How to stop fucking up triconsonantal languages for novices

[Yng]

How to stop fucking up triconsonantal languages for novices
Or: Discontinuous morphology for beginners

Inspired by Whimemsz's excellent thread on polysynthesis, I thought it might be useful to have a similar, reasonably detailed guide to another kind of language that with the best will in the world new conlangers often fuck up incredibly - in this case, triconsonantal root languages modelled on the Semitic language family. Almost all of us have been tempted - me included - and have produced predictably awful results. Mec's previous thread is very useful, and I'll be drawing on information found there quite a bit - but times have changed, there's a lot more expertise around the board and in any case, a more accessible, organised summary without the intervening discussion and occasional uninformed interruption (everyone seems to have been a bit stupider back then) will probably be helpful.

However, I don't want this to be simply a discussion of triconsonantal root languages. Although there is significant variation within the Semitic family, any language with the premise of triconsonantal roots alone, if done well, will probably be little more than a Semitoclone (perhaps with some extra features not found in Semitic - ergativity, say). Instead, I want to move away from a fetishisation of the Semitic system - which is a particularly mad and all-pervasive example - and discuss other types of discontinuous morphology.

Like Whim's thread, there will obviously be some jargon and some reasonably srs linguistix - but if you're not familiar with the terms used and left undefined, chances are you need to learn some more of the basics before having a read of this.

What is discontinuous morphology?

Discontinuous or non-concatenative morphology is a morphological behaviour in which a language, rather than straightforwardly affixing morphemes to one another (perhaps with some surface distortion because of morphophonology), modifies the form of lexical morphemes or, sometimes, inserts morphemes within the boundaries of other morphemes (i.e. infixing). There are various different types of discontinuous morphology. The Semitic system is the most advanced and pervasive type which I am aware of and is usually analysed as the insertion of root consonants into a given derivational pattern, which has slots ready for them:

كتب
k{a}t{a}b-a
write{PERF}-3sg.MASC
he wrote

كاتب
k{ā}t{i}b
write{ACT.PART}
writer

كتّب
k{a}t{ta}b-a
write{CAUS_PERF}-3sg.MASC
he made (someone) write

We will discuss the Semitic system in more detail shortly and how it fits in with other kinds of discontinuous morphology. Closer to home, we have less productive systems like English's strong verbs, or various other (now frozen) vowel patterns:

wr{o}te
write{PRET}

f{e}ll
fall{CAUS}

Tagalog makes use of infixes to control valency:

bumilí ng saging ang lalaki
b{um}ilí ng saging ang lalaki
buy{PATIENT} IND banana DIR man
The man bought bananas

binilí ng lalaki ang saging
b{in}ilí ng lalaki ang saging
buy{AGENT} IND man DIR banana
The man bought the banana

Bambara modifies tones to mark definiteness:

bâ
b{â}
river{definite}
the river

bá
b{á}
river{indefinite}
a river

Vietnamese, on the other hand, has traces of tone being used to produce distal demonstratives from proximal ones:

bà nầy
woman this
this woman here

bà nấy
woman that
that woman there

It is clear then that discontinuous morphology is found all over the world. In this summary we'll look at a few different case studies of productive discontinuous morphology.

What is a triconsonantal root

Since this is the kind of discontinuous morphology most commonly seen in new conlangs, since it's the kind I know most about, and since it's there in the thread title - and simply for its whack value - the Semitic system deserves a brief in-depth treatment of its own.

The Semitic system is often quite profoundly misunderstood, or at least, in my opinion, misanalysed. Brief summaries of the kind found on Wikipedia follow the native Arabic grammatical tradition and Western pedagogical materials in describing Semitic languages as having a 'root-and-pattern' system where roots - sets of three or occasionally four consonants, possessing an inherent, abstract meaning such as 'writing' - are inserted into patterns of vowels and sometimes consonants, themselves carrying largely predictable meanings like 'place of', to produce an Arabic word. This is often a useful approach pedagogically, but in my opinion it is more useful to conceptualise roots not as meaningful units in themselves, but rather as something extracted from an input word and inserted into a new pattern to produce an output word, with the meaning taken from the input word and augmented by the derivational process (just as in any other language). Taking into account the input word not only allows us to explain new coinages and theorise about the etymologies of specific words (maktabah 'library' < kitāb 'book'), it also allows us to explain why, a given noun, for example, takes a given plural form (the Arabic broken plural is formed by discontinuous morphology and to a very significant extent the choice of plural form is conditioned by the shape of the singular noun). When we look at the system this way, it becomes nothing more than a particularly complex form of discontinuous morphology, where the consistent element between the input word and the output word is what we call the triconsonantal root.

This interpretation is supported by the fact that native speakers are perfectly capable of coping with words which do not have a clear root-pattern correspondence, be they loanwords (banafsajan 'violet', telfīziyyūn 'television') or unanalysable native terms. Furthermore, natives often reanalyse words' roots (taʾaslama 'be Islamised' has the root ʾ-S-L-M, but is derived from ʾislām, which is a regular verbal noun from the word ʾaslama which has the root S-L-M), or extract roots from loanwords (film has a broken plural ʾaflām). We will discuss new coinages, innovations and analogies within Semitic later.

[Yng]

Where does discontinuous morphology come from?

[Yng]

Case study: Semitic verbal derivations

All Semitic languages make considerable use of discontinuous morphology - Arabic in particular, but Hebrew and Aramaic too. Their systems of verbal derivations - called 'binyanim' in Aramaic and Hebrew and 'measures' or 'forms' in Arabic - are particularly ripe for discussion and for comparison. These forms are historically largely (although not entirely) valency modifiers, and we can still see many examples of relatively straightforward valency-based derivations in the modern Semitic languages. However, seeing them as inflections rather than derivations is unwise; whilst some of them blur the line between inflectional and derivational morphology, whether a given form exists or not is essentially conventional and often verbs have unpredictable meanings due to semantic drift. Furthermore, at least in Arabic many of these forms can be used to produce denominative verbs; what looks like a 'passive' or 'reflexive' verb with a ta- prefix may not have an active or non-reflexive form.

