Swahili's locative noun-classes ... not cases?

A forum for discussing linguistics or just languages in general.
Post Reply
User avatar
Imralu
roman
roman
Posts: 960
Joined: 17 Nov 2013 22:32

Swahili's locative noun-classes ... not cases?

Post by Imralu »

So, an aspect of Swahili language that I'm slowly getting my head around is the so-called "locative noun classes". These seem to be something that's a feature of Bantu languages and I think I broke a few blood vessels in my brain looking at some Zulu grammar tables. Swahili seems to be much more simple, which makes sense seeing as it's a lingua franca.

For anyone who doesn't know about Swahili (or more generally, Bantu) noun classes, here's a short summary.
Spoiler:
Basically, there are a set of genders of nouns which are marked by prefixes, generally with different prefixes for plural and singular, and there is agreement between a noun and a following adjective (kiti kikubwa "big chair", viti vikubwa "big chairs") as well as genitive/associative modifiers (kiti changu "my chair", viti vyangu, viti vya rafiki yangu "the chairs of my friend") as well as subject and object verbal prefixes and relative syllables, which can appear in one of two locations in a verb or attached to the end of the relative pronoun amba-. All in all, it's quite complex but actually not that difficult to learn.

The noun-classes themselves have strong but not absolute semantic justifications and they are frequently used as a source of derivation. For example, taking mti/miti "tree/s" from class 3/4 (common for plants and long things) and putting it in class 7/8 (common for human made creations and diminutives), we get kiti/viti "chair/s".

The class 9/10 word nyumba means "house(s)", and the class 7/8 word chumba/vyumba is its diminutive: "room/s" and the class 5/6 version, jumba/majumba is its augmentative "mansion/s, building/s". Prefixes can even be reanalysed as part of the stem, meaning prefixes can end up stacked as in kichumba/vichumba "concubine/s" which, in the plural can also mean "household appliances", although I don't know if that sense is used in the singular at all. Another example is mchumba/wachumba meaning "fiancé(e)/s", which is essentially adding the 1/2 prefix for the human class onto the new stem "-chumba" meaning "room" ... "room person"
Wikipedia has a table of the usual agreement markers here.

I've made a slightly more detailed table for my own purposes here, with a few corrections I've noticed need to be made along the way and the standard Bantu numbers down the side as well as other ad hoc systems that I come across and some notes for myself down the bottom that I've gleaned from elsewhere.
Spoiler:
Image
So, that brings me to the locative classes themselves. When I started learning Swahili, everything kept saying Swahili has no case (it's head marking, with roles of things generally indicated on the verb, so that makes sense) except then that seems to be obviously untrue because there's a locative suffix -ni. For example, nyumba means "house" and nyumbani means "at home". Soko means "market" and sokoni means "at the market".

And then I saw that, depending on the verb, these can also mean "to or from". Ninaenda nyumbani "I'm going home", Ninatoka nyumbani "I'm coming from home."

There's a funny kink, however, in that not every noun can take this suffix. Proper nouns, for example, cannot, so there's no *Tanzaniani. You can simply say Tanzania and it means the same thing. Ninatoka Tanzania "I come from Tanzania." If you really need to make it clear, you can place a basically classifying noun before the proper noun and give it the locative suffix, so Ninatoka nchini Tanzania "I come from (the country) Tanzania." (country-LOC).

At this stage, it still seemed like a case to me ... but I couldn't make sense of why there are these "locative noun classes" in the tables or how their agreement particles could work. I think I get it now and I can kind of see how this is not a case, but I still feel uncomfortable saying it's "not a case" because it does what is usually done by case, just in a kind of different way.

So, the three locative classes are apparently:

pa- (16) exact, specific, proximate location
ku- (17) general, inexact, distant location, direction
mu- (18) internal location, along

When we add the locative suffix -ni, it essentially derives a locative-class noun from it, but it is ambiguous as to exactly which locative class it belongs to. To distinguish them, you need to see some agreement.

