(L&N) Q&A Thread - Quick questions go here

[Dormouse559]

Just a little hypothetical I thought of this morning:

The Romance languages derive their words for "liver" from Latin ficatum, meaning "fig-stuffed". Evidently there was some kind of medieval dish called iecur ficatum "fig-stuffed liver". The actual word for liver, iecur, dropped out of usage entirely and ficatum was reanalyzed to mean liver as food, and eventually the liver as a body part (hence Sp. hígado, It. fegato, Fr. foie, Ro. ficat). iecur is an odd Latin word with an odd structure (an r/n-stem neuter that doubled up both stems in the oblique, like iter), so I'm not surprised it has no modern descendants. But if it did, what would they look like? What would the alternate Spanish, Italian, and French words for liver be?

Thanks for any input

Yay diachronics! I think French would give /ʒiʁ/, probably spelled <gire>. Latin /ɛ/ likes to become French /i/ in contact with palatals, and iecur puts one on either side. The spelling with <g> instead of <j> is based on inherited French words with /ʒi/ (gîte "shelter", gît "[he/she/it] lies, is located"); I'd guess that's a disambiguating choice from the time when <j> and <i> weren't distinct letters.

[Salmoneus]

Well, this is a stab in the dark because I'm certainly not an expert, but I'm guessing that in Spanish it would now be juezno.

Why do I think this?

1. The Appendix Probi warns people to say 'iecur non iocur'. AP prohibited pronunciations can't be taken as universal and successful late latin forms, but they're generally more likely than not to have won out. So I'll assume that iecur>iocur.

2. Iocur/iocinoris is a really weird declension. However, 3rd declension neuter nouns with -oris genitives aren't weird, they just usually have -us in the nominative (tempus/temporis, etc). So it would seem natural that at some point irregular iocur/iocinoris > regular iocinus/iocinoris.

3. Neuters were geneally lost; in the West, 3rd declension neuters in -us were assimilated into 2nd declension neuters in -us, and thence into 2nd declension masculines in -us. Either because the accusative in -um was then taken as the basis of further evolution, or becuase the -s was reanalysed as a plural ending and removed, these nouns end up in Spanish ending in -o.

4. The first syllable of iocinus is the same as of 'iocus', and we can expect a similar development, with fricativisation and diphthongisation. The second and third syllables are the same as of the suffix -icinus, which yields -ezno in Spanish. It's possible that the difference of the vowel preceding *k would change how it palatalises, but I can't see any suggestion that this was the case, so I assume that the reflex would be juezno.

[KaiTheHomoSapien]

Thanks for your answers, guys. Very interesting :D This is the kind of hypothetical linguistics I love.

[VaptuantaDoi]

Following Sal's conjecture, the French word would probably evolve like *iocinus > /dʒu̯ɔi̯znə/ > /ʒɥin/ probably written juîne. Finding examples of pre-consonantal elided CI is harder in French, but from examples like *COCERE > cuire and *DECIMUS > dîme it looks like it just generated a yod (or yod + /s/ which ends up as the same thing in modern French). I'm less sure about the Italian, but maybe giuocino /ˈdʒu̯ɔtʃino/.

[Creyeditor]

Italian could also be giocino, right? Judging from the fact that game is gioco in Standard Italian and giouco is considered archaic.
Also, I really like how the similarity of joke, game words and liver words could yield to a reanalysis keeping the idea from Latin that the liver is tbe seat of passion.

[Omzinesý]

Genitive often derives from an old dative or ablative, like English "of".
It is very common that ablatives or datives are used as several kind of adverbials. Source and goal actually are adverbials.
But the IE s-genitive is ancient. And in modern Germanic languages, it is basically exclusively used for possessors. In German it some adpositions govern it.
But, probably dropping the adposition from a PP is more common than what I have thought.

Thank you for answers.

[Salmoneus]

Genitive often derives from an old dative or ablative, like English "of".

English 'of' neither is nor is derived from an old dative or ablative. [it derives from an adverb, ultimately, as most or all English prepositions do (thanks, word order changes!)]

But the IE s-genitive is ancient. And in modern Germanic languages, it is basically exclusively used for possessors.

At least in the case of English, to the extent that that is true (it's also still used adjectivally ["a summer's day"], and nominatively ["my reading the book", "the city's destruction"]), it's largely because pretty much all the old uses of the genitive have been shifted over to the "of" construction.