...

[Yng]

Case study: Germanic umlaut and ablaut

[Yng]

Case study: Tonal morphology

[Yng]

Case study: Reduplication

[Yng]

Case study: Austronesian infixes

[Yng]

The power of analogy: Innovations and borrowings

[patiku]

Wow. That's all I can say. Please write more, I'm really looking forward to your posts. *microwaves some popcorn*

[Bristel]

Thanks for this Yng, I look forward to reading the thread. I'm working on a 3con language right now, but I have no idea where to start, or how to make it less like Hebrew.

[Whimemsz]

so is anything happening with this or what

[Astraios]

All Semitic languages make considerable use of discontinuous morphology - Arabic in particular, but Hebrew and Aramaic too.

How do Hebrew and Aramaic make less use of it than Arabic?

at least in Arabic many of these forms can be used to produce denominative verbs

Also in Hebrew and Neo-Aramaic.

[Khvaragh]

All Semitic languages make considerable use of discontinuous morphology - Arabic in particular, but Hebrew and Aramaic too.

How do Hebrew and Aramaic make less use of it than Arabic?

The Arabic grammatical tradition, or naḥw, is a massive one, directly comparable in pre-modern times to perhaps vyākaraṇa in India (perhaps also the Latin and Greek traditions, but I'm not aware of them being as elaborate); and like the Indian version, emerged out of exegetical study. There is considerable artifice, shall we say, in the system, because of its religious significance. There is a large tendency in Arabic grammar to treat roots and patterns as being Cartesian categories, with a predictable or at least rationally justifiable outcome from inserting X root into Y pattern. Based on my study of Arabic grammar (as well as the opinions of linguists, see for example the article on Arabic in Comrie's "the World's Major Languages," this tendency is extremely pervasive, and much more so than in Hebrew and Aramaic. These two languages do of course have elaborate derivation morphology, but Arabic is extreme. There's also the issue of the collapse of certain roots into each other by sound change, which can obscure the derivation, and hence they are not obvious synchronically, only through diachronic comparison. Keep in mind that the grammatical traditions of Hebrew and Aramaic (the only really significant one of the latter is for Syriac) are posterior to the Arabic one and largely influenced by it (for example, Sa`adia Gaon's Hebrew linguistics was clearly heavily influenced by the Arabic one); there is less prescriptive artifice (or the attempt to justify X root in Y pattern, etc.) in the Hebrew grammatical tradition simply because it isn't as large or old. You will even find in grammars of Hebrew the warning that the meaning of a given word is not always predictable from its root and pattern. Also, in terms of a specific example, Hebrew and Aramaic both have regular (or relatively regular) plural formations, whereas Arabic makes use of discontinuous morphology in the vast majority of its plural forms (as does Ge`ez, but the number of broken plural patterns is smaller). The segolate plurals of Hebrew (Aramaic has less; they occur mostly when affixes like pronoun suffixes are attached to the morpheme, and these can be singular or plural words) seem to mostly derive from the loss of short vowels or epenthesis, etc. due to syllable weight (though there are some arguments that some of the segolates do preserve ancient PS broken plurals).

[Astraios]

You will even find in grammars of Hebrew the warning that the meaning of a given word is not always predictable from its root and pattern.

I'll grant you the rest but this one is surely true of Arabic too. Like وقع 'happen' in form I which is وقع 'to sign' in II, how is that predictable? Unless it's one of the collapsed distinctive roots, but.

[Yng]

uhhhh I basically just meant that in Arabic it pervades everything - there's very little morphology which isn't at least partially ablaut-based. Hebrew has the 8 binyanim, but the majority of Arabic's plurals are formed by

also guys cool your jets it'll happen OK

[Khvaragh]

You will even find in grammars of Hebrew the warning that the meaning of a given word is not always predictable from its root and pattern.

I'll grant you the rest but this one is surely true of Arabic too. Like وقع 'happen' in form I which is وقع 'to sign' in II, how is that predictable? Unless it's one of the collapsed distinctive roots, but.

With certain verbs, there is a very wide range of semantic possibilities. وقع doesn't just mean "happen," it also means "fall," "occur," "meet," "alight," and many other idiomatic meanings. Form II also has the causative meaning of "cause to fall, push down." It's difficult to know which one of these meanings is primary, which one is derived by analogy, and which is a construction of the grammatical tradition. Please note that I many not saying that all Arabic roots have predictable meanings, rather that there are quite a few, more so than in most other Semitic languages I'm familiar with. I attribute most of this predictability to a combination of massive analogy with prescription of the grammatical tradition.

And yes, as Yng states, Arabic has very little morphology which isn't ablaut-based; more so than the other languages mentioned.

[Whimemsz]

tick








tock

[Culla]

I would like to bump this as well. I want to work on a language with a bit of non-concatenate morphology and this would be a nice intro before I work on.

[Dewrad]

no pressure, Yng. no pressure.

[Kereb]

[Trebor]

I would like to bump this as well. I want to work on a language with a bit of non-concatenate morphology and this would be a nice intro before I work on.

And I'm seriously considering adding grammatical pitch-accent/tone to my conlang...

Plus, I'm interested in finding out more about Bambara, which I hope to learn someday.

[sirdanilot]

Posting to say I approve of this thread and would like to see it finished !