Slot-and-filler vs templatic

A forum for discussing linguistics or just languages in general.
Post Reply
User avatar
eldin raigmore
korean
korean
Posts: 6352
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Slot-and-filler vs templatic

Post by eldin raigmore »

What is slot-and-filler morphology,
and what is templatic morphology?
And how do they differ?
From each other,
and from triconsonantal-root morphology,
and from agglutinative morphology,
and fusional morphology,
and polysynthetic
and oligosynthetic
and isolating
and analytic morphologies?
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: Slot-and-filler vs templatic

Post by Creyeditor »

This question would require a whole into class to morphology as an answer [:D] Maybe someone can write one aimed at conlangers if he/she/they have the time.
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
Sequor
sinic
sinic
Posts: 438
Joined: 30 Jun 2012 06:13

Re: Slot-and-filler vs templatic

Post by Sequor »

"Nonconcatenative morphology" is another key term.
hīc sunt linguificēs. hēr bēoþ tungemakeras.
User avatar
eldin raigmore
korean
korean
Posts: 6352
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: Slot-and-filler vs templatic

Post by eldin raigmore »

There seems to be some confusion among professional or academic linguisticians.

Or maybe, I should say, when I made that post I was wondering whether they were confused or had merely confused me;
now that I’ve done more research I realize that the confusion originates with them.

Some use the term “templatic morphology” to refer to the type of nonconcatenative morphology that 3Cons have.
The triconsonantal root is the template, and the transfix is the filler.

Others use the term “templatic morphology” for the extremely concatenative type of morphology also known as slot-and-filler morphology. These are for languages whose word-building can be described by a regular expression without any Kleene stars, or finite-state-automata without any cycles or recursion of states.

Since the latter use is the one I originally thought was meant, it is clearly the better usage!
[;)]

....

(Really now that I’ve read the explanation I can see why someone might refer to the 3Cons root as a “template”. But I don’t think we need another term for that phenomenon, and I do think we need a better term for “slot-and-filler”, so I’m going to call “slot-and-filler” morphology “templatic”.)

......

As I understand it, in most human natlangs in which words are strings of morphemes, and in all compiler languages, the “lexer” — the part that recognizes words and assigns them to parts of speech — is (in the case of compiler languages) or can be theoretically imitated by (in the case of natlangs with concatenative morphology) a Type 3 grammar on the Chomsky Hierarchy; able to be recognized by a finite-state automaton with no extra memory such as a push-down stack nor a Turing “tape”.
Spoiler:
In compiler languages the parser, which decides on the grammaticality of an utterance and also decides its semantics, is a Type 2 grammar on the Chomsky Hierarchy, and can be implemented by a Push-Down Automaton, an FSA with a push-down stack for memory. In other words, a context-free generative grammar, writeable for instance in Backus-Naur form.

Originally Chomsky hypothesized that natlangs’ syntaxes could also be represented by CFGs.
And that may be true of some of them.
But the emerging consensus is that some natlangs’ syntaxes require a slightly more powerful kind of grammar, called “mildly context-sensitive grammars”, that are not Chomsky Type 2.
Chomsky Type 1 grammars, however, (that is, context-sensitive grammars in full generality), include many that are way wilder than anything encountered among natlangs.
Last I heard “they” were still working on what “mildly context-sensitive” means, exactly.
But anyway, the point is, that morphology in natural languages whose morphology is strictly-speaking “concatenative”, can be handled by regular expressions; and a subset of these, namely those which don’t require recursion at the morphology level, have been called “slot-and-filler” morphology and also called “templatic” morphology.

.....