In German it some adpositions govern it.
But, probably dropping the adposition from a PP is more common than what I have thought.

It's best not to see languages as deviations from German or English. Using the genitive for functions other than possession - partitives, adjectives and adverbs, compositives, measuring and counting, ergatives, etc - is extremely common cross-linguistically and is not generally best advertised as simply dropping a genitive-governing preposition from a prepositional phrase.

[Omzinesý]

English "of" still expresses source 'from' in many contexts, i.e. ablative function.

[All4Ɇn]

English 'of' neither is nor is derived from an old dative or ablative. [it derives from an adverb, ultimately, as most or all English prepositions do (thanks, word order changes!)]

Wow I didn't realize this. Does this mean that verb-preposition phrases were originally just verb-adverb phrases or were they already perceived of as prepositions by the time these started to pop up?

[Salmoneus]

English "of" still expresses source 'from' in many contexts, i.e. ablative function.

It expresses the "of" function, which, yes, does sometimes indicate sources. The ablative is a noun case (which can also sometimes indicate sources). Prepositions are not just expressions of cases, so calling the function of a preposition the "ablative function" or "dative function" or the like is misleading - prepositions and cases may have overlapping semantics, cross-linguistically and within a single language, but need not equate.

[Salmoneus]

English 'of' neither is nor is derived from an old dative or ablative. [it derives from an adverb, ultimately, as most or all English prepositions do (thanks, word order changes!)]

Wow I didn't realize this. Does this mean that verb-preposition phrases were originally just verb-adverb phrases or were they already perceived of as prepositions by the time these started to pop up?

Weeeellll, it's actually a really long and only partially understood story, which I rather simplified there. We can divide this into two parts, the big picture and the small picture.

In the big picture, PIE had nonconfigurational particles, which we can easily interpret as prototypically adverbs. These adverbs could be more semantically linked to noun phrases, to verbs, to both, or to neither, in particular sentences. Over time, these particles have tended to become more configurational, becoming locked in to certain positions, and into certain functions along with those positions, in different ways and to differing extents in different languages, developing into conjunctions, adpositions, adverbs and verbal prefixes. This has resulted in turnover in some situations, with new derivatives of these particles being formed as the particles themselves became locked into certain roles. This is why both within individual languages, and between languages, there can be a whole panoply of words derived from any one of these particles. Proto-Germanic however was a bit conservative, and the distinctions between parts of speech were not quite as clearcut as in, say, Classical Latin.

[this origin story explains many oddities. The pre-verbal adverbs tended to be interpreted as postpositions in SOV languages, but were free to move into preverbal position in other languages. They often came to be semi-attached to the verb, forming separable verbs, which only later locked into simply prefixed verb (a change that occured in the history of Latin (early Latin prayers still feature separable prefixes), and centuries later in English), and in the process created weird tmetic patterns with captured pronouns and particles (as most famous for their role in creating the nightmare of Old Irish verbs). Adverbs were often more explicitly linked to nouns by placing the noun into a case, the same way prepositions can link adverbs (or adjectives) and nouns today in English [out OF the fire, away FROM the goat, large IN size] - when adverbs came to be reinterpreted as adpositions, this lead to adpositions 'governing' specific cases.]


In the small picture, meanwhile, we can be much more precise. Historically in Germanic, these particles were still found before, and modifying, clause-final verbs (as well as elsewhere). These were originally all separable, but gradually become locked onto the noun. This become permanent during later Old English as the shift from SOV to SVO became overwhelming - the particles came with the verbs, becoming wholly derivational (and often phoneticaly reducing, eventually completely).

However, stronger, more independent adverbs did not come with the verb, but remained in place. These formed a new class of phrasal verb in Middle English: instead of preposition+verb, they were verb+adverb.

Many phrasal verbs are still verb+adverb: look AWAY, throw UP, kick OUT. However, over time:
- many of these adverbs have developed into prepositions in their own right, through the loss of linking prepositions. So adverbs like "outside" or "above" have become prepositions.

- some phonologically distinct preposition/adverb pairs in PGmc, where the adverb was the 'stronger' of the two due to strengthening derivational affixes or simply stress, have merged in English, either by replacement of one by the other or by regular diachronic merger. So Old English allative adverb "inn" and locative adverb "inne" have merged with the Old English preposition "in".

