[Lingtyp] Structural congruence: objectivity
Martin Haspelmath
haspelmath at shh.mpg.de
Wed Jan 20 10:33:03 UTC 2016
Indeed, objectivity is really crucial, as emphasized by Volker Gast. As
Bill Croft said: "we must be careful to define comparative concepts in
consistent and crosslinguistically valid ways, that is, using formal
properties that are crosslinguistically valid and consistently applied
across languages". This gives us objective comparative results.
But whether we define "adjective" or "subject" in one way or another is
not a matter of subjective decisions, but a matter of terminological choice.
Some people seem to think that there is one correct set of optimal
comparative concepts, and that comparative concepts should not be based
on intuition or chosen "arbitrarily" (in Gilbert Lazard's words). But
this is wrong: There are myriad ways of comparing languages, and thus
myriad possible comparative concepts. Which kinds of concepts are the
most productive concepts, most likely to yield deeper insights, is a
matter for research. In fact, finding good comparative concepts is one
of the most important ingredients of the creative process for successful
comparative research.
Thus, in Volker Gast's discussions on the typology of reciprocals, it
may well be that all sides were right, in that different ways of
comparing languages would have given different, but equally important
insights...
Best,
Martin
On 19.01.16 11:19, Volker Gast wrote:
>
> Dear all,
> when I studied Linguistic Typology in the 1990s (with W. Bisang), one
> of the first things I learned was that, for non-nativists, linguistic
> categories are language-specific, and that the definition of 'tertia
> comparationis' for language comparison is one of the major
> difficulties of the discipline. Twenty years later, I think the
> problem is still with us. (Paradoxically, typology has been rather
> successful, however.)
>
> Personally, I have to say that I don't regard Martin's "comparative
> concepts", Paolo's "linguists' categorizations" or Gilbert Lazard's
> ''cadres conceptuels arbitraires" as the ultimate solution to the
> problem. In my view, they fail to meet one of the three widely
> recognized quality criteria of empirical research, objectivity (the
> other two criteria are validity and reliability). Note that
> objectivity issues may multiply in the process of data collection:
> Grammar writers interpret the raw data in a specific way, and
> typologists interpret the grammars in their own ways (unless they have
> acces to the primary data). That introduces a lot of subjectivity into
> the data.
>
> Anyone who has ever tried to build a typological database would
> probably agree that this is not just an 'academic' problem, but a very
> practical one. I remember spending hours and hours discussing the
> appropriate classification of reciprocal strategies with colleagues
> when we created BURS (http://languagelink.let.uu.nl/burs/database/). I
> assume that word order typology is seemingly simpler than semantic
> typology -- but probably the adverb 'seemingly' is important here.
>
> Best,
> Volker
>
> Am 19.01.2016 um 10:25 schrieb Paolo Ramat:
>> Martin Haspelmath has written (Jan.18.): “I find it important to
>> recognize that typology works with a heterogeneous class of
>> comparative concepts, which may be defined in a variety of ways
>> (formally, functionally, with respect to discourse, with respect to
>> translation equivalence, etc.). Typology does not (necessarily) work
>> in terms of the descriptive categories that are the most useful in
>> analyzing languages, and it need not define its concepts in a uniform
>> way.”
>> I agree. All the subsequent interventions in the discussion seem to
>> ignore the concepts of ‘prototype’ and ‘tertium comparationis’. As I
>> tried to argue in my 1999 article on /Linguistic categories and
>> linguists’ categorizations/. “Linguistics” 37: 157-80, there are
>> semantic/functional concepts which are universal (e.g. ‘attribution
>> of a quality to an X’ : you may call it “ADJ”). These concepts
>> (‘tertia comparationis’) may be implemented in different languages
>> via different strategies (e.g. via a relative clause) and it is by no
>> means said that the category ADJ is present in language A or B [see
>> Rijkhoff’s ex. from Ross :
>>
>> _Kiribati _(Ross 1998: 90)
>>
>> (1)///teueeaeetikiraoi/(relative clause)
>>
>> artflower rel 3sg.s be.pretty
>>
>> ‘a pretty flower’ (lit. ‘a flower that pretties’) ]
>>
>> On its turn a prototypical ADJ will be formally defined by a series
>> of properties (‘features’ having different /values/), like +/-
>> agreement with its head, Consequently, there are forms which are more
>> or less adjectival, according to the grammars of individual languages.