So I guess I’ve answered my own question:
If the answer is:
Calling 3Cons “templatic” introduces unneeded and inconvenient polysemy to the term “templatic”, and it should probably be used as a synonym for “slot-and-filler” instead. (Though one might remark that such a synonym, while IMO convenient, is questionably necessary!)
But leaving aside considerations of necessity and convenience, “templatic” reasonably could apply to both types.
User avatar
eldin raigmore
korean
korean
Posts: 6352
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: Slot-and-filler vs templatic

Post by eldin raigmore »

Creyeditor wrote: 24 Oct 2020 11:27 This question would require a whole into class to morphology as an answer [:D] Maybe someone can write one aimed at conlangers if he/she/they have the time.
Just answer the first five lines. Or as many of those first five as you can and/or want to.
eldin raigmore wrote: 24 Oct 2020 08:34 What is slot-and-filler morphology,
and what is templatic morphology?
And how do they differ?
From each other,
and from triconsonantal-root morphology,
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: Slot-and-filler vs templatic

Post by Creyeditor »

eldin raigmore wrote: 24 Oct 2020 18:01
Creyeditor wrote: 24 Oct 2020 11:27 This question would require a whole into class to morphology as an answer [:D] Maybe someone can write one aimed at conlangers if he/she/they have the time.
Just answer the first five lines. Or as many of those first five as you can and/or want to.
eldin raigmore wrote: 24 Oct 2020 08:34 What is slot-and-filler morphology,
and what is templatic morphology?
And how do they differ?
From each other,
and from triconsonantal-root morphology,
AFAIK and as far as I have seen the term used, templatic morphology is a term used in morphological typology to describe languages where morpheme order is describable (only) by non-hierarchical positive declarative statements. You basically stipulate a total ordering among all morphemes. This often coincides with the claim that the morpheme order is independent of syntatic or semantic constraints. Semantic scope should not influence morpheme order. In a way, morpheme order is supposed to be purely morphological. This is an interesting claim for some linguists, since most mainstream theories of generative morphology assume some influence of semantic or syntax on morpheme order. Note that this different from prosodic templates in morphophonological theory, which are used e.g. to derive Arabic and Hebrew 3cons patterns, see below.
A slot-and-filler model of templatic morphology formalizes the templates in the following way. First, a finite number of slots and a total order is posited. This is the non-hierarchical positive declarative statement mentioned above. Now each morpheme is linked to a slot. Each slot can be filled by one morpheme at a time, but in principle several morphemes can be linked to the same slot. It is generally assumed that these morphemes block each other, so that we do not actually get a total order of all morphemes. It is also important to mention that the linking between slot and filler is ideally completely accidental, i.e. not influenced by semantic or syntactic factors.
Triconsonantal root morphology is a language specific property of certain languages. The phonological shape of a word is determined by morphological/lexical factors in three ways. The root contributes (usually) three consonants. The morphology contributes vowels and a prosodic template. This template basically determines in which way the consonants and vowels are interspersed. It is called prosodic because it influences word shape and syllable structure- It is called a template because it is a positive declarative non-hierarchical statement.

All of this may be wrong or slightly inaccurate, but I answered to the best of my knowledge.
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
eldin raigmore
korean
korean
Posts: 6352
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: Slot-and-filler vs templatic

Post by eldin raigmore »

All of this may be wrong or slightly inaccurate, but I answered to the best of my knowledge.
Looks right to me, thanks!

I think in some morphologies, in particular tlHingaan morphology, semantic scope does count.
There is a negating morpheme that can negate just one of the word’s other morphemes.
Maybe it’s the one just before it; maybe it’s the one just after it.
At any rate this negating morpheme is a “floater”; there are several places in a word it can be used, and though I can’t remember an example I think it can have two or more instances in the same word, if two or more of the word’s other morphemes need negation.
I don’t know of a natlang example of that.

In some (possibly only theoretical?) morphologies, strings of morpheme-types can be repeated in a single word.
This would be the result of applying a Kleene star to some sub-expression of a regular expression that already involved more than one type.
So this would be more complicated than the idea (I think) you said of having two or more consecutive occurrences of morphemes of the same type in a given slot.
I don’t think I’d call that ‘templatic” nor “slot-and-filler”.

But the idea of optionally having two or more consecutive morphemes of a given type, is more complex than the morphologies I have recently become interested in. In effect, for them, the Kleene star is available as the innermost operation; but once some operation has been applied, the Kleene star can’t be applied subsequently.
I think I would call those “templatic” or “slot-and-filler”.