House, on its own is a 9/10 noun, meaning you need agreement to distinguish plural and singular anyway.
  • nyumba yangu
    ny-umba y-angu
    9/10-house(s) 9-GEN.1s
    My house

    nyumba zangu
    ny-umba z-angu
    9/10-house(s) 10-GEN.1s
    My houses
With a noun like room, we can distinguish this on the noun itself:
  • chumba changu
    ch-umba ch-angu
    7-room 7-GEN.1s
    My room

    vyumba vyangu
    vy-umba vy-angu
    8-room 8-GEN.1s
    My room
So, I knew that nyumbani means "to/at/from house(s)", chumbani means "to/at/from room" and vyumbani means "to/at/from rooms" but it clicked when I found out that I can't say
  • *chumbani changu
    *ch-umba-ni ch-angu
    7-room-LOC 7-GEN.1s
    At/in/to/from my room
When a noun is locative, it no longer belongs in the noun-class indicated by its prefix. Just as mchumba "fiancé(e)" no longer belongs to the ki/vi class but instead to the m/wa class, chumbani is a derivation and can now only belong to on of the locative classes. With further agreement, we can distinguish the meaning a bit more clearly.
  • chumbani kwangu
    ch-umba-ni kw-angu
    7-room-LOC 17-GEN.1s
    "at/to/from my room"

    chumbani mwangu
    ch-umba-ni mw-angu
    7-room-LOC 18-GEN.1s
    "inside/into/out of my room"



It's often said that the noun mahali "place" (and it's dia/idiolectal variants mahala, pahali, pahala) is the sole noun in class 16 and that class 17 and 18 don't have any nouns of their own. I've also read that this is a fairly arbitrary thing to say and that many nouns are inherently locative, such as all place names. I'm not sure I quite believe that though, since, for example, saying "the place is good/nice/beautiful" and "Tanzania is good/nice/beautiful" is quite different.
  • Tanzania ni nzuri
    Tanzania ni ny-zuri
    Tanzania COP 9/10-good
    Tanzania is good.

    Mahali ni pazuri
    mahali ni pa-zuri
    place.16 COP 16-good
In googling, I did find a few instances of Tanzania ni pazuri and I think that essentially means "Tanzania is a nice place."

One way I've seen it explained is that -ni doesn't mark something as a syntactical location, as a role in the sentence, but essentially means "place", so "nyumba" is "a house as a physical structure" and "nyumbani" is "a house place". So, if you ask me what I think of your house, I think I could answer:
  • ninaipenda
    ni-na-i-pend-a
    1sg-PRES-9-like-INDIC
    I like it (as a thing)

    ninapapenda
    ni-na-pa-pend-a
    1sg-PRES-18-like-INDIC
    I like it (as a location)
... I'm not sure how much the answer would have to depend on the way the question was phrased. In any case, I know that the phrase "I like it here" is:
  • Ninapapenda hapa.
    ni-na-pa-pend-a
    1sg-PRES-18-like-INDIC PROX.18
Here's a gratuitous extra example from the wild, where the sense of "into" is conveyed by the verb "enter" as well as the mw- prefix on the following genitive attribute:
  • ... kuingia nyumbani mwa watu bila kualikwa ...
    ku-ingi-a ny-umba-ni mw-a wa-tu bila ku-alik-w-a
    INF-enter-INDIC 9/10-house-LOC 18-GEN 2-people without INF-invite-PASS-INDIC
    walking into people's homes uninvited
These three noun classes can also be used as prefixes for verbs and to create locative "be" verbs. Cf.
  • nina
    ni-na
    1sg-with
    I have

    nilikuwa na
    ni-li-ku-wa na
    1sg-PST-INF-be with
    I had

    kuna
    ku-na
    17-with
    there is ... (in general)

    kulikuwa na
    ku-li-ku-wa na
    17-PST-INF-be with
    there was ... (in general)

