Comments for The World Atlas of Language Structures Online

Comment on Datapoint for feature 65A and language wals_code_wch by Neele Müller

Neele Müller — Wed, 08 Feb 2012 15:02:07 +0000

According to Terraza, Wichí does have a marked imperfective and an unmarked perfective.

Terraza, Jimena. 2009. Grammaire du Wichi: Phonologie et Morphosyntaxe. Dissertation. Université du Québec á Montréal. p.167.

Comment on Datapoint for feature 65A and language wals_code_tuc by Neele Müller

Neele Müller — Wed, 08 Feb 2012 14:52:57 +0000

As to the Tucano data, there is one Tucanoan language that probably does have a morpho-syntactically marked perfective. For Desano, Miller gives a perfect suffix -a, but this is more likely a perfective.

Miller, Marion. 1999. Desano Grammar. The Summer Institute for Linguistics and the University of Texas at Arlington. p.77.

Comment on Datapoint for feature 68A and language wals_code_hun by Alexandra Kajdi

Alexandra Kajdi — Fri, 20 Jan 2012 17:23:41 +0000

There is perfectiveness in Hungarian, although we do not express this with the help of a separate time, but with “igekötők” (prefixes-like elements), eg.: “meg”, “el” and they indicate that the action has been finished or will be finished.
For more details, see eg. Szili (2006)

Comment on Order of Subject, Object and Verb by Henry Davis

Henry Davis — Wed, 18 Jan 2012 06:27:04 +0000

Though I’m not a big fan of major constituent order as a meaningful typological category, I may as well point out that the Salish languages represented here are more uniform than the maps make them look. To be precise, all of them are V-initial, with flexible post-predicative order (VOS/VSO). In most languages/dialects (all of Central Salish, Lower Lillooet, Thompson, Okanagan), the ‘unmarked’ order (in sentences presented out of the blue containing two animate arguments) is VSO; in some Interior languages (Upper Lillooet, Moses Columbia, Coeur D’Alene) it is VOS. Nearly all Salish languages also allow SVO under more or less restricted circumstances (more restricted in Central Salish, less so in the Interior). The only exception I know of is Bella Coola, which has rigid VSO order without an SVO alternant.
However, Salish data points are spread across maps 81and 81b, falsely giving the impression, for example, that Lillooet differs fundamentally from Shuswap and Thompson, its closest relatives. As often, this is a problem with sources, which exaggerates cross-linguistic differences: van Eijk (1997) simply didn’t have enough data to characterize word order accurately, given the paucity of sentences with two overt arguments in the traditional narratives which formed the empirical basis of his grammar.

Comment on Datapoint for feature 51A and language wals_code_bsq by Matthew Dryer

Matthew Dryer — Tue, 17 Jan 2012 23:15:19 +0000

I don’t understand this comment. Basque is already coded as a language with postpositional clitics.

Comment on Datapoint for feature 37A and language wals_code_bul by Matthew Dryer

Matthew Dryer — Tue, 17 Jan 2012 21:37:48 +0000

As explained in the chapter, phonologically bound elements whose placement is determined syntactically are treated as separate words, not affixes. The Bulgarian definite article is an example of such since its position depends on the structure of the noun phrase.

Comment on Datapoint for feature 120A and language wals_code_iri by Stefan Schumacher

Stefan Schumacher — Tue, 10 Jan 2012 15:05:34 +0000

The statement that copula deletion is impossible in Irish is wrong. Ó Siadhail 1989, which is referred to without page reference, actually has a section “Deletion of copula” (pp.244–5), at the outset of which the following statement is made:
“A very general rule which seems to apply to all dialects is that the copula form _is_ may be deleted at the beginning of an utterance. This can be stated negatively by saying that the copula may not normally be deleted when marked for mood, tense, negation, interrogation or when embedded in a sentence”
In fact, this statement of Ó Siadhail’s strikes me as very appropriate, and it is my own observation that this is accurate.
Therefore I suggest changing the value of the datapoint for Irish to “possible”.

Comment on Datapoint for feature 37A and language wals_code_bul by Dimitǎr

Dimitǎr — Thu, 05 Jan 2012 22:25:53 +0000

The definite article in Bulgarian is affixed! I was very suprised to read otherwise at this datapoint. I could only check some snippets from Hubenova et al. 1968 in google docs but I don’t think it treats the Bulgarian article as aseparate word. For a detailed discussion of the topic check “Deﬁniteness Marking in the Bulgarian” by Ascander Dost and Vera Gribanova available online here:
http://www.lingref.com/cpp/wccfl/25/paper1441.pdf
There, it’s demonstrated that the article acts like an affix.

