Pernilla Hallonsten Halling

waNkatu-ya

be.high-aDv

he] be

‘The mountain stands highly. / The mountain was high.’^3
3 The English translation alternatives come from Ingham (2003: 45), but it seems that (4.8a) could equally well be translated as ‘The mountain stands high’.

2. 
   [taku
   

something
waN
one
ska-yela

be.white-aDv

he] be

‘something stood whitely / there was something white there’


The existential verb he ‘be’ used in (4.8) is not a copula, but a full lexical verb. The suffixes -ya and -yela are two of a number of Adverb-deriving suffixes in Lakota, which attach primarily to verbal stems, but sometimes also to nominal stems (Ingham 2003: 43- 44). However, -ya and -yela can be distinguished from other Adverb-deriving suffixes in that the Adverbs that they form “are often used in a construction with the existential verbs -haN/he or -yaNka/e ‘be in a place’ to describe an object” (2003: 45). This is illustrated with he in (4.8a) above, where it is impossible to distinguish between pred (e.g. ‘The mountain was high.’) and adv (e.g. ‘The mountain stands highly.’). This is not a construction that is a prototypical instance of either pred or adv. It is rather exactly between the two functions, and illustrates how their edges can be blurred. Loeb- Diehl (2005: 243) treats Lakota forms with -ya as participles. Depending on where the line is drawn, they may also be seen as converbs. More importantly, if only the lexeme or root and not the whole construction were analyzed, the fact that this construction is intermediate between pred and adv could not have been accounted for.
Let us now consider what logical combinations of encoding overlap there are for attr, pred, and adv, regardless of language and level (lexeme, root, word form, or construc- tion). Firstly, the three functions may all show different encoding, i.e. no overlap. Sec- ondly, they may all be encoded in the same way. Thirdly, there may be pairs of encoding overlaps, with the remaining function being encoded in a separate way. This results in five potential types of encoding overlap, as illustrated in figure 4.6.

(e)
Figure 4.6. Potential combinations of separating and overlapping encoding


As we have seen above, a certain semantic notion or even a single lexeme may occur in only one or two of the three functions in a specific language. Along the same lines, a certain language may also have different types of encoding overlap, depending on what semantic notions or lexemes that are examined, and more importantly, depending on whether the focus is the lexeme, the root, the word form, or the whole construction. Accordingly, to capture attr, pred, and adv in a specific language, several of the combinations of encoding patterns in figure 4.6 may be needed to account for all relevant examples. When encoding overlaps covering two functions are attested in this thesis, these are usually not treated as opposed to any specific encoding in the function that is not included in the overlap. Rather, when an overlap of two functions is discussed, this simply implies that the third function is encoded in a different way.
In summary, individual languages may have several different encoding patterns in one and the same function. Whether or not these are instances of major or minor patterns, it must be possible to account for all of them. Moreover, when comparing the three func- tions, the focus may be on individual roots, lexemes, word forms, or whole constructions. For a full picture, one must be able to make comparisons at all these levels (although the word-form level applies only marginally, see section 6.3). In order to do so, I have cho-

sen to use the constructional-typological approach (Koch 2012). This approach employs a notation that captures constructions in their entirety for each language and function examined, and thus provides a level of abstraction at which they can be compared. In this way, it becomes possible to analyze whole constructions as well as the roots, word forms, or lexemes inside constructions. In (4.9), the English examples discussed above are captured in their entirety in constructional-typological format.

(4.9) attr, pred, and adv in English

attr

Function: property modification within referring expression
Form: art ADJ N
Example: (4.7a)

preD

Function: property predication
Form: S be_cop ADJ Example: (4.7b)

aDv

Function: property modification within predicating expression
Form 1: S V ADJ-ly _advz Form 2: S V ADJ_subset Example: (4.7c), (4.4c)
The notation in (4.9) contains three different types of lines for each of the functions attr, pred, and adv. The first one is Function, which gives a definition of the function in question. This is language-independent, or in other words, constant across languages. The second line is Form, which contains the language-specific encoding form found in the function, in constructional notation (see more on this in section 4.3), here for English. The third one is Example, which contains cross-references to actual examples. As illustrated for adv in (4.9), there can be more than one instance of Form, i.e. more than one way of encoding the function. In that case, the forms are numbered. In adv in (4.9), Form 1 refers to such examples as sadly, and Form 2 refers to the example with fast (the examples with alone are not included here, since they are considered to be outside the scope of this study).⁴ Example may contain more than one cross-reference, according to how forms that are found and examined in the function in question for a specific language.
There are several advantages to presenting and analyzing examples in this way. First, we neither have to limit the analysis to the level of the root, word form, or lexeme, nor to a set number of elements surrounding them. Instead, the construction as a whole can be analyzed, in addition to the more delimited levels of root and lexeme. Secondly, the study does not have to be limited to one or even a few forms for each function and language. This reduces the risk of skewed results considerably. The forms and examples
⁴ In adv in (4.9), the forms could be seen as being ranked according to productivity, since Form 1 is a major pattern, but Form 2 is a minor one. However, this is not a principle that I will employ generally, since it is often not possible to tell which patterns are major and minor in individual languages.

can be as many as necessary according to what is found in a specific language. In order not to miss relevant data, or not to focus on certain forms at the expense of others, all possible forms occurring in a specific function must be considered, to the extent this is possible. Most importantly though, the constructional-typological notation is suitable for making a cross-linguistic comparison of attr, pred, and adv in a consistent way. In the next section, the constructional-typological approach is described and evaluated in greater detail.

3.