    pana
    pa-na
    16-with
    there is ... (at an exact location)

    palikuwa na
    ku-li-ku-wa na
    17-PST-INF-be with
    there was ... (at an exact location)

    mna
    m(u)-na
    18-with
    there is ... (inside something)

    mlikuwa na
    m(u)-li-ku-wa na
    18-PST-INF-be with
    there was ... (inside something)
And also ...
  • Vitabu vipo mezani.
    vi-tabu vi-po meza-ni
    8-books 8-be.LOC.16 table-LOC
    The books are on the table.

    Pana vitabu mezani
    pa-na vi-tabu meza-ni
    16-with 8-books table-LOC
    There are books on the table.

    Vitabu vimo chumbani.
    vi-tabu vi-mo ch-umba-ni
    8-books 8-be.LOC.18 7-room-LOC
    The books are in the room.

    Mna vitabu chumbani
    m-na vi-tabu ch-umba-ni
    16-with 8-books 7-room-LOC
    There are books in the room.
So, I was wondering if anyone had any input into this. The orthodox description of these is as noun-classes, but is that the most accurate description? (Seeing as the most common description of tense and object verbal prefixes is as "infixes", even though they don't go inside any other morpheme. A lot of resources are very nice, descriptively, but their use of grammatical terms is a bit ... at odds with other things.) Are they cases? If you came across this in a language that hadn't been described, or came up with this way of doing things for a conlang, how would you describe it?
Glossing Abbreviations: COMP = comparative, C = complementiser, ACS / ICS = accessible / inaccessible, GDV = gerundive, SPEC / NSPC = specific / non-specific, AG = agent, E = entity (person, animal, thing)
________
MY MUSIC | MY PLANTS
Salmoneus
MVP
MVP
Posts: 3030
Joined: 19 Sep 2011 19:37

Re: Swahili's locative noun-classes ... not cases?

Post by Salmoneus »

If I've understood correctly, they don't look like cases.

You give an example of the, as it were, "native" class-16 noun as the subject of a copula. Can derived class-16s stand as subjects of copulas? Can they stand as subjects of other verbs?

If so, then the locative isn't governing the word's syntactic role in the sentence, so it doesn't look like case. Similarly, if, as you say, only some nouns can take the suffix, that suggests (though doesn't prove) it may be a derivation rather than an inflexion.

Your example with Tanzania also shows that a word can be semantically/syntactically locative even with the "locative class" marker, which again suggests it's not a case.

So to me it looks simplest just to say that 'ni' creates a derived noun indicating a place.


Similarly, consider English suffixes like -most and -ward. -most is often found in substantives - "the topmost is blue" - but is also found, in 'adjectives' and 'adverbs', in locative senses, like "I placed it topmost". And, as here, it can effectively be added to nouns, like "top" (although of course you could argue there's zero-derivation to an adjective first). Similarly, though less often, "-ward" can be used in substantives, and is added to nouns, and is also found as locatives, and unlike -most, -ward was actually originally a nominal suffix (which is why the genitive form, -wards, arose), so in OE apparently it worked more purely for both substantives and locatives. Another more limited example would be "-side". It occurs in locatives, like "he stood pitchside of the manager", "he stayed pitchside", but also as a subject, as in "the pitchside was crowded".

Now, our suffixes like this don't trigger special agreement on adjectives - but then, our adjectives don't agree for noun class at all, so that's hardly significant. More importantly, our suffixes are very patchy - although most words can take them, few do regularly, and there's a tendency for them to develop idiomatic meanings or to become fixed either in substantivised or in adjectival/adverbial form ("fireside" is almost always a noun, for instance). So our suffixes are probably MORE derivational and LESS case-like than -ni. But you wouldn't have to make them much more regularly to make them look pretty similar, so far as I can see.