Comment on Datapoint for feature 108A and language wals_code_kab by Peter Arkadiev

Peter Arkadiev — Tue, 20 Dec 2011 17:25:10 +0000

Data from Kabardian is misinterpreted due to the inadequacy of the source. There is a mistake in Colarusso’s example on p. 106 (the object of an antipassive verb, if it is expressed at all in the first place, which is normally not the case, cannot appear in the Absolutive case). The situation with antipassives in Kabardian (and in the closely related Adyghe) is complex; see e.g. Smeets 1992, Kumakhov & Vamling (2006). Generally, antipassivization renders the verb monovalent, i.e. the original object can no longer be expressed at all. With some verbs the object can be expressed as an adjunct in the Instrumental case, and another couple of verbs have a bivalent antipassive with the original O expressed as a indirect object cross-referenced in the verb and marked with the Oblique case used for non-Absolutive arguments.

Reference:
Kumaxov, Muxadin A., Vamling, Karina. 2006. Ėrgativnost’ v čerkesskix jazykax. Malmö: Malmö University: School of International Migration and Ethnic Relations. 2006.
Smeets R. On valencies, actants and actant coding in Circassian. // G. Hewitt (ed.), Caucasian Perspectives. München, Newcastle: LINCOM Europa, 1992, pp. 98—144.

Comment on Datapoint for feature 104A and language wals_code_ika by Anne Goergens

Anne Goergens — Fri, 16 Dec 2011 14:19:30 +0000

Hi!
The example for Iko shows, that P is a prefix, while S is a suffix. Still, it is classified as “A precedes P”. Either the translation of the example or the classification must be wrong.

Greets,
Anne Goergens

Comment on Datapoint for feature 51A and language wals_code_bsq by ISHIZUKA Masayuki

ISHIZUKA Masayuki — Thu, 15 Dec 2011 10:29:28 +0000

Although many authors describe the case markers in Basque as suffixes, they should be considered postpositional clitics in the light of the definition: ‘while they[=the adpositional clitics] attach phonologically to some word, the word they attach to need not be a noun, and which word they attach to is determined syntactically.’

The case markers in Basque do attach to a noun, an adjective or a determiner. In fact the word they are fixed to is the final one of noun phrases. See the examples below:

1) bost gizon=i
five man=DAT
‘to five men’
2) gizon haue=i
man these=DAT
‘to these men’
3) bost gizon gazte=ri
five man young=DAT
‘to five young men’

The dative marker =(r)i is attached to a noun, a determiner, and an adjective in each sentence.

Comment on Datapoint for feature 77A and language wals_code_map by Ole Nedergaard Thomsen

Ole Nedergaard Thomsen — Mon, 12 Dec 2011 10:47:20 +0000

Concerning: No grammatical evidential in Mapudungun (you refer to Smeets 1989!!!). To the best of my knowledge, this claim is simply not true. In Smeets (1989, p. 322 ff.) it says quite clearly: ”[-(ü)rke] … indicates that the denoted situation has not been witnessed by the speaker himself. [It] indicates that the speaker is informed by others, has heard rumours, or has deduced a conclusion. Although the source of information is left unspecified, it is usually evident who the source of information is. I found [it] in indicative forms only. [It] is very frequent in narratives.”

In fact evidentiality or mediativity (plus mirativity) is mentioned in every handbook/grammar/descripition of the language by the specialist of the language, right from the beginning! How come you haven’t consulted the relevant sources, before ruling out a category for a language.

Comment on Datapoint for feature 66A and language wals_code_akn by Ato Quansah

Ato Quansah — Tue, 06 Dec 2011 08:07:18 +0000

There is a past tense. It is signaled by a tonal system. For example, with the simple past tense, the syllables of the verb clause remain constant until the last vowel (on the last syllable of the verb) , where this vowel is elongated and a downstep is applied to its tone. With the past participle, an up-step is applied to the beginning of the noun clause (on the first syllable of the noun). This might not be apparent to non-native speakers, but this tonal ( tone-terracing) system is very real. Sometimes the tone is applied to the vocal accompaniment to nasal consonants making it ever more undetectable for non-native speakers yet still existent and necessary communicating in the language.