>>
>> *From:* Martin Haspelmath <mailto:haspelmath at shh.mpg.de>
>> *Sent:* Monday, January 18, 2016 9:20 PM
>> *To:* lingtyp at listserv.linguistlist.org
>> <mailto:lingtyp at listserv.linguistlist.org>
>> *Subject:* Re: [Lingtyp] Structural congruence
>> Jan Rijkhoff and Randy LaPolla are completely right that word order
>> studies have sometimes been based on formally defined comparative
>> concepts. This has long been recognized (but perhaps not emphasized
>> sufficiently), e.g. in Dryer's (2005) WALS chapter on relative
>> clauses, he defines a relative clause as follows: " A construction is
>> considered a relative clause for the purposes of this map if it is a
>> clause which, either alone or in combination with a noun, denotes
>> something and if the thing denoted has a semantic role within the
>> relative clause" (http://wals.info/chapter/90). Thus, relative
>> clauses must be clauses, i.e. simple adnominal adjectives do not count.
>>
>> (This is in contrast with Comrie's (1981) definition of relative
>> clause, which is purely semantic and thus (counterintuitively)
>> includes adnominal adjectives. This worked for Comrie's purposes,
>> because he was not interested in the ordering possibilities of
>> relative clauses, and for the generalizations that he considered, the
>> inclusion of adnominal adjectives did not make a difference.)
>>
>> By contrast, Dryer indeed includes relative clauses in his chapter on
>> the order of adjective and noun. For example, he says about Ojibwa,
>> which lacks a dedicated class of adjectives: "Because words
>> expressing adjectival meaning are really verbs iin Ojibwa, instances
>> in which such words modify nouns, like (6a), are, strictly speaking,
>> relative clauses" (http://wals.info/chapter/87).
>>
>> Here it might have been better to use the term "property word" rather
>> than "adjective", but in practice, it is often very hard to say
>> whether a language has a "dedicated" class of adjectives (Dixon 2004
>> even claims that all languages have one, even if the distributional
>> differences may be very small). Thus, it is not the terms that count,
>> but the definitions, and these are generally very clear in Dryer's
>> WALS chapters.
>>
>> When Dryer says that adjectives are non-branching elements, as
>> opposed to relative clauses which are branching elements, he
>> evidently means the most frequent types of adnominal property words
>> and adnominal clauses. Adjective phrases can be long ("very proud of
>> his achievements"), and relative clauses can be short ("who left"),
>> but it is clear that overall, relative clauses (a formally defined
>> concept) tend to be longer than property-word modifiers (a
>> semantically defined concept).
>>
>> In general, I find it important to recognize that typology works with
>> a heterogeneous class of comparative concepts, which may be defined
>> in a variety of ways (formally, functionally, with respect to
>> discourse, with respect to translation equivalence, etc.). Typology
>> does not (necessarily) work in terms of the descriptive categories
>> that are the most useful in analyzing languages, and it need not
>> define its concepts in a uniform way.
>>
>> Best wishes,
>> Martin
>>
>> On 18.01.16 13:41, Jan Rijkhoff wrote:
>>>
>>> I think the last word has not been said about Greenbergian word
>>> order correlations, mainly because semantic categories and formal
>>> categories have not always been clearly distinguished in
>>> post-Greenberg (1963) word order studies (Rijkhoff 2009a).* For
>>> example, both Hawkins (1983: 12) and Dryer (1992: 120) claimed that
>>> they followed Greenberg (1963: 74) in ‘basically applying semantic
>>> criteria’ to identify members of the same category across languages,
>>> but in practice these semantically defined forms and constructions
>>> are treated as formal entities.
