WALS lists a language called Jamsay as having a feature called double-headed relative clause, but the site does not define what this means, and I can’t find a definition anywhere.

Does it mean it marks relative clauses with particles of some sort on both ends of the clause, or does it mean something else?

1 Answer 1


This description appears to be taken from Jeffrey Heath’s A Grammar of Jamsay, which starts section 14 Relativization of with this preamble:

The basic relative clause pattern has an internal head NP (marked by tone-dropping), and a participle agreeing in gender-number features with this head NP (rather than an inflected verb agreeing in all pronominal features with the clause subject). […] The participle itself therefore behaves somewhat like an adjective modifying the head noun in several respects […].

It is also possible to expand this core relative clause structure […] by adding a copy of the head noun (not the full head NP), as a special kind of external head. Specifically, this copied noun is “possessed” by the relative-clause proper, which functions here as the possessor NP […]. The complete construction may be suggested (crudely) by a structural paraphrase, whereby ‘a/the dog that I saw’ is expanded as as ‘dog of [a/the dog that I saw]’, though of course the linear order in Jamsay is very different (with the external ‘dog’ at the end). The copied external head noun is optional, and occurs in fewer than half of relative clauses occurring in my texts, but it is nonetheless quite well-attested. […]

Thus the maximal relative-clause construction might be described as double-headed. However, the Jamsay construction (when examined in detail) is rather unique, and I question how useful it would be to assign it to a general “double-headed” typological category.

So rather than being particles marking the relative clause at both ends, it is that there are two co-indexed slots for the head of the relative clause: one inside the clause (the internal head), and one after the clause that ‘possesses’ the relative clause and is marked with the possessive particle .

Jamsay has two tone levels: H[igh] and L[ow]. All stems have a lexical tone contour that includes H somewhere – it can be H, HL, LH or LHL – but in the internal head in a relative clause, that lexical tone is changed to L, which doesn’t otherwise exist as a stem contour and is thus always distinctive. This is called tone-lowering.

The internal head is not extracted, however, meaning that it is not moved to a particular place in the clause in the way that relative pronouns (or prepositional phrases containing relative pronouns) in English are always moved to the front of the clause, but remains in its underlying position within the clause.

That means the only thing that marks a relative clause as being a relative clause can be that the internal head is tone-lowered – and of course the fact that it’s embedded in a containing clause. Having an explicit external head may be one way to emphasise the relative nature of the preceding clause.

For example, using the clause ‘the dog I saw’ used in the quote above, the structure would be ‘dogᴸ I saw dog’ (where ‘dogᴸ’ means ‘dog’ with tone-lowering). An example from Heath (481) is ‘the position where you can stand’:

[ ìjɛ̀         è        íjɛ́     bèrε̂ː  ]   mà     ìjέ
[ positionᴸ   you.pl   stand   can.pl ]   poss   position

As cited above, Heath states that he thinks this feature is of limited value as a general typological category, and he may be right, but WALS have nevertheless done just that and made it a category. Their data should always be taken with a pinch of salt, but it is interesting that the languages they have marked as exhibiting this feature cluster together geographically, with two in central-western Africa (Jamsay and Mina) and three in Papua New Guinea (Kobon, Kombai and Yagaria).


Another example of double-headedness (sort of)

I’m not familiar with any other languages that repeat the head noun in the way it works in Jamsay, but a similar feature that could perhaps be subsumed under ‘double-headedness’ is resumptive/pleonastic pronouns in indirect relative clauses, which is found at least in Irish (and I assume also in Scottish and Manx). Irish relative clauses come in two shapes: direct and indirect.

In a direct clause, the head is the subject or direct object in the clause and is only indexed once, in the relativiser, which is always moved to clause-initial position. This is exactly the same as how it works in English, except that Irish is VSO, so when the relativiser moves to the front, we get SV[O] or OSV in English, but SV[O] or OVS in Irish: ‘the man [who married her]‘ ~ ‘an fear [a phós í]’ vs ‘the man [whom she married]’ ~ ‘an fear [a phós sí]’. Direct relative clauses use direct relativisers and independent verb forms.

In indirect clauses, the head is indexed twice, both in the relativiser and in another overt element which appears in the head’s underlying, pre-movement position. This is different from English, which leaves only a gap when moving the relativiser out of its underlying slot (e.g., ‘the manᵢ [whomᵢ she gives the book to __ᵢ]’). Indirect clauses use indirect relativisers and dependent verb forms.

This double indexing is reminiscent of the double-headedness seen in Jamsay, except that in Irish, the ‘repeated’ element, termed resumptive or pleonastic, is pronominal, agreeing with the head noun (which only appears outside the relative clause), whereas in Jamsay, it is the actual head noun itself. Not unexpected, really, since in Irish relative clauses come after the external head that they modify, making the resumptive anaphoric; while in Jamsay they come before it, making the internal head (had it been a pronoun) cataphoric (typologically much rarer).

Most commonly, the resumptive pronoun is the object of a preposition:

an    fear   aᵢ        thugann       an    leabhar   [di]
the   man    REL.dir   gives.indep   the   book      [to-her]
‘The man who gives the book [to her]’

an    fear   aᵢ          dtugann     sí      an    leabhar   dóᵢ
the   man    REL.indir   gives.dep   she     the   book      to-him
‘The man she gives the book to’

Relativisers do not distinguish subject and object forms, and pronouns only do so in 2sg (optionally), 3sg and 3pl. Since the movement rules result in SVO and OVS structures in direct relative clauses (see above), these are ambiguous if the element after the verb is one of the non-distinguishing pronouns:

an    madra   a         chonaic     mé
the   dog     REL.dir   saw.indep   1sg.subj/obj
‘The dog that I saw’ // ‘The dog that saw me’

A way to disambiguate is to use an indirect relative clause, repeating the indirect object as a standalone pronoun. This structure seems particularly close to Jamsay’s double-headedness to me:

an    madra   aᵢ         bhfaca    mé    éᵢ
the   dog     REL.indir  saw.dep   1sg   3sg.obj
‘The dog that I saw’

an    madra   aᵢ         bhfaca    sé         méᵢ
the   dog     REL.indir  saw.dep   3sm.subj   1sg
‘The dog that saw me’


Disclaimer: I’d never heard of Jamsay before this question and know nothing of the language. The Jamsay part of this answer is the result of a quick Google search, so feel free to correct anything I may have misunderstood.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.