Conlang Random Thread

[jal]

I'm too depressed to keep working on this.

I'm sorry to hear that.


JAL

[bradrn]

Thanks everyone who replied!

While people have been good at explaining what a (past) imperfective is, the specific term ‘imperfect’¹ is a lot more confusing. It was originally used for a specific Latin tense that happened to mostly match a past imperfective, and then got borrowed across Europe for various past tense/aspect combinations, including cognate ones where the meaning had diverged², and so it is a mess with no clear meaning. Because of this is has fallen out of use except as a name for specific verb forms in languages that used it, and the clearer (im)perfective terminology was introduced.⁴

Ah, so basically this is a term that conlangers don't really need to use for their grammars, because one can use other, better terms instead? (Unless you're doing some althistory conlanging where it makes sense to use the same terms as the conlang's parent/sisterlangs.)

In short, yes, although noöne will care what you call [verbform x], as long as the name doesn't actively create confusion.

This is really the key to most terminological debates. The function of a morpheme is about its syntactic and semantic distribution, not the term you happen to choose to summarise that distribution.

(Of course it’s nice for names to correspond to the underlying function in some way, but the alternative isn’t actually wrong, merely unhelpful.)

[xxx]

it's the curse of language...

like Echo, condemned,
repeating others says,
giving them new meanings...

what metamorphosis,
to be a conlanger,
the Gods condemn us to...

perhaps like Narcissus...

[Richard W]

In short, yes, although noöne will care what you call [verbform x], as long as the name doesn't actively create confusion.

This is really the key to most terminological debates. The function of a morpheme is about its syntactic and semantic distribution, not the term you happen to choose to summarise that distribution.

Terminological question: Does a tense have a morpheme? It feels hard to say that the English simple past has a morpheme, and this is not an isolated example - good examples can be found in all of the 3 classical IE languages.

[quinterbeck]

Terminological question: Does a tense have a morpheme? It feels hard to say that the English simple past has a morpheme, and this is not an isolated example - good examples can be found in all of the 3 classical IE languages.

Does a tense have a morpheme? Not necessarily.
Can a tense have a morpheme? Yes!
Is a tense morpheme present wherever that grammatical tense is expressed? Not necessarily.

I would pretty confidently say that the English simple past has a morpheme <-ed>, and that English has one morphological class of verbs that take <-ed>, and other classes* of verbs that express the simple past with different morphological strategies.

This might come down to the semantics of 'have a morpheme'.

*well, at least one other class

[bradrn]

This might come down to the semantics of 'have a morpheme'.

Pretty much, yes. In English I would say that <-ed> is a morpheme whose sole function is as a past tense; other languages may have different arrangements with more complicated form-to-function correspondences.

[jal]

Terminological question: Does a tense have a morpheme? It feels hard to say that the English simple past has a morpheme, and this is not an isolated example - good examples can be found in all of the 3 classical IE languages.

Let's first get clear what a morpheme is: it's the smallest bearer of information in a word (and let's not get into the definition of "word" for now). A morpheme can be free, in which case it can be used on its own, i.e. as a stand-alone word, or bound, in which case it can only be used attached to another word (typically a free morpheme, though exceptions exist) - in this case it's typically an affix.

A tense (or any other grammatical category) can be expressed by a morpheme, like English past tense marker -ed (which has phonetic allopmorphs /d/ and /t/). I don't think it's correct to say the tense "has" a morpheme, if we're talking about terminology.

I'm not sure why you say that the English simple past isn't expressed by a morpheme (or at least that that feels "hard to say"), as it's clearly -ed. That doesn't mean that there aren't other possible ways to form the simple past - many verbs have a supletive form, and some verbs have more than one (e.g. "to be" with "was" and "were"). But note that these suppetive forms are themselves morphemes as well!

As for the "three classical IE languages", are you referring to Greek, Latin and Sanskrit? Those are all inflectional languages (of course, because IE), so they typically don't have a single morpheme for tenses (or aspects, or moods), but morphemes that at least include number and gender (and often tense, aspect and mood combined as well). Looking at other language families, it's often easier to point to morphemes indicating past tense; especially in agglutanative languages, tense, aspect and mood, as well as number and gender, are often seperate morphemes.


JAL

[jal]

Pretty much, yes. In English I would say that <-ed> is a morpheme whose sole function is as a past tense

<-ed> doubles as a morpheme to form the past participle. So the past tense is not its sole function.


JAL

[Richard W]

This might come down to the semantics of 'have a morpheme'.

Pretty much, yes. In English I would say that <-ed> is a morpheme whose sole function is as a past tense; other languages may have different arrangements with more complicated form-to-function correspondences.

