Universal #9
Legacy number: 9
| Number | 9 |
| Original | If in apposition the proper noun usually precedes the common noun, then the language is one in which the governing noun precedes its dependent genitive. With much better than chance frequency, if the common noun usually precedes the proper noun, the dependent genitive precedes its governing noun. |
| Standardized | IF in apposition the proper noun usually precedes the common noun, THEN the head noun precedes the attributive noun. IF in apposition the common noun usually precedes the proper noun, THEN, with much better than chance frequency, the attributive noun precedes the head noun. |
| Formula | Proper Common ⇒ N G, Common Proper ⇒ G N |
| Keywords | order, apposition, proper name, attributive, genitive |
| Domain | syntax |
| Type | implication |
| Status | achronic but presumably diachronically motivated |
| Quality | statistical |
| Basis | 17 lgs of the 30 languages of Greenberg 1963 sample |
| Source | Greenberg 1963: 89-90, #23 |
Discussion
Counterexamples
Slavic languages are mentioned in the Russian translation of Greenberg 1963 as counterexamples to part 2.In Old English (Germanic, IE), proper nouns precede their appositive common nouns in close-knit apposition (e.g., Ælfred cyning ‘Alfred king’; when the common noun is accompanied by a determiner and/or modifier, however, the order can be reversed: ‘the good king Alfred’ or ‘Alfred the good king’); genitives can precede or follow their heads (or also be split up: ‘with God’s help the almighty’s’), but a good case can be made for pre-head position of genitives being more common/basic.The eventual reversal of the order in apposition has been claimed to have been inspired by a single, culturally salient Latin model borrowed into Old English: Dominus Christ ‘Lord Christ’. Conceivably, left to its own devices, English might thus have continued to infringe on this universal.
Notes
See also #10 -- to be DELETED.Greenberg's Note 19:Languages with common noun–proper noun are Greek, Guarani, Italian, Malay, Serbian, Swahili, Thai, Welsh, Zapotec. Those with proper noun–common noun are Basque, Burmese, Burushaski, Finnish, Japanese, Norwegian, Nubian, and Turkish.*****Concerning the counterexample status of Old English, perhaps the universal shouldn't really be assumed to apply to such personal name constructions in the first place. They consist of a term of address (title or kin term) as the common part and a personal name as the proper part (e.g., King Alfred, Captain Ahab, Master Copperfield, Father Christmas, Uncle Joe); they are NOT diachronically derivative of constructions consisting of head noun and an attributive nominal. The universal might have to be limited to name constructions which ARE diachronically derivative, such as for places and times, as copiously exemplified in English (New York State, Calton Hill, Cramond Island, Wivenhoe Park, London Town, Tower Bridge, George Square, David Hume Tower, Scotland Yard, Edinburgh University, etc.). Assuming that attributive genitives used to be pre-head in earlier English (as 's genitives still are), and assuming further that the actual source of these appositive name constructions were attributive constructions, with the attributive marker dropped (Hudson's Bay > Hudson Bay, Humboldt's Current > Humboldt Current, Princes' Street > Princes Street), one would expect the ordering to remain the same. The correlation that Greenberg states as an achronic universal would thus be of a diachronic nature: as one construction (attribution) is reanalysed as another construction (close apposition), the ordering of the parts of these constructions remains unaltered. (Greenberg does reckon with some sort of a diachronic connection: constructions of the type of 'City X' are said to be possibly "assimilated" to genitive constructions, 'the city of X', "and may therefore be expected to show a similar order". My claim here is that they are diachronically derived from genitive constructions, and for this reason alone -- without any achronic universal demanding such harmony, but from simple inertia -- may be expected to show a similar order.) Among place names, those with River (River Thames), Lake (Lake Windermere), Loch (Loch Ness), Cape (Cape Horn), and Mount (Mount Everest) deviate from the otherwise unexceptional Proper–Common pattern in Modern English. They retain the ordering in the source languages of these borrowed common nouns, namely French (perhaps Latin) and Gaelic.
Comment
It seems to have gone unnoticed that the data in Greenberg's table 9, on which
Universal 23 is based (number 9 in the Archive), are inconsistent. The 9 languages in the first row
displaying the order Common Noun - Proper Noun are in the first and second
column (i.e. VSO and SVO), but in the second part of the table, where the
order of nouns in apposition is correlated with GN order, 8 of them are
GN. While 6 of the 8 languages displaying the order Proper Noun -Common Noun
are in the 3rd column (i.e. SOV), NONE in the second part of the table has the order GN.
If one then looks at the languages listed in note 19, one finds that the languages
with the order common noun - proper noun are Greek, Guarani,
Italian, Malay, Serbian, Swahili, Thai, Welsh, Zapotec. Of these, at least two, not only one,
as the second part of the table would suggest, are strictly NG (e.g. Italian and Welsh)
and were counted as such in Greenberg's data;
on the contrary, the languages with the order proper noun - common noun are
Basque, Burmese, Burushaski, Finnish, Japanese, Norwegian, Nubian and Turkish;
and here again it is impossible that NONE of them is GN.
Thus, it is clear that in the second part of table 9 the two column headings GN
and NG must be reversed, i.e. there are 8 languages with CN-PN and NG (not GN)
and one with CN-PN and GN (not NG) (this language is Guaraní), and 8 languages with PN-CN and GN (not
NG). The only two SVO languages which have the order PN-CN have GN (these are Nowegian and Finnish).
