Use of final –n for independently used quantifiers
Eric Hoekstra
Morfologiedagen 2011, Radboud Universiteit Nijmegen, 21-22 december 2011
2. Two uses of quantifiers
• Attributive usage
• Independent use
3. Attributive use
Beide fammen (wiene oan it dûnsjen)both women were at the dancing“Both young women were dancing.”
4. Independent use
Beiden wiene oan it dûnsjenboth were at the dancing“Both were dancing.”
The same facts hold of Dutch.
5. Prescriptive rules for Dutch independently used quantifiers
They are written with –n (ANS 1997:366) iff(i) they don’t have an antecedent within the text(ii) they refer to humans.Otherwise, write –e.
6. Relevant factors (hypotheses)
1. Human / nonhuman reference (ANS)
2. Specific quantifier involved
3. Intratextual / extratextual antecedent (ANS) => type of construction (Den Hertog 1973, Popkema 1979)
7. Type of construction
Partitive construction automatically entails an intratextual antecedent:
Sommige fan ‘e feintsjessome of the boys
8. Relevance of my paper
The claims made in the literature are all based on the author’s intuitions, not on corpus based research, with the exception of Popkema (1979).
9. Popkema (1979)
• Does not calculate significance.• Does not calculate phi-value (explanatory value).• Uses a smaller data set, that is, the collection of
written citations of the Dictionary of the Frisian language
• Does not investigate the mentioned factors systematically, since he tries to establish a prescriptive rule.
10. Factor 1
Is there a correlation between the use of –e or –en, and the presence of a human or nonhuman antecedent?
(Significance and phi-coefficient:http://faculty.vassar.edu/lowry/tab2x2.html)
11. Quantifier: beide(n) ‘both’
beide(n) + human - human-en 71 14-e 219 97
p = 0.9 % phi = 13% source: Popkema
-EN correlates with a human antecedent.The correlation does not explain much of the observed variation (phi = 13 %)
12. Quantifier: inkelde(n) ‘a few’
p < 0.1 % phi = 66% source: Popkema
-EN correlates with a human antecedent.The correlation explains a lot of the observed variation.
inkelde(n) + human - human-en 46 3-e 2 6
13. Sommige and somlike ´some´
p < 0.1 % phi = 60% source: Popkema
sommige(n) + somlike (n)
+ human - human
-en 44 2-e 5 7
14. Ferskate ‘several’
p > 5 % source: Frisian Language Corpus
No correlation!
ferskate(n) + human - human-en 3 0-e 18 12
15. Conclusion
The presence of a human antecedent promotes the use of -EN, for some quantifiers. However, this correlation is far from being as absolute as suggested by the ANS.
beiden‘both’
inkelden‘a few’
somliken ‘some’
ferskaten‘several’
+ + + -
16. Factor 2
Is there a correlation between the specific quantifier involved and the choice of –e / -en?
(Data on the previous slide already suggested this.)
17. Comparison inkelde ‘a few’ with ferskate ‘several’
p < 0.1 % phi = 76 % source: Popkema
There is a correlation between the choice of quantifier and –E / -EN, and it explains a lot of the observed variation.
inkelde ferskate-en 49 2-e 8 26
18. Comparison sommige+somlike ‘some’ with ferskate ‘several’
p < 0.1 % phi = 68% source: Popkema
The correlation explains a lot of the observed variation.
Sommige+somlike ferskate-en 46 2-e 12 26
19. Conclusion
There is a correlation between the use of –e or –en, and the specific quantifier involved. Low degree quantifiers show a correlation between human antecedents and –EN. The medium degree quantifier ferskate has a preference for–E. This factor is not accommodated in the ANS.
20. Factor 3
Is there a correlation between the construction type involved and the choice of suffix (–e / -en)?
21.Partitive construction
* Beide(n) fan ‘efeintsjes both of the boys
Inkelde(n) / somlike(n) / ferskate(n) fan ‘ea fewsomeseveral of thefeintsjesboys
Beide ‘both’ is anyhow excluded from the partitive.
22. Sommige ‘some’
`p < 0.1 % phi = 29 % source: FLC
The partitive correlates with –E, but it does not explain much of the observed variation.
sommige(n) partitive other-en 10 102-e 12 24
23. Somlike(n) ‘some’
p < 0.1 % phi = 32 % source: FLC
The data indicate that the partitive construction exhibits relatively more the suffix –E with this quantifier.
somlike(n) Partitive other-en 40 307-e 15 12
24. Inkelde(n) ‘a few’
p = 1,7 % phi = 41 % source: Popkema
The partitive correlates with –E.
inkelde(n) partitive other-en 2 47-e 3 5
25. Ferskate ‘several’
p > 5% source:Popkema
• The partitive does not correlate with –E/-EN for this quantifier. (NB Few instances.)
ferskate(n) partitive other
-en 0 2
-e 10 6
26. Conclusion
The partitive shows a preference for –E as compared with other constructions, for low degree quantifiers. For medium degree quantifiers, there seems to be no such correlation.
27. Wy beide(n) ‘we both’
Clausal edge (one constituent):Wy beidebinneklearwe both are ready
Middle field (one or two constituents):Dan binnewy beideklearthen are we both ready
Only beide ‘both’ can enter these constructions.
28. Wy beide(n) ‘we both’
p = 0.7% phi = 26 % source: FLC
Construction type is relevant, but explanatory value is low.
clausal edge middle field
wy beide 14 154wy beiden 4 5
29. Conclusion 1
Construction type (partitive, middle field, clausal edge) promotes the choice of –E over –EN, but not to the same degree. Most –EN is found in the clausal edge position.
30. Conclusion 2
• The partitive, the clausal edge and the middle field constructions all involve intratextual antecedents.
• Thus intratextual antecedenthood promotes the choice of –E over –EN.
• This tendency is in accordance with the rules of the ANS, but it is far from being as absolute as suggested by the ANS.
31. Overall conclusions
The following three factors are relevant forthe choice of suffix (–E / -EN):
1. Human / nonhuman reference2. Specific quantifier involved3. Type of construction
The ANS promotes major and minor tendencies to absolute prescriptive rules (as far as Frisian is concerned, but by and large Dutch seems to exhibit similar facts).