And then there's <-en>, which marks the past participle, and some would say in weak verbs generally has the same form as <-ed>.

Of course, for English, there's the claim that <i>dreamt</i> and <i>dreamed</i> are not synonymous, but have a subtle difference in aspect.

[bradrn]

Pretty much, yes. In English I would say that <-ed> is a morpheme whose sole function is as a past tense

<-ed> doubles as a morpheme to form the past participle. So the past tense is not its sole function.

Aargh, yes, I knew I was missing something. (The perils of trying to write on a phone…) So let’s say that ‘past participle’ and ‘past tense’ are expressed by two different morphemes with overlapping allomorphs.

(I agree with your previous post, too, though I tend to regard the ‘inflectional’ vs ‘agglutinative’ distinction as oversimplified.)

[jal]

(I agree with your previous post, too, though I tend to regard the ‘inflectional’ vs ‘agglutinative’ distinction as oversimplified.)

One cannot describe languages in general without oversimplification :D.


JAL

[jal]

And then there's <-en>, which marks the past participle, and some would say in weak verbs generally has the same form as <-ed>.

The past participle is marked by -ed, just like the past tense. -en isn't synchronically a past participle morpheme, I consider forms like "written" to be supletive.


JAL

[Richard W]

Terminological question: Does a tense have a morpheme? It feels hard to say that the English simple past has a morpheme, and this is not an isolated example - good examples can be found in all of the 3 classical IE languages.

Let's first get clear what a morpheme is: it's the smallest bearer of information in a word (and let's not get into the definition of "word" for now). A morpheme can be free, in which case it can be used on its own, i.e. as a stand-alone word, or bound, in which case it can only be used attached to another word (typically a free morpheme, though exceptions exist) - in this case it's typically an affix.

A tense (or any other grammatical category) can be expressed by a morpheme, like English past tense marker -ed (which has phonetic allopmorphs /d/ and /t/). I don't think it's correct to say the tense "has" a morpheme, if we're talking about terminology.

I'm not sure why you say that the English simple past isn't expressed by a morpheme (or at least that that feels "hard to say"), as it's clearly -ed. That doesn't mean that there aren't other possible ways to form the simple past - many verbs have a supletive form, and some verbs have more than one (e.g. "to be" with "was" and "were"). But note that these suppetive forms are themselves morphemes as well!

Saying that an English strong verb contains /-ed/ in its simple past feels wrong, and doesn't do well with synonymous simple past tense forms such as arrived and arrove. It gets even worse when one works with a strongly templatic language such as the Semitic ones.

As for the "three classical IE languages", are you referring to Greek, Latin and Sanskrit? Those are all inflectional languages (of course, because IE), so they typically don't have a single morpheme for tenses (or aspects, or moods), but morphemes that at least include number and gender (and often tense, aspect and mood combined as well). Looking at other language families, it's often easier to point to morphemes indicating past tense; especially in agglutanative languages, tense, aspect and mood, as well as number and gender, are often seperate morphemes.

Yes; are not they the 'classical' IE language? What had in mind was:

For Greek: The first and second aorists, which are formed from the stem quite differently and have different endings. The former adds /s/, which may have been absorbed. The latter instead uses zero grade of the stem and has the inflections of the imperfect. Some verbs have both, I think in general without any difference in meaning.

For Latin: (i) The future has -bō, -bis, ... in the first and second conjuɡation, but -am, -ēs, ... in the third and fourth conjuɡations.
(ii) The perfect has several modes of formation - adding /v/ to the stem, adding /s/ to the stem, and reduplication.

For Sanskrit: The 'aorist' has both thematic and athematic forms. I'm not sure whether the synthetic and periphrastic futures are synonymous; I don't recall a claim of subtle differences. If we regard tenses as functional definitions, then in Classical Sanskrit the preterite is composed of the inherited imperfect, aorist and perfect, which are of course formed quite differently.

[Richard W]

And then there's <-en>, which marks the past participle, and some would say in weak verbs generally has the same form as <-ed>.

The past participle is marked by -ed, just like the past tense. -en isn't synchronically a past participle morpheme, I consider forms like "written" to be supletive.

I hope you mean 'irregular'; 'suppletion' means the substitution of a different stem altogether, as in the usual singular-plural pair person and people.

How do you express the analyses of wrote and written into morphemes?

[þeprussianfrog]

Yes; are not they the 'classical' IE language?

Middle Persian probably counts as well, I guess?
(Old Persian is too poorly attested)

[bradrn]

Saying that an English strong verb contains /-ed/ in its simple past feels wrong, and doesn't do well with synonymous simple past tense forms such as arrived and arrove. It gets even worse when one works with a strongly templatic language such as the Semitic ones.