These simple guidelines are just illustrative and their simplicity is their error because there is a highly complex system underlying the system, much like any language.

Comment on Nominal and Verbal Conjunction by Henry Davis

Henry Davis — Sun, 04 Dec 2011 07:35:32 +0000

Here’s a textual example of clausal coordination with múta7 in St’át’imcets/Lillooet from Lisa Matthewson’s 2005 volume of personal histories of St’át’imc elders: (the example is in the van Eijk practical orthography in common use in Lillooet territory):

[xzum ti=q’úmqen-s=a] múta7 [xzum ti=tsútsin-s-a]
[big det=head-3poss=exis] and [big det=head-3poss=exis]
“It’s head was big and it’s mouth was big.”

The reason why múta7 is less common as a clausal conjunction than as a phrasal conjunction is because unlike English ‘and’ it never encodes temporal sequencing, so it is limited to cases like the one above where there is no temporal ordering between clauses.

Henry Davis

Comment on Zero Copula for Predicate Nominals by Matthew Dryer

Matthew Dryer — Fri, 02 Dec 2011 01:09:25 +0000

The problem raised by Henry Davis is actually more general than the Pacific Northwest. In his book Intransitive Predication, Stassen has a more fine-grained typology in which languages in the Pacific Northwest are coded as languages in which predicate nominals have verbal encoding, taking affixes that also occur on predicate verbs. His typology also distinguishes languages which do not employ copulas but where predicate nominals do not take verbal encoding, which he classifies as nominal encoding. When his more fine-grained typology was collapsed into two types for WALS, those with verbal encoding were, for complex historical reasons, classified with languages where a zero copula is impossible. This applies to all languages with verbal encoding, not just those in the Pacific Northwest. Stassen now agrees that this is very misleading and is planning on producing a version of this map in a future online WALS that employs a more fine-grained typology, in which these languages with verbal encoding will be treated as a separate type.

Comment on Nominal and Verbal Conjunction by Martin Haspelmath

Martin Haspelmath — Thu, 01 Dec 2011 15:18:24 +0000

Thanks for this comment. In van Eijk’s grammar, múta’ is described as combining NPs, but not as combining clauses (only nilh is mentioned for the latter function).

Comment on Datapoint for feature 2A and language wals_code_sor by David Stampe

David Stampe — Tue, 29 Nov 2011 21:12:56 +0000

Sora does not have 5-6 vowels but nine: i e ɛ ɨ ə ɑ u o ɔ, without phonemic length.
Ramamurti’s Telugu-influenced analysis with length, stress, etc., does not match any
of the dialects he investigated. Actually even Ramamurti’s analysis can’t be taken
to have fewer than nine vowels.

The nine-vowel analysis, with occasional variations in symbols, has been
independently arrived at in published and unpublished works produced since 1962
by H. S. Biligiri, David Stampe, Stan Starosta, Bijoy Prasad Mahapatra, Raganayaki
Mahapatra, Arlene R. K. Zide, Khageswar Mahapatra, and Arun Ghosh. Gregory D. S.
Anderson’s report in his edited Munda Languages volume was a hurried replacement
article based on very brief field work, but despite its uncertainties it is not
incompatible with the majority analysis.

Incidentally, the Sora New Testament of the Canadian Baptist Mission also
nine vowels written i I e ü ə ɑ u U o (based on Ramamurti, w/o length or stress),
where ü was Ramamurti’s misunderstanding of Sweet’s notion “mixed vowel”, really ɨ.
The full Bible was published in the English alphabet, but except for writing ə and ɑ
both as a, it also distinguishes the nine vowels (though with many mistakes).
A detailed description of Sora phonology and phonetics is in preparation by
Patricia Donegan, donegan@hawaii.edu.

Comment on Lateral Consonants by Henry Davis

Henry Davis — Mon, 28 Nov 2011 06:49:41 +0000

Bella Coola has both an ejective lateral affricate and a lateral fricative, contrary to what you have on the map.

Henry Davis

Comment on Presence of Uncommon Consonants by Henry Davis

Henry Davis — Sat, 26 Nov 2011 07:36:53 +0000

More Salish mistakes. Squamish has no pharyngeals. Shuswap does. Did you mix up the two Kuipers grammars?

Comment on Locus of Marking in the Clause by Henry Davis

Henry Davis — Sat, 26 Nov 2011 06:29:26 +0000

The datapoint for Squamish (Salish) is incorrect. Like all other Salish languages, Squamish is head-marking: objects are obligatory encoded as suffixes on the verb, but there are no case distinctions and dependent marking is limited to a distinction between obliques and non-obliques.