The corresponding universal, which apparently was formulated on the basis
of a look at the table with a slip, should evidently be inverted.
The implication is exactly the opposite (and in fact it is the Proper Noun
which is "assimilated to the genitive", as Greenberg suggests in the comments just above the table):
none of the languages with PN-CN is NG.
The reasons why this inconsistency based on a trivial mistake has gone
unnoticed may be several: the fact the the data here are incomplete (only
17 languages are taken into account), the fact that the data are almost the
same in figures (8 vs. 1 and 8 vs. 0 for the two rows), the fact that in Greenberg's article the
basic data on the order GN/NG are not represented directly in a table or
in the data listed for each language in Appendix 1 (and must then be
deduced from other data), and maybe also the fact that the universal in general is not often
taken into consideration (for example it is not in WALS), perhaps because too many exceptions surfaced (of course).
Thus the evolution of English, commented upon in the Archive, does
not pose any particular problem: Old English was basically PN-CN and GN, although
it allowed some freedom in word order both for genitives and for proper names,
while Modern English has both genitive orders but the new genitive order is harmonic with
its new CN-PN order.
Universal 23 is based (number 9 in the Archive), are inconsistent. The 9 languages in the first row
displaying the order Common Noun - Proper Noun are in the first and second
column (i.e. VSO and SVO), but in the second part of the table, where the
order of nouns in apposition is correlated with GN order, 8 of them are
GN. While 6 of the 8 languages displaying the order Proper Noun -Common Noun
are in the 3rd column (i.e. SOV), NONE in the second part of the table has the order GN.
If one then looks at the languages listed in note 19, one finds that the languages
with the order common noun - proper noun are Greek, Guarani,
Italian, Malay, Serbian, Swahili, Thai, Welsh, Zapotec. Of these, at least two, not only one,
as the second part of the table would suggest, are strictly NG (e.g. Italian and Welsh)
and were counted as such in Greenberg's data;
on the contrary, the languages with the order proper noun - common noun are
Basque, Burmese, Burushaski, Finnish, Japanese, Norwegian, Nubian and Turkish;
and here again it is impossible that NONE of them is GN.
Thus, it is clear that in the second part of table 9 the two column headings GN
and NG must be reversed, i.e. there are 8 languages with CN-PN and NG (not GN)
and one with CN-PN and GN (not NG) (this language is Guaraní), and 8 languages with PN-CN and GN (not
NG). The only two SVO languages which have the order PN-CN have GN (these are Nowegian and Finnish).
The corresponding universal, which apparently was formulated on the basis
of a look at the table with a slip, should evidently be inverted.
The implication is exactly the opposite (and in fact it is the Proper Noun
which is "assimilated to the genitive", as Greenberg suggests in the comments just above the table):
none of the languages with PN-CN is NG.
The reasons why this inconsistency based on a trivial mistake has gone
unnoticed may be several: the fact the the data here are incomplete (only
17 languages are taken into account), the fact that the data are almost the
same in figures (8 vs. 1 and 8 vs. 0 for the two rows), the fact that in Greenberg's article the
basic data on the order GN/NG are not represented directly in a table or
in the data listed for each language in Appendix 1 (and must then be
deduced from other data), and maybe also the fact that the universal in general is not often
taken into consideration (for example it is not in WALS), perhaps because too many exceptions surfaced (of course).
Thus the evolution of English, commented upon in the Archive, does
not pose any particular problem: Old English was basically PN-CN and GN, although
it allowed some freedom in word order both for genitives and for proper names,
while Modern English has both genitive orders but the new genitive order is harmonic with
its new CN-PN order.
Log in to join the discussion.
*****
Concerning the counterexample status of Old English, perhaps the universal shouldn't really be assumed to apply to such personal name constructions in the first place. They consist of a term of address (title or kin term) as the common part and a personal name as the proper part (e.g., King Alfred, Captain Ahab, Master Copperfield, Father Christmas, Uncle Joe); they are NOT diachronically derivative of constructions consisting of head noun and an attributive nominal. The universal might have to be limited to name constructions which ARE diachronically derivative, such as for places and times, as copiously exemplified in English (New York State, Calton Hill, Cramond Island, Wivenhoe Park, London Town, Tower Bridge, George Square, David Hume Tower, Scotland Yard, Edinburgh University, etc.). Assuming that attributive genitives used to be pre-head in earlier English (as 's genitives still are),and assuming further that the actual source of these appositive name constructions were attributive constructions, with the attributive marker dropped (Hudson's Bay > Hudson Bay, Humboldt's Current > Humboldt Current, Princes' Street > Princes Street), one would expect the ordering to remain the same. The correlation that Greenberg states as an achronic universal would thus be of a diachronic nature: as one construction (attribution) is reanalysed as another construction (close apposition), the ordering of the parts of these constructions remains unaltered. (Greenberg does reckon with some sort of a diachronic connection: constructions of the type of 'City X' are said to be possibly "assimilated" to genitive constructions, 'the city of X', "and may therefore be expected to show a similar order". My claim here is that they are diachronically derived from genitive constructions, and for this reason alone -- without any achronic universal demanding such harmony, but from simple inertia -- may be expected to show a similar order.) Among place names, those with River (River Thames), Lake (Lake Windermere), Loch (Loch Ness), Cape (Cape Horn), and Mount (Mount Everest) deviate from the otherwise unexceptional Proper–Common pattern in Modern English. They retain the ordering in the source languages of these borrowed common nouns, namely French (perhaps Latin) and Gaelic.