Pernilla Hallonsten Halling

The root level

The root level is the smallest domain at which the encoding of attr, pred, and adv is analyzed. Recall from chapter 4 that root is defined as a form that cannot be divided into smaller morphological forms. Individual roots that occur in more than one of the three functions are identified. In (6.1), an English example from the previous chapter is repeated to illustrate this.

(6.1) English [attr pred adv] root

1. 
   The sad person over there is Sheila. attr
   
2. 
   Sheila is sad. pred
   
3. 
   Sheila smiled sad-ly. adv
   

In (6.1), the same root sad is used in all three functions, meaning that a total overlap on the root level is attested in English. The following subsections are devoted to a description of the three encoding overlaps at the root level: that of attr and pred in section 6.2.1,

that of pred and adv in section 6.2.2, and that of all three functions in section 6.2.3.
Note that an overlap of attr and adv has only been marginally attested on the root level in the case of Imonda (Border) which has intensifiers used in attr that also attach to Verbs in adv (see (83) and (84) in appendix B).

1. [attr pred] root overlap

Over a third of the sample languages (23/60) show an encoding overlap of attr and pred at the root level. One of these is Alamblak (Sepik), which has a closed Adjective class with around 45 members. In attr, Adjectives may either come before or after the head of the NP within which they modify, with some shifts in emphasis, as illustrated in (6.2). A clitic indicating person, number, and gender attaches to the last element of the NP, which may or may not be the Adjective.

(6.2) Alamblak (Sepik) (Bruce 1984: 90) [attr pred]

1. 
   1nd
   

dem
bro

big

fëh=r
pig=3sg.m
attr

‘the big pig’

2. 
   1nd
   

dem
fëh
pig
bro-r
big=3sg.m
attr

‘the big pig’

In the pred function, the copula -e attaches to the Adjective.

(6.3) Alamblak (Sepik) (Bruce 1984: 181) [attr pred]

yima-r
man=3sg.m
bro-e=r
big-cop=3sg.m
pred

‘The man is big.’

As can been seen when comparing (6.2) and (6.3), the same root is used, although it carries the enclitic in attr when it occurs as the last element of the NP, and has a copula attached to it in pred.

On a much smaller scale, Yagua (Peba-Yagua) has only two items that can be classified as Adjectives (Payne & Payne 1990: 416). These are já˛á˛mu ‘big’ and pasiiy ‘little’, and at least já˛á˛mu ‘big’ occurs in both attr and pred (1990: 416).
(6.4) Yagua (Peba-Yagua) (Payne & Payne 1990: 416, 96)

1. 
   já˛á˛mu
   

big

rooriy
house
attr

‘big house’

2. 
   Néé
   

neg
ja˛a˛mu-qu11-nu-qu11
big-long-clf.an.sg-long
ráy:
1sg
ja˛a˛mu-qu11-nu-qu11
big-long-clf.an.sg-long
j11ta
cli

Tomasa.
Tom
‘I am not tall; Tom is tall.’
pred

All languages with an overlap of attr and pred on the root level are listed in table 6.1, and plotted on the map in figure .

Table 6.1. Languages with [attr pred] root overlap

Alamblak (Sepik) Maltese (Afro-Asiatic)
Bambara (Mande) Mam (Mayan)
Bukiyip (Nuclear Toricelli) Mapudungun (Auracanian) Gooniyandi (Bunaban) Marathi (Indo-European)
Guaraní (Tupian) Mian (Nuclear Trans New Guinean)
Hup (Nadahup) Nishnaabemwin (Algic)
Jamsay (Dogon) Paumarí (Arawan) Jamul Tiipay (Cochimi-Yuman) Sango (Atlantic-Congo) Kalaallisut (Eskimo-Aleut) Yagua (Peba-Yagua)

Kambera (Austronesian) Yankunytjatjara (Pama-Nyungan) Kham (Sino-Tibetan) Yimas (Lower Sepik-Ramu) Kilivila (Austronesian)

overlap no overlap

Figure 6.1. Languages with [attr pred] root overlap

2. [pred adv] root overlap

Five sample languages have a root overlap of pred and adv. One of these is Jamsay
(Dogon), for which it was illustrated in section 5.3.4 that its ideophonic Adverbs can be used both in pred and adv (see example 5.19). Another is Waiwai (Cariban), which has simple adverbs that can be used both in pred, with a copula, and in adv. This also implies a lexeme overlap of pred and adv (cf. section 6.4.2).
(6.5) Waiwai (Cariban) (Hawkins 1998: 126, 89)

1. 
   Ehcopo
   

unequal
Ø-xakne
3.sbj-be
noro
3
y-apo-rî.
gen-arm-poss
pred

‘His arms were unequal (in length).’

2. 
   Kayka,
   

let’s.go
yarî
in.your.direction
k-mokyasî
1sbj-come
yamoro.

slowly

adv

‘Let’s go, I will come your way slowly’
Note that although different property words are used in (6.5a) and (6.5b), they belong to

the same class of simple Adverbs. All languages with a root overlap of are listed in table 6.2.

Table 6.2. Languages with [pred adv] root overlap

pred and adv

Jamsay (Dogon)
Mian (Nuclear Trans New Guinean) Waiwai (Cariban)

Yankunytjatjara (Pama-Nuyngan) Yimas (Lower Sepik-Ramu)

Four of the languages with a root overlap of
pred
and adv have the corresponding

overlap with the same roots on the level of the lexeme (see section 6.4.2).