Comment on Order of Object, Oblique, and Verb by Henry Davis

Henry Davis — Fri, 25 Nov 2011 07:26:50 +0000

Wrong again on Gitksan: strictly VSOX.

Comment on Order of Object and Verb by Henry Davis

Henry Davis — Fri, 25 Nov 2011 07:22:55 +0000

No Tsimshianic language ever allows OV order; the Gitksan value is anomalous here.

Henry Davis

Comment on Order of Subject and Verb by Henry Davis

Henry Davis — Fri, 25 Nov 2011 07:14:13 +0000

There’s another Gitksan~Nisga’a mistake in here. As I commented earlier, these two languages are mutually intelligible ends of the Interior Tsimshianic dialect continuum: they are not going to differ in major typological features.

In fact, both languages have very strict VSOA order. I’m not sure where the ‘no dominant order’ value comes from for Gitksan, unless the compiler is treating pre-predicative ergative clitics as full arguments (but in that case, French would count as SOV whenever object clitics showed up).

Henry Davis

Comment on Definite Articles by Matthew Dryer

Matthew Dryer — Mon, 21 Nov 2011 18:21:34 +0000

What Henry Davis says is consistent with how I have Gitksan coded in my database. The current coding is due to a programming error.

Comment on Zero Copula for Predicate Nominals by Henry Davis

Henry Davis — Sat, 19 Nov 2011 08:19:00 +0000

Unfortunately, the entire Pacific NW section of this map is completely wrong. Wakashan and Salish languages, in particular, are *famous* for lacking predicative copulas (that is one of the bases of Swadesh’s notorious claim that they lack lexical category distinctions altogether). No Salish or Wakashan language has a copula in nominal predicae constructions: this includes Lilooet, Squamish and Kalispel (Salish) and Heiltsuk, Kwak’wala, and Nuuchahnulth (Wakashan). This is a pretty egregious error and should be fixed as soon as possible.

Henry Davis

Comment on Definite Articles by Henry Davis

Henry Davis — Sat, 19 Nov 2011 08:02:14 +0000

Nisga’a and Gitksan are mutually intelligible, and form part of an Interior Tsimshianic dialect continuum. They have identical nominal semantics, yet they are given different values on the map here! In fact, the Nisga’a value is correct: articles in IT are unspecified for (in)definiteness, and there is no definite affix (the Gitksan value must be a misreading of Rigsby’s grammar).

Henry Davis

Comment on Nominal and Verbal Conjunction by Henry Davis

Henry Davis — Sat, 19 Nov 2011 07:49:04 +0000

You cite Lillooet (Salish) as exemplifying a ‘differentiated’ system. But Lillooet uses its general purpose conjunction múta7 for coordination of all constituents, including clauses and DPs. It also uses its clefti marker nilh to connect sequences of clauses, but nilh is not really a conjunction, but a subordinating predicate. On your criteria, I’d therfore say that Lilllooet uses an ‘identical’ strategy.

Henry Davis

Comment on Absence of Common Consonants by Doug Marmion

Doug Marmion — Wed, 05 Oct 2011 07:04:24 +0000

Another language lacking velar consonants is Wutung (Sko; Sandaun, Papua New Guinea), a very close relative of the already-cited Dumo. This is mentioned in my PhD thesis on the language.

Comment on Datapoint for feature 14A and language wals_code_bqs by Javier

Javier — Wed, 28 Sep 2011 16:48:24 +0000

I wonder why Basque was assigned the feature \first stress\. Hualde & Urbina (2003: 65) point out \Usually the stress falls on the second syllable of the word or the phrase, frequently with secondary prominence on the last syllable [...]. Bisyllabic words ending in a wowel receive initial stress instead [...]. But, as mentioned, stress is not phonologically contrastive and may also fall on other syllables\.

Comment on Datapoint for feature 122A and language wals_code_kha by Henri de Solages

Henri de Solages — Sun, 25 Sep 2011 10:01:59 +0000

Although of limited use, there are relative clauses and relative pronouns in Mongolian. Example:
“⁅Хэн даарна⁆ тэр хүүхэд ханиад амархан хүрдэг.” ( http://www.tvhicheel.mn/file/232.shtml ) “A child ⁅who is cold⁆ gets a caught easily.”