Cross-linguistic split flagging patterns (= patterns of case and adpositional marking) have often been described in terms of referential-prominence scales such as those in (1a-c), which have sometimes been put together as the “extended animacy scale” in (2).
(1a) person scale: 1st/2nd (locuphoric) > 3rd (aliophoric)
(1b) nominality scale: person form > full nominal
(1c) animacy scale: human > animal > inanimate
(2) extended animacy scale: 1st/2nd > 3rd person form > human > animal > inanimate
The claim is that accusative flagging is more likely on arguments higher on the scales, and ergative flagging is more likely on arguments lower on the scale (e.g. Dixon 1979; Bossong 1991; Haspelmath 2019).
Now Helmbrecht et al. (2018) ask the question: What about person names – do they have their own place on one of these scales? Can they be incorporated into the extended animacy scale (“1st/2nd > 3rd person form > person name > human > animal > inanimate”)? This had actually been claimed by Silverstein (1976), Dixon (1979), and others following them.
Helmbrecht and colleagues examined 34 languages with split case-marking and “hierarchical marking systems” and conclude that there is not enough evidence to put person names in the extended animacy hierarchy. Here is their abstract (they use the term “Animacy Hierarchy” for the extended animacy scale; and “morphosyntactic coding” for flagging):
The Animacy Hierarchy (AH) is an important construct employed for the description and explanation of variation and splits in case marking and agreement in various grammatical domains. The AH is a scale that combines person, definiteness and semantic animacy and is used to state clear preferences of certain morphosyntactic coding types over others. One assumption of the AH is that proper names occupy an intermediate place between personal pronouns and common nouns. Despite the large body of research since its first extensive formulation in Silverstein (1976), it is astonishing that there has been almost no empirical evidence published for this claim. Since the AH has been formulated mostly on the basis of case marking and agreement phenomena in languages with split ergativity or hierarchical alignment, we compiled a sample of more than 30 such languages in order to find data on the morphosyntactic coding of proper names. While there are only a very few instances that confirm the claim, there are more instances that contradict it. We conclude that proper names should be removed from the AH, since their assumed position has no predictive value for typological generalizations.
It is indeed very good occasionally to reexamine received wisdom – some apparent truths become textbook knowledge but later turn out not to be true after all. And the place of names on the extended animacy scale has indeed become received wisdom in some circles (e.g. in the Wikipedia article on animacy). In particular, the most widely cited article on DOM from this century, Aissen (2003: 437), mentions names as occurring above common (human) nouns on a “definiteness scale” (personal pronoun > person name > definite nominal).
Helmbrecht and colleagues find only two languages, Arabana and Yidiñ (two Pama-Nyungan languages of Australia), which confirm the position of person names on the scale: These languages have accusative marking on names, but not (to the same extent) on common human nouns. All the other languages with split flagging that they examine (actually only 21 altogether, because of missing data) have different kinds of splits. According to Aissen’s interpretation, Pitjantjatjara (another Pama-Nyungan language) is also a language of this type (the original description is by Bowe 1990; I did not go back to that source for this post), as seen in these examples (2003: 452):
(3a) Tjitji-ngku Billy-nya/ngayu-nya nya-ngu.
child-ERG Billy-ACC/1SG-ACC see-PAST
‘The child saw Billy/me.’
(3b) Billy-lu tjitji nya-ngu.
Billy-ERG child see-PAST
‘Billy saw the child.’
Here the name ‘Billy’ is accusative-marked, like the personal pronoun ‘me’, while the common noun ‘child’ is not accusative-marked.
To be sure, two (or three) languages confirming the position of names on the scale is not a lot, so one may indeed conclude that the evidence for having a special position of names on the scale is not strong. And indeed, Helmbrecht et al. suggest that the idea of names as part of the scale “has little predictive value and should be abandoned” (p. 386) (later, they say more cautiously that they have “doubts that this hypothesis can be maintained”, p. 398).
It is indeed good to be reminded that the evidence in favour of treating names separately is not overwhelming, but I find it important to stress that Helmbrecht and colleagues found no counterevidence: There is no language in which names lack an accusative flag, whereas common human nouns have an accusative flag. Thus, while we do not gain a whole lot by distinguishing names from common nouns, we do not lose anything. Hence, it seems to me, there is no case for “abandoning the hypothesis”. On the contrary, the hypothesis may lead linguists to look more closely at what happens to person names – so far, many grammatical descriptions do not pay special attention to names and may have missed distinctions that are actually there.
Surprisingly, Helmbrecht et al. seem to be missing an important class of confirming cases: Languages with split accusative marking that do not have any ergative marking at all. For some reason, they restrict themselves to languages that have some ergative marking – of course, these languages were the basis for Silverstein’s orginal formulation of the idea, but I do not see why only such languages should be considered (as a result, their sample is rather skewed: 13 Pama-Nyungan languages, 4 languages of northern South Asia, 3 Amazonian languages, plus Squamish).