3. [attr pred adv] root overlap

An overwhelming majority of the sample languages (50/60) display an overlap of attr, pred, and adv at the root level. This is the most common overlap for the root level. Swahili (Atlantic-Congo) is one such instance. Since it does not really make sense to separate Nouns from Adjectives here, property words will be labeled Nouns in the present description. Nouns take class markers, which are identical for the property-denoting Nouns in attr and pred.
(6.6) Swahili (Atlantic-Congo) (Myachina 1981: 64–65)

1. 
   m-sichana
   

i-girl
m-zuri
i-pretty
attr

‘a pretty girl’

2. 
   m-toto
   

i-child
wako
your
ni
cop
m-zuri

i-beautiful

sana
int
pred

‘Your child is very beautiful.’
The same Nouns can be used in adv, but then they require a different class marker than the one they take in attr and pred (but see exception in appendix B). In the example below, class marker I m- is thus exhanged for class marker VIII vi-.
(6.7) Swahili (Atlantic-Congo) (Marilena Thanassoula, p.c.)

A-li-soma
3sg-pst-read ‘She read well.’
vi-zuri.

8. good

adv

Since the class marker is exchanged in adv, Swahili has a clear root overlap of all three functions.

Another instance of an overlap of attr, pred, and adv is found in Imbabura Quechua (Quechuan). Here, property words are attested among Nouns. In attr, they are used in simple form, but in pred, they require the Copula ka-, except in present tense third person, where the Copula may be omitted. Here, a ‘validator’ -mi is also required (Cole 1985: 67). In adv, Adverbs are derived from Nouns with the suffix -ta, which is the accusative marker, although it can also mean ‘through’ or ‘via’ (1985: 186, cf. table 5.5 in section 5.3.1).

(6.8) Imbabura Quechua (Quechuan) (Cole 1985: 77, 67, 162, 186)

1. 
   yuraj
   

white

wasi
house
attr

‘a/the white house’

2. 
   ñuka
   

my
wasi-ka
house-top
yuraj-mi
white-val
ka-rka
be-pst.3
pred

‘My house was white.’

3. 
   maymi
   

very
sumaj-mi
pretty-val
ka-ngui
be-2
pred

‘You are very pretty.’

4. 
   tayta-ka
   

father-top
sumaj-ta

beautiful-acc/aDv

trabaja-rka
work-pst.3
adv

‘Father worked well.’


Examples (6.8a) and (6.8b) show the same root yuraj ‘white’ in attr and pred respec- tively, while (6.8c) and (6.8d) show the same root sumaj ‘beautiful; well’ in pred and adv, respectively. The examples belong to the same type of property words whose roots occur in all three functions, although semantically yuraj ‘white’ is hardly expected in adv. For sumaj ‘beautiful; well’, a certain semantic shift also occurs from use in pred to adv, but this is fairly subtle (for a discussion of such semantic shift, see section 8.8). All languages with an overlap of attr, pred, and adv are listed in table 6.3. It should be noted that the language-specific patterns vary greatly in terms of what language- specific class the root in question belongs to, as well as the productivity of the overlap. Accordingly, a certain language may have a root overlap of all functions filled by, e.g., a subset of stative verbs, but still have a class of simple adverbs as the major encoding strategy in adv. It should be kept in mind that the overlap below is most often not the only encoding possibility for attr, pred, and adv. No claim is made here in terms of whether the encoding pattern in table 6.3 is more productive than another in attr,
pred, and adv for any specific language.

overlap no overlap

Figure 6.2. Languages with [attr pred adv] root overlap

Table 6.3. Languages with [attr pred adv] root overlap
Abau (Sepik) Koyra Chiini (Songhay) Abkhaz (Northwest Caucasian) Krongo (Kadugli-Krongo) Acoli (Nilotic) Lahu (Sino-Tibetan)
Ainu (isolate) Lakota (Siouan)
Basque (isolate) Lezgian (Nakh-Daghestanian) Bininj Gun-Wok (Gunwinyguan) Ma’di (Central Sudanic)
Bora (Witotoan) Maltese (Afro-Asiatic)
Cavineña (Tacanan) Mapudungun (Auracanian)
Cherokee (Iroquoian) Marathi (Indo-European)
Cuicatec (Otomanguean) Mian (Nuclear Trans New Guinea) Dutch (Indo-European) Nama (Khoe-Kwadi)
Estonian (Uralic) Nuu-chah-nulth (Wakashan)
Ewe (Atlantic-Congo) Paumarí (Arawan)
Georgian (Kartvelian) Pirahã
Hdi (Afro-Asiatic) Sahaptin (Sahaptian)
Hup (Nadahup) Sango (Atlantic-Congo) Imbabura Quechua (Quechuan) Slave (Athapaskan-Eyak-Tlingit) Imonda (Border) Swahili (Atlantic-Congo)
Jamul Tiipay (Cochimi-Yuman) Tagalog (Austronesian) Kalaallisut (Eskimo-Aleut) Turkish (Turkic) Kambera (Austronesian) Urarina (isolate)
Kewa (Nuclear Trans New Guinea) Warekena (Arawakan) Kilivila (Austronesian) Yagua (Peba-Yagua)

Kiowa (Kiowa-Tanoan) Yankunytjatjara (Pama-Nyungan) Koasati (Muskogean) Yimas (Lower Sepik-Ramu)

4.