The morphologies I have been thinking about lately wouldn’t allow more than one representative of the class of morpheme that goes in a given slot, to be put in that slot simultaneously with each other, to make an acceptable, well-formed word.
Within the words of any given part-of-speech, any two morphemes of different types that both occur in the word, would have to occur in a particular order that depends (only) on their types.
However probably most slots are optionally empty.
At least one would have to be mandatory, though.

....

I suppose a morphology could be “almost templatic”.
Maybe 99.9% of the time questions about which kinds of morphemes can go where in which kinds of words, and which orders which kinds of morphemes can come in in words, can be deterministically answered by knowing the class of the word and the type(s) of the morpheme(s).
And the other 0.1% are handled in some other, more complicated manner.

....

It’s possible, or at least it seems possible to me, that somewhere among natlangs and/or conlangs, there is a morphology describable by a CFG but not by a finite-state automaton.
If there is I either never saw it or can’t remember a thing about it.

.....

How would one go about guaranteeing that no two distinct words shared an initial substring that was more than 2/3 of one and more than 3/4 of the other?
And/or no two distinct words shared a final substring that was more than 3/5 of one and more than 2/3 of the other?
Etc.?
If there is an answer is it related to these questions?
What does any of this post do to compound words?
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: Slot-and-filler vs templatic

Post by Creyeditor »

I just wanted to point out some natlang examples that you might want to look at.
eldin raigmore wrote: 24 Oct 2020 22:18 I think in some morphologies, in particular tlHingaan morphology, semantic scope does count.
Just wanted to mention that some Bantu languages have sets of voice morphology that can occur in either order yielding different semantic interpretations according to the scope. These might also count as repepetition of morpheme classes.

eldin raigmore wrote: 24 Oct 2020 22:18 In some (possibly only theoretical?) morphologies, strings of morpheme-types can be repeated in a single word.
IIRC, Turkish allows for repetitions of the passive marker. There is also less inflectional examples like German Ur-ur-ur-großvater "Great great great grandparent"
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
eldin raigmore
korean
korean
Posts: 6352
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: Slot-and-filler vs templatic

Post by eldin raigmore »

Creyeditor wrote: 25 Oct 2020 07:52 I just wanted to point out some natlang examples that you might want to look at.
Just wanted to mention that some Bantu languages have sets of voice morphology that can occur in either order yielding different semantic interpretations according to the scope. These might also count as repepetition of morpheme classes.
Didn’t know that! Thanks!
… repepetition …
Is this r-epep-e-titi-on or re-pepe-titi-on?
Maybe it’s just a minor typopographicical error?
[;)]
But if I wrote repetion instead, that would be a different kind of error; a haplogy.
IIRC, Turkish allows for repetitions of the passive marker.
I knew both Turkish and Hindi use double passivization; but didn’t know they involved the same morpheme twice.
I also didn’t think of them. If I had thought of them, maybe I would have looked them up? I guess we’ll never know now!
There is also less inflectional examples like German Ur-ur-ur-großvater "Great great great grandparent"
That wouldn’t have occurred to me. I wouldn’t have thought it an interesting example, so I didn’t think of it at all.
But now I think it is an example, or at least might be one; and might be interesting after all!
Thanks!
User avatar
eldin raigmore
korean
korean
Posts: 6352
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: Slot-and-filler vs templatic

Post by eldin raigmore »

Not slot-and-filler, but speaking of semantic scope.
...
A conlang in which the morphemes are each assigned a one-digit rank from 0 to 9.
A morpheme with even rank has scope over all the lower-ranking morphemes after it (“to its right”) up to but not including the next morpheme with the same or higher rank.
A morpheme with odd rank has scope over all the lower-ranking morphemes before it (“to its left”) back to but not including the last previous morpheme with the same or higher rank.
....
That probably belongs in “morphology ideas”, or whatever the thread-name is, rather than here.
Post Reply