If one includes languages without ergative marking, then immediately some more languages with special accusative marking of names come to mind: Yiddish (mentioned by Aissen 2003: 456), Middle High German (mentioned by Krifka 1999: 14), Old Russian (Ilja Seržant, p.c.), and apparently also a number of Romance varieties (e.g. Old Catalan, cf. Escandell-Vidal 2009).
True, special treatment of person names is far less common than special treatment of animate nominals, or special treatment of definite nominals – this is indeed important to keep in mind. And no language has been found where person names get special treatment in ergative splits (i.e. no language where ergative marking is restricted to all common nouns and does not occur with personal pronouns plus person names) (Helmbrecht et al, p. 392).
(It is true that there are some exceptions in the way personal pronouns behave, and Helmbrecht et al. list a few Australian languages with such exceptions. But these do not affect the behaviour of person names, which never show less accusative, or more ergative, marking than common nouns.)
I conclude that there is no reason to dismiss person names in an explanatory theory of split flagging of arguments. I have argued (Haspelmath 2019), following Bossong (1991) and others, that special case marking is explained by low predictability and coding efficiency (i.e. the universals of flagging are just special cases of fairly well-understood coding asymmetries in grammar). I would expect that names have a somewhat greater propensity to be used in agent function, and a lower propensity to be used in patient function, compared to human nouns (i.e. “Lee me the woman” would be more likely than “The woman met Lee” in corpora). If this is so, then we would have an explanation for the tendency that has been part of the textbook wisdom since the 1970s – and which has not been disconfirmed, even though Helmbrecht et al.’s study makes us aware that it is probably not as common as other kinds of case-marking splits.
Aissen, Judith. 2003. Differential object marking: Iconicity vs. economy. Natural Language & Linguistic Theory 21(3). 435–483.
Bossong, Georg. 1991. Differential object marking in Romance and beyond. In Douglas Kibbee & Dieter Wanner (eds.), New analyses in Romance linguistics, 143–170. Amsterdam: Benjamins.
Bowe, Heather J. 1990. Categories, constituents, and constituent order in Pitjantjatjara: an aboriginal language of Australia. London: Routledge.
Dixon, R.M.W. 1979. Ergativity. Language 55. 59–138.
Escandell-Vidal, M. Victoria. 2009. Differential object marking and topicality: The case of Balearic Catalan. Studies in Language. International Journal sponsored by the Foundation “Foundations of Language” 33(4). 832–884. doi:10.1075/sl.33.4.02esc.
Haspelmath, Martin. 2019. Role-reference associations and the explanation of argument coding splits. https://ling.auf.net/lingbuzz/004047
Helmbrecht, Johannes, Lukas Denk, Sarah Thanner & Ilenia Tonetti. 2018. Morphosyntactic coding of proper names and its implications for the Animacy Hierarchy. In Sonia Cristofaro & Fernando Zúñiga (eds.), Typological hierarchies in synchrony and diachrony, 377–401. Amsterdam: Benjamins.
Krifka, Manfred. 1999. Kasus-Synkretismus im Deutschen in typologischer Perspektive. Vortrag am Fachbereich Sprachwissenschaft der Universität Konstanz. http://publikationen.ub.uni-frankfurt.de/frontdoor/index/index/docId/14196.
Silverstein, Michael. 1976. Hierarchy of features and ergativity. In Robert M. W. Dixon (ed.), Grammatical categories in Australian languages, 112–171. Canberra: Australian National University.
Thanks for this post! I’ve used the Autotyp dataset (version 0.1.0, by Bickel et al. 2017) for a case study on differential case marking. Maybe this information can be useful. There’re about two hundred languages with some kind of Silversteinian case-flagging split in A or P. I’ve found only a couple of splits which involve proper names (in Pomoan languages and Djinang), and one split against the scale (in Gitksan). Unfortunately, the raw data with flags are not provided, but this sparseness probably suggests that the authors of the paper have a point. No traces of animals, either (lame pun intended), although I don’t know if Bickel et al. paid much attention to that category when they were analyzing the data.
Yes, they consider Kala Lagaw Ya, but they merely say that it provides counterevidence to the ranking of personal pronouns and person names. Thanks for pointing out that it’s actually another example of a language where person names are treated differently from common nouns in the expected direction.
Do they consider & refute Kala Lagaw Ya, as described by Comrie (1981) in AJL; which would bring it up to 4. Somewhere I can probably dig up a copy of Ephraim Bani’s morphological sketch of Mabuiag, which is very similar.