[In English, the locative uses of these words are called adjectives and adverbs, but they bear no overt marking as such and can almost always be zero-derived into nouns, so this seems something of a moot point. If, for instance, English adjectives and adverbs always took certain inflections that these nouns didn't, we might just call them "locative nouns" instead.]
User avatar
gach
MVP
MVP
Posts: 513
Joined: 07 Aug 2013 01:26
Location: displaced from Helsinki

Re: Swahili's locative noun-classes ... not cases?

Post by gach »

I'd still call them noun classes since that's how they function grammatically. If you start to describe the locative classes as cases, you'll end up making statements along the lines that these cases still look and function like the regular noun classes. Inventing such a terminological divides isn't terribly economical.

It certainly feels funky when you encounter grammatical categories interacting in less common ways, like noun classification and adverbial marking here. Motuna felt even more surprising when I first encountered it since it includes both a "local" gender and a "manner" gender and its grammatical descriptions indeed go as far as to use the term "gender" rather than "noun class".
ImageKištaLkal sikSeic
User avatar
Imralu
roman
roman
Posts: 960
Joined: 17 Nov 2013 22:32

Re: Swahili's locative noun-classes ... not cases?

Post by Imralu »

Salmoneus wrote:You give an example of the, as it were, "native" class-16 noun as the subject of a copula.
Googled and found:
  • Mahali hapa ni pazuri, lakini pamejaa watu.
    mahali hapa ni pa-zuri | lakini pa-me-ja-a wa-tu
    place.16 PROX.16 COP | but 16-PRF-become.full-INDIC 2-people
    This place is nice, but it's full of people.
Perfect example ... and highly relatable too.

Incidentally, mahali/mahala/pahali/pahala is not even a native word. It's a loan from Arabic.
Salmoneus wrote:Can derived class-16s stand as subjects of copulas? Can they stand as subjects of other verbs?
I tried googling and found there's a song called Nyumbani ni nyumbani, which is potentially not very elucidating. Then I found this:
Mulango inayekwenda njiani upotevu ni mupana na watu wingi wanatembea, lakini wanaenda kwa upotevu. Lakini mlango ya uzima takatifu ni mwembamba
At first I thought "mupana" was 18-broad, in agreement with njiani, way-LOC, and I couldn't make sense of some other things till I realised that mulango is actually mlango "door". For some speakers, I think /m̩/ merges with /mu/, so that "mupana" is simply actually mpana, referring to mlango, and I realised it's that Christian thing, the English wording of which I can't remember right now, but the one about the narrow road to good and the wide road to iniquity and I can't make sense of a lot of the agreement there, so ...

... but then:
In Kiswahili, you will say “chumbani ni mahali pazuri”; or you may say “chumbani ni mahali kuzuri”; or you may say “chumbani ni mahali muzuri”.
Found here.
  • chumbani ni mahali pazuri / kuzuri / muzuri
    room-LOC COP place 16-good / 17-good / 18-good
That shows "room-place" as the subject but with the adjective actually agreeing with "mahali", which shows that it's not bound to class 16.
Salmoneus wrote:So to me it looks simplest just to say that 'ni' creates a derived noun indicating a place.
Yeah, that's what it seems like to me. I suppose the thing I find weirdest about it is that the noun itself doesn't actually specify if it means "exact place", "approximate place" or "internal space" but that it's marked through agreement on other words ... and if you don't have any agreeing words with it, that will just remain ambiguous, unless you specifically add some just to indicate that.

Also, I just find it interesting that structures that require a locative, such as the verb kwenda "to go", it means that basically, you can't go to a thing, you can only go to the place of a thing which ... in my mind is the same thing/place, seeing as the thing is, by definition, in its place ... but it's just interesting that it's grammatically conceptualised that way.

Actually, they do have prepositions to and the prepositions remove the need to have locatives ... although now I'm wondering about agreement markers with these prepositional phrases.