>>>
>>> If Hawkins and Dryer applied semantic criteria in their
>>> cross-linguistic studies, this implies, for example, that their
>>> semantic category Adjective must also have included verbal and
>>> nominal expressions of adjectival notions (such as relative clauses
>>> and genitives), which are typically used in languages that lack a
>>> dedicated class of adjectives:
>>>
>>> _Kiribati _(Ross 1998: 90)
>>>
>>> (1)///teueeaeetikiraoi/(relative clause)
>>>
>>> artflower rel 3sg.s be.pretty
>>>
>>> ‘a pretty flower’ (lit. ‘a flower that pretties’)
>>>
>>> _Makwe_(Devos 2008: 136)
>>>
>>> (2)/ muú-nuw-á=ki-búúli/(genitive)
>>>
>>> nc1-person pp1-gen=nc7-silence
>>>
>>> ‘a silent person’ (lit. ‘person of silence’)
>>>
>>> Relative Clause and Genitive are, however, also semantic categories
>>> in their own right in word order studies by Dryer and Hawkins.
>>>
>>> When these authors subsequently formulate rules and principles on
>>> the basis of the data they collected, the semantic category labels
>>> (Adjective, Genitive, Relative Clause, but also e.g. Demonstrative
>>> and Numeral) appear to stand for _formal_ categories, i.e.
>>> categories whose members are defined on the basis of structural or
>>> morphosyntactic criteria. This apparent change of category is not
>>> explained, but can be seen in the case of the ‘Heaviness
>>> Serialization Principle’ (Hawkins 1983: 90-91) and the ‘Branching
>>> Direction Theory’ (Dryer 1992).
>>>
>>> Hawkins defined ‘heaviness’ in terms of such non-semantic criteria
>>> as (a) length and quantity of morphemes, (b) quantity of words, (c)
>>> syntactic depth of branching nodes, and (d) inclusion of dominated
>>> constituents.
>>>
>>> (3)///Heaviness Serialization Principle/: Rel≥_R Gen≥_R A≥_R Dem/Num
>>>
>>> Thus a member of the (semantic? formal?) category Relative Clause is
>>> ‘heavier’ than a member of the (semantic? formal?) category
>>> Adjective. But Hawkins’s semantic category Adjective must also have
>>> included members of the ‘heavy’ formal categories Genitive and
>>> Relative Clause (see (1) and (2) above). It is not clear whether the
>>> original members of the single semantic category Adjective were
>>> later ‘re-categorized’ and distributed over the formal categories
>>> Adjective, Genitive and Relative Clause in the /Heaviness
>>> Serialization Principle/.
>>>
>>> Dryer’s ‘Branching Direction Theory’ refers to a structural feature
>>> of the internal syntactic organization of a constituent. According
>>> to the ‘Branching Direction Theory’, relative clauses and genitives
>>> are phrases, i.e. members of a branching category, whose position
>>> relative to the noun correlates with the relative order of Verb and
>>> Object, whereas adjectives are non-branching elements, whose
>>> position relative to the noun does not correlate with OV or VO order
>>> (Dryer 1992: 107-8, 110-1). In this case, too, one may assume that
>>> the semantic category Adjective also included members of the formal
>>> categories Genitive and Relative Clause (see examples above). Again
>>> we do not know what happened to the branching/phrasal members of the
>>> erstwhile(?) semantic category Adjective (relative clauses,
>>> genitives) when this category was turned into the formal
>>> (non-branching) category Adjective that is part of the ‘Branching
>>> Direction Theory’.
>>>
>>> So as to avoid categorial confusion in cross-linguistic research
>>> (and so as to make it possible to produce more reliable results), it
>>> is necessary to keep formal and semantic categories apart, as
>>> members of these two categories have their own ordering rules or
>>> preferences. I also think it is an illusion to think we can give a
>>> satisfactory account of the grammatical behaviour of linguistic
>>> units -including word order- without taking into consideration
>>> functional (interpersonal) categories or ‘discourse units’ (Rijkhoff
>>> 2009b, 2015).
>>>
>>> * Greenberg (1963: 88) made it clear that he sometimes used formal
>>> criteria to remove certain members of a semantic category before he
>>> formulated a universal, as in the case of his Universal 22.
>>>
>>> *References*
>>>
>>> Devos, M. 2008. /A Grammar of Makwe/. München: Lincom Europa.
>>>
>>> Dryer, M. S., 1992. The Greenbergian word order correlations.
>>> /Language/ 68-1, 81-138.
>>>
>>> Greenberg, J. H. 1963. Some universals of grammar with particular
>>> reference to the order of meaningful elements. In J. H. Greenberg
>>> (ed.), /Universals of Language/, 73-113. Cambridge MA: MIT.
