. He also concludes in passing that it is best to analyse the light verbs as auxiliaries, not as light verbs, because their lexical meaning is so minor.
The paper I mentioned in my earlier post was JSwarm: A Jingulu-Inspired Human-AI-Teaming Language for Context-Aware Swarm Guidance, by Hussein A. Abbass, Eleni Petraki and Robert Hunjet (2022). That paper doesn’t mention Baker’s book