|
lycklig-t
sad-neut.sg
slut.
end
attr
‘It was a sad ending.’
Lisa
Lisa
är
be.prs
lycklig.
sad.comm.sg
pred
‘Lisa is sad.’
Slut-et
end-n eut
var
be.pst
lycklig-t.
sad-neut.sg
pred
‘The ending was sad.’
Lisa
Lisa
log
smile.pst
lycklig-t.
sad-neut.sg
adv
‘Lisa smiled sadly.’
To analyze the Swedish examples as instantiating encoding overlaps, a word form level must be acknowledged. Thus, it could be argued that Swedish has a word form overlap of all three functions in the case of the neuter ending -t.2
The utility of the word form level varies across languages. As we shall see, very few languages in my sample have tendencies towards such patterns as Swedish. These are briefly discussed in section 6.3. The point here is nonetheless that the word form level may be necessary for analysis of attr, pred, and adv in certain languages. Assuming word forms as a level of analysis also leads to the morphosyntactic problems surrounding words and whether they can be distinguished from affixes (see e.g. Haspelmath 2011). While this is a complex issue, it does not pose any problems for the distinctions drawn here.
So far, we have focused on the root, lexeme, and word form levels. But there are other ways to analyze the English examples discussed in (4.2-4.4). If the focus is shifted to constructions as wholes, the interpretation of overlaps will necessarily also shift. In (4.7), we revisit the examples with sad, paying attention to entire constructions (here indicated with square brackets in the examples) instead of lexemes or roots (the property word is still highlighted).
2 Note that Swedish also has a couple of simple adverbs such as fort ‘fast’ and sakta ‘slowly’, which means that the lexeme level would also be relevant here (cf. examples in 3.22 in section 3.3.4).
(4.7) English [attr] [pred] [adv] construction
[The sad person] over there is Sheila. attr
[Sheila is sad]. pred
[Sheila smiled sadly]. adv
In (4.7a), the construction which contains a property word that functions as a modifier within a referring expression is the NP made up of the definite article, the Adjective sad and the Noun person. In pred, the construction that predicates a property consists of the subject Sheila, the third person singular copula is, and the Adjective sad. Finally, in adv, the construction that contains a property word that is a modifier within a predicating expression is made up of the subject Sheila, the Verb smiled and the derived Adverb sadly. From a constructional point of view, English (as expected) has three different ways of encoding the three functions.
Figure 4.5. The English constructions in attr, pred, and adv
If the examples with fast (see 4.4) are analyzed on the level of constructions, this will result in the same three different types of encoding. In the case of alone, there will be no overlap either. Rather, pred and adv are encoded in different ways.
The shift to focus on the level of the construction does perhaps not appear to be very fruitful for English, in comparison to focusing on the levels of the lexeme and the root. At most, it shows us what we already know, that the constructions in the three functions differ. But there are languages in which it is necessary to capture whole constructions in order to understand how attr, pred, and adv are related. For instance, one cross- linguistically recurring pattern is a construction that is intermediate between pred and adv. This is attested in Lakota (Siouan). Lakota has several other constructions found in pred and adv, more central to the two functions (cf. appendix B). The examples of the intermediate construction are taken specifically in order to illustrate why the construction level is important for the analysis of Lakota. It is neither possible to distinguish two different constructions nor two different functions here. This is illustrated in the examples in (4.8), where both examples can be interpreted as either pred or adv.
(4.8) Lakota (Siouan) (Ingham 2003: 45)
pred/adv
construction
[paha
mountain
ki
top
Do'stlaringiz bilan baham: |