Salmoneus wrote:Similarly, consider English suffixes like -most and -ward. -most is often found in substantives - "the topmost is blue" - but is also found, in 'adjectives' and 'adverbs', in locative senses, like "I placed it topmost".
Hmm, my internal grammar says that last sentence is ungrammatical, but I appreciate the point you were making with it, and the rest of everything else you said.

I've also thought -ward or -wards would be something cool to turn into an allative or even dative case marker in some kind of future English, even though my gut says it's something that's probably more likely to fade away that become hyper productive and attach to everything. Still possible though.
gach wrote:It certainly feels funky when you encounter grammatical categories interacting in less common ways, like noun classification and adverbial marking here. Motuna felt even more surprising when I first encountered it since it includes both a "local" gender and a "manner" gender and its grammatical descriptions indeed go as far as to use the term "gender" rather than "noun class".
Yeah, I started watching a video of someone teaching Swahili and I just groaned when he says its easy because there are no grammatical genders like European languages. Sorry buddy, you've only got SHITLOADS MORE than any European language would dream of, lol.

What makes it not too painful is that the gender of nouns can very often be remembered because it's indicated reasonably clearly on the noun. It's a bit like Spanish or Italian with the final endings that usually tell you, but don't always. Class 1 and class 3 are generally held apart semantically. Aside from cases where something has no prefix but looks like a prefix, or has one prefix that looks like another and things like that, the thing I'm finding the hardest is sorting out the class 5/6 nouns from the class 9/10 nouns because in the singular, they generally don't have any obvious prefix ... but I'm not alone and nouns seem to drift a bit between these two classes. If you know the singular and plural of a word, you should generally know its noun class. There are some strange ones. I thought mbavu was rib and mibavu was ribs, but it turns out mbavu is the class 10 plural of the class 11 ubavu.

In any case, most people are non-native speakers and some people essentially just use 1/2 agreement for all animates (largely correct in standard language) and 9/10 agreements for inanimates (incorrect in standard language). An old colleague of mine who knows Swahili seems to use it this way as he said kiswahili yako "your Swahili", where the native-like way would be kiswahili chako. It makes me wonder how the locative agreements fare in non-native speech of this type.
Glossing Abbreviations: COMP = comparative, C = complementiser, ACS / ICS = accessible / inaccessible, GDV = gerundive, SPEC / NSPC = specific / non-specific, AG = agent, E = entity (person, animal, thing)
________
MY MUSIC | MY PLANTS
User avatar
Imralu
roman
roman
Posts: 960
Joined: 17 Nov 2013 22:32

Re: Swahili's locative noun-classes ... not cases?

Post by Imralu »

Oh, another thing I find weird is that locations can be unmarked if they're place nouns. You can just throw them into the sentence, they don't have to be verb compliments.

Nilimwoma simba (nchini) Tanzania.
I saw a lion (country-place) Tanzania.

With prepositions, I suppose there are optional ways of marking them, so it makes it no different from the sign languages I know, but it's just weird to me - especially because -ni superficially looks like case when you ou first start learning it.
Glossing Abbreviations: COMP = comparative, C = complementiser, ACS / ICS = accessible / inaccessible, GDV = gerundive, SPEC / NSPC = specific / non-specific, AG = agent, E = entity (person, animal, thing)
________
MY MUSIC | MY PLANTS
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: Swahili's locative noun-classes ... not cases?

Post by Creyeditor »

Just avfew unordered comments. Proper place nouns nopt taking prepositions is not unheard of. There is the so called 'Haltestellen'-construction in colloquial German.

Am besten Sie steigen Hauptbahnhof aus.
SUPERLAT good.SUPERLAT 2SG get.of main.station PARTICLE
You should get off at the main station.


Some languages do this more regularily, but I can't think of one right now.