>>>
>>> Hawkins, J. A., 1983. /Word Order Universals: Quantitative analyses
>>> of linguistic structure/. New York: Academic Press.
>>>
>>> Rijkhoff, J. 2009a. On the (un)suitability of semantic categories.
>>> /Linguistic Typology/ 13-1, 95‑104.
>>>
>>> Rijkhoff, Jan. 2009b. On the co-variation between form and function
>>> of adnominal possessive modifiers in Dutch and English. In William
>>> B. McGregor (ed.), /The Expression of Possession/ (The Expression of
>>> Cognitive Categories [ECC] 2),51‑106. Berlin and New York: Mouton de
>>> Gruyter.
>>>
>>> Rijkhoff, J. 2015. Word order. In James D. Wright (editor-in-chief),
>>> /International Encyclopedia of the Social & Behavioral Sciences
>>> (Second Edition)/, Vol. 25, 644–656. Oxford: Elsevier.
>>>
>>> Ross, M. 1998. Proto-Oceanic adjectival categories and their
>>> morphosyntax. /Oceanic Linguistics/ 37-1, 85-119.
>>>
>>> Jan Rijkhoff
>>>
>>> ------------------------------------------------------------------------
>>> *From:* Lingtyp [lingtyp-bounces at listserv.linguistlist.org] on
>>> behalf of Alan Rumsey [Alan.Rumsey at anu.edu.au]
>>> *Sent:* Monday, January 18, 2016 12:23 PM
>>> *To:* lingtyp at listserv.linguistlist.org
>>> *Subject:* Re: [Lingtyp] Structural congruence as a dimension of
>>> language complexity/simplicity
>>>
>>> Many thanks to all of you who responded to my posting on this topic,
>>> both online and off. All the readings you have pointed me to have
>>> indeed been highly relevant and very useful, including an excellent
>>> recent publication by Jennifer Culbertson that she pointed me to in
>>> her offline response, at
>>> http://journal.frontiersin.org/article/10.3389/fpsyg.2015.01964/abstract
>>> Thanks especially to Matthew Dryer for pointing out that the
>>> Greenbergian ‘universal’ I had used as an example – the putative
>>> association between VSO and noun-adjective order — had been
>>> falsified by his much more thorough 1992 study “The Greenbergian
>>> Word Order Correlations”. My reading of that article and further
>>> correspondence with him has confirmed that, by contrast, Greenberg’s
>>> universals no 3 and 4 were solidly confirmed by his study, namely
>>> that SOV languages are far more likely to have postpositions than
>>> prepositions and that the reverse is true for VSO languages.
>>> Drawing on all your suggestions, Francesca and I have now finished a
>>> draft of the paper referred to in my posting, called 'Structural
>>> Congruence as a Dimension of Language Complexity: An Example from Ku
>>> Waru Child Language’.**If any of you would like to read it please
>>> let me know and I’ll send it to you.
>>> Alan
>>>
>>
>> --
>> Martin Haspelmath (haspelmath at shh.mpg.de)
>> Max Planck Institute for the Science of Human History
>> Kahlaische Strasse 10
>> D-07745 Jena
>> &
>> Leipzig University
>> Beethovenstrasse 15
>> D-04107 Leipzig
>>
>>
>>
>>
>>
>> ------------------------------------------------------------------------
>> _______________________________________________
>> Lingtyp mailing list
>> Lingtyp at listserv.linguistlist.org
>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>>
>>
>> _______________________________________________
>> Lingtyp mailing list
>> Lingtyp at listserv.linguistlist.org
>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>
> --
> Prof. Volker Gast
> English and American Studies
> Ernst-Abbe-PLatz 8
> D-07743 Jena
>
> Fon: ++49 3641 9-44546
> Fax: ++49 3641 9-44542
>
>
> _______________________________________________
> Lingtyp mailing list
> Lingtyp at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
--
Martin Haspelmath (haspelmath at shh.mpg.de)
Max Planck Institute for the Science of Human History
Kahlaische Strasse 10
D-07745 Jena
&
Leipzig University
Beethovenstrasse 15
D-04107 Leipzig
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20160120/dbbe5add/attachment.htm>
More information about the Lingtyp
mailing list