246 5 Noun phrase structure
'5' saemmos
'6' sadis (saedis)ad sasmmos ad saemmos-aet'7' aessa
'8' aettam
'9' taezza ι ι
'10' mserawad sadis ι
ad aessa
ad aettam
ad taezza ι <
ad maerawad sadls-aet ι
ad aessdy-ast
ad asttam-ast
ad taezzdy-aet
ad masraw-aetThe trailing forms begin with Comitative preposition d (ad) 'with, and'
(§6.4.2). The trailing forms are added to other numerals denoting ten-somes or
larger units, e.g. 'ten and five' = '15'. See the examples 'fifteen men' and
'fifteen women' in §5.1.2.4, below.
The d- in d-ly-aen Ί', d-assin '2', and variant d-aekkoz '4' may also have
originated as the same Comitative preposition ('and' would make sense for
numerals above Ί', as in a counting sequence Ί, and 2, and 3, ...'), but is now
a frozen part of certain V-initial numerals and co-occurs with ad in the relevant
trailing forms.
At least in the R dialect, the combination ad d- seems to have been
reinterpreted as a unit prefix in the trailing forms. It then extends to all cases
where the numeral stem proper begins with a V: masculine add aessa '7', add
asttam '8', feminine add aekkoz-aet '4', add aessay-aet '7', add aettam-aet '8'.
One could alternatively segment as e.g. ad d-aettam where the d- is a
prevocalic extension of the numeral; this would "preserve" the identity of ad
as the Comitative preposition.
For many dialects (including T-ka, T-md, R), the counting and masculine
trailing forms are unaccented. We get default accents in the counting forms,
and phrasal accent on the preposition ad in the masculine trailing forms.
However, the A-grm, K, and I speakers gave counting and masculine trailing
forms for '2' to '10' with accent on the second syllable, e.g. K-d d-assin '2',
kaerdd '3'. Note also A-grm mseraw ad assln 'twelve-Ma'.
Even the T-ka and R speakers accented the bare numerals '2' to '10' when
used as NPs in a sentence, in the absence of the implied referent nouns. Thus
anhdy-aer maerdw Ί have seen ten', anhdy-asr assin Ί have seen two',
ajlae-n maerdw 'ten went'. If the referent nouns were present, the numeral
would have the same accent, so one way to think of these examples is as
truncations of fuller forms like Ί have seen ten men'.
For '4', '7', and '8' there is dialectal variation between as and a in the first
syllable. T-ka and some other Timbuktu-area dialects have ae, but I recorded a
for A-grm and in some other Timbuktu-area dialects. In the cases of '7' and
*8', where the second syllable has a clear a, this variation may reflect the usual
T-ka penchant for Short-V Harmony (§3.2.6) within these numeral stems. In
'4' this is moot since the phonetic [o] before ζ (a BLC, §3.1.2.2) in the second
syllable could be taken as lol or /u/.
Whether melodic harmony is at work in numerals is also a factor in the
lexical representation of '6'. The stem is pronounced [sae'dis] in T-ka and
some other dialects. Because of the BLC (d), a melodically harmonic