Also I think noun class can be thought of as a morphomic class, because the morphemes pattern similarly wrt morphology, independent of their meaning. With the more functional side, some people have argued for three categories: core cases, oblique cases and semantic cases (or flagging if you like). The locative classes seem to behave more like an oblique case, IMHO.
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
User avatar
Lao Kou
mongolian
mongolian
Posts: 5089
Joined: 25 Nov 2012 10:39
Location: 蘇州/苏州

Re: Swahili's locative noun-classes ... not cases?

Post by Lao Kou »

Creyeditor wrote:Am besten Sie steigen Hauptbahnhof aus.
You should get off at the main station.
Head just imploded (with accompanying 'just had a stroke' face). Why not:

Am besten steigen Sie Hauptbahnhof aus.?

Colloquial?
道可道,非常道
名可名,非常名
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: Swahili's locative noun-classes ... not cases?

Post by Creyeditor »

Yes, but even colloquially yours is okay, too. I just wanted to give it that feel, you know [;)]
Mine is more like the start of a description, while yours is more likely to be at the end of some advice. At least that's what my intuition says.
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
User avatar
Lao Kou
mongolian
mongolian
Posts: 5089
Joined: 25 Nov 2012 10:39
Location: 蘇州/苏州

Re: Swahili's locative noun-classes ... not cases?

Post by Lao Kou »

Creyeditor wrote:Yes, but even colloquially yours is okay, too. I just wanted to give it that feel, you know [;)]
Mine is more like the start of a description, while yours is more likely to be at the end of some advice. At least that's what my intuition says.
Kewl. As a non-native speaker, just never seen it. I thought moving conjugated verb from S2 was echt verboten, not matter what register you were speaking in. Not the case? Nuance? I can cope. [:)] Thanks.
道可道,非常道
名可名,非常名
User avatar
Adarain
greek
greek
Posts: 511
Joined: 03 Jul 2015 15:36
Location: Switzerland, usually

Re: Swahili's locative noun-classes ... not cases?

Post by Adarain »

Lao Kou wrote:
Creyeditor wrote:Yes, but even colloquially yours is okay, too. I just wanted to give it that feel, you know [;)]
Mine is more like the start of a description, while yours is more likely to be at the end of some advice. At least that's what my intuition says.
Kewl. As a non-native speaker, just never seen it. I thought moving conjugated verb from S2 was echt verboten, not matter what register you were speaking in. Not the case? Nuance? I can cope. [:)] Thanks.
Basically the way to analyze it is that in a quite colloquial register for many people it's grammatical to have a sentence modifier of sorts (such as "am besten") precede a whole sentence, much like some conjunctions. In my dialect it's clearly ungrammatical, but I've certainly seen it, I think mostly from Germans.
At kveldi skal dag lęyfa,
Konu es bręnnd es,
Mæki es ręyndr es,
Męy es gefin es,
Ís es yfir kømr,
Ǫl es drukkit es.
User avatar
Imralu
roman
roman
Posts: 960
Joined: 17 Nov 2013 22:32

Re: Swahili's locative noun-classes ... not cases?

Post by Imralu »

That's how my landlord talks.
I've never really thought of it as anything systematic other than just dropping function words where context is enough.

I just thought I'd share a couple of compounds from Swahili that use a locative noun as an attribute.
  • taa ya barabarani
    taa y-a barabara-ni
    light 9-GEN road-LOC
    traffic light
    (on-road light, light of road-location)


    simu ya mkononi
    simu y-a mkono-ni
    phone 9-GEN hand-LOC
    mobile phone
    (in-hand phone, phone of hand location)
Like, I understand this but semantically, I find it hard to hard to really see what's so important about the difference between a thing and it's place, so I guess that's why the syntactic differences in their use arouse my attention a bit more.
Glossing Abbreviations: COMP = comparative, C = complementiser, ACS / ICS = accessible / inaccessible, GDV = gerundive, SPEC / NSPC = specific / non-specific, AG = agent, E = entity (person, animal, thing)
________
MY MUSIC | MY PLANTS
Post Reply