‘Arrove’? Never seen or heard that one…

More generally, I think you might find it interesting to read more about current theories of morphology. Corbett’s very interesting book Agreement has a concise overview:

We can approach the different types of morphological theory from a basic question: how do we relate an inflected form, like English sits, to the stem sit? I first distinguish lexical theories from inferential theories. In lexical theories, the affix -s has a lexical entry, which specifies it as 'third-person singular subject agreement', 'present tense' and 'indicative mood', rather as sit has a lexical entry. Thus sits corresponds to two lexical entries. In inferential theories, on the other hand, the systematic relations between a stem like sit and an inflected form like sits are expressed in terms of rules or formulas. The existence of an inflected form like sits is inferred from the existence of sit by a rule which associates the appearance of -s with the feature specification 'third-person singular', and so on.

There is a second distinction, cross-cutting the first, which contrasts incremen­tal and realizational theories. In incremental theories words gain morphosyn­tactic feature values only together with the relevant exponents. This means that sits gains the feature values 'third-person singular subject agreement', 'present tense' and 'indicative mood' only by adding -s (whether -s is inserted from the lexicon or is introduced by a rule or formula). In realizational theories the stem sit is associated with a particular set of morphosyntactic feature values ('third-person singular', and so on) and this licenses the introduction of the inflec­tional exponents for them (whether this is by lexical insertion or by applying a rule).

[…]

Stump (2001: 3-12) evaluates the four types of theory and concludes that the data point towards inferential-realizational theories. […] In such theories a stem is associated with a morphosyntactic feature specification, and these license the inflected forms.

By contrast you seem to be implicitly working from some sort of ‘lexical-incremental’ theory in which each ‘morpheme’ is its own autonomous entity, and features are added to a word only by adding that ‘morpheme’ to a stem. That works well enough for very regular agglutinative languages, but as you can see it quickly runs into problems with others. It’s probably for this reason that modern linguists seem to avoid the term ‘morpheme’ except in informal use — indeed, Corbett only uses the term twice in the whole book. (See also Anderson’s 1992 book A-Morphous Morphology.)

[jal]

I hope you mean 'irregular'; 'suppletion' means the substitution of a different stem altogether

You're right, "suppletion" is not quite the right term here, as "written" is diachronically related to "write".

How do you express the analyses of wrote and written into morphemes?

They're their own, free, morphemes. Words with an ablaut ("wrote") or a synchronically unanalyzable form ("written") do not have morphemes other than the main one.


JAL

[Richard W]

In short, yes, although noöne will care what you call [verbform x], as long as the name doesn't actively create confusion.

This is really the key to most terminological debates. The function of a morpheme is about its syntactic and semantic distribution, not the term you happen to choose to summarise that distribution.

Terminological question: Does a tense have a morpheme? It feels hard to say that the English simple past has a morpheme, and this is not an isolated example - good examples can be found in all of the 3 classical IE languages.

By contrast you seem to be implicitly working from some sort of ‘lexical-incremental’ theory in which each ‘morpheme’ is its own autonomous entity, and features are added to a word only by adding that ‘morpheme’ to a stem. That works well enough for very regular agglutinative languages, but as you can see it quickly runs into problems with others. It’s probably for this reason that modern linguists seem to avoid the term ‘morpheme’ except in informal use — indeed, Corbett only uses the term twice in the whole book. (See also Anderson’s 1992 book A-Morphous Morphology.)

You're the one who switched from the naming of tenses to the meaning of morphemes.

An issue with the realisational approach is that it allows one to claim person and number for English relative pronouns. While that is probably correct for several forms, it does allow a great deal of other marking.

I have been pondering whether almost every English word can have a possessive of some form. This comes about because of the group genitive, which get complicated for words that already have arguable marking in <s>.

[Richard W]

How do you express the analyses of wrote and written into morphemes?

They're their own, free, morphemes. Words with an ablaut ("wrote") or a synchronically unanalyzable form ("written") do not have morphemes other than the main one.

Are you saying that write, wrote and written have but a single morpheme between them, or are you saying they're three different morphemes?

[Travis B.]

I hope you mean 'irregular'; 'suppletion' means the substitution of a different stem altogether

You're right, "suppletion" is not quite the right term here, as "written" is diachronically related to "write".

How do you express the analyses of wrote and written into morphemes?

They're their own, free, morphemes. Words with an ablaut ("wrote") or a synchronically unanalyzable form ("written") do not have morphemes other than the main one.

-en is a semi-productive morpheme in my dialect. For instance, new strong past participles with -en can be derived from strong preterites in cases, such as aten and dranken. (Yes I will say and have heard both of these, and they did not sound ungrammatical to me or to the person with whom I was speaking.)