- gradually these two developments (and the loss of case) have led to a merger of the "verb + PP" and "verb+adverb" constructions, with a shift of the object in the latter construction from postverbal to postadverbial, further eroding the adverb/preposition distinction. There are still nuances that can be teased out sometimes, and occasional minimal pairs (adverbial "I ran him through" and prepositional "I ran through him" don't mean the same thing!), but they're increasingly minor.

[there's also the case of 'of', where, contrary to most words, the weak prepositional form ('of') and the strong adverbial form ('off') have actually diverged rather than converged, which is why you can't have phrasal verbs with 'of'... if it's a phrasal verb, it has 'off' instead!]


--------


So, to answer your question, there are basically three types of complex verb, diachronically:
- from PIE through to PGmc or afterward, many adverbs were more or less strongly associated with verbs, forming separable verbs
- where the adverbial element was semantically and prosodically weak, this resulted into prefixal verbs once verbs had decisively moved forward in th sentence. Eg "forgo", "undergo", "betoken", etc.
- where the adverbial element was strong, it remained separate from the verb as a distinct adverb. This then formed a modern phrasal verb - "look up", "look out", "throw off", etc. This could also happen with new adverbs not formed from single particles, like "go down" (where down < adune < ofdune < ab dunai, "off the hill")
- where the PIE adverb was only weakly connected to the verb, it could instead become a preposition governing an object. Common combinations of verb and preposition could then develop their own specific semantics - thus "look into the case", "hear of the events", etc, where "look into" means investigate and "hear of" means become aware of, rather than having their literal meanings.

These have theoretically distinct syntax: the first is a verb with a prefix; the second is a verb with an adverb, which can have an object between them; the third is a verb with a prepositional phrase. However, the three categories have been blurred: adverbs have become prepositions, or merged with them, so many of the second category now look like the third, but more flexible ("I threw down", "I threw it down" or "I threw down the paper" (but NOT "I threw down it"), whereas "I looked into the case", but NOT "I looked into" or "I looked the case into"). Some have also created new versions of the first construction as well (often via nominal or adjectival constructions).

So in theory: the first construction (prefixal verbs) developed at the same time that the adverbs were turning into prepositions when not in that construction; the second construction developed when a new set of adverbs were still adverbs (and some still are); the third construction developed when the old adverbs were already prepositions. But, as I say, some of the boundaries between these classes in individual cases are blurred.



....so, that's my understanding of it, anyway!

[All4Ɇn]

Thank you for the in depth post! Very interesting

[Davush]

I have a question about morpho-phonological processes.

Essentially: can morphemes "dictate" certain phonological processes limited to themselves and/or "take precedence" over normal sound changes? This means that they don't operate as "ordinary" across-the-board sound changes.

For example, let's say there's an "ordinary" sound change in which unstressed high vowels are deleted in open syllables: *túkuma > tókma (with lowering due to new closed syllable). This is straightforward enough, but let's also say it fails to apply in some environments caused specifically by morphemes:

The suffix -ta "dictates" that all final vowels of certain word-shapes are removed before it is applied, so *túkuma+ta yields > túkumta, not tókmta~tókmata (which is the expected outcome according to the "ordinary" sound change above). This means -ta has blocked, or occurs before, the previous rule of vowel deletion. In this language, indiscriminate final-vowel deletion by suffixes is not universal: it only applies to -ta, so this also cannot be an "ordinary" sound change.

Also, the suffix -mi behaves differently again: vowel deletion does not occur when -mi is responsible for producing the environment for it: *xánimi > xánimi, NOT ?xánmi with ordinary vowel deletion. On the other hand, *túkuma-mi does yield tókmami, as the environment for deletion is not caused by -mi itself.

This looks like the morphemes -ta and -mi stipulate special rules only for themselves, which "override" the ordinary, purely phonetic-based sound change (high-vowel deletion).

I know it is often said that sound-changes are purely phonetic and are blind to things like this, but I think similar processes where morphemes stipulate their "own" rules do occur?

Any advice appreciated! Thank you!

[Salmoneus]

In the words of Reverend Lovejoy: short answer yes with an if, long answer no with a but.

No, it's not wise to think of morphemes as dictating their own diachronic rules. However, it's possible to set up regular or irregular sound changes that superficially might look as though that's what's happening.

Your two main ways for doing this:

- correct phasing of sound changes and morphological changes can result in different affixes having different morphophonetics. In a very simple example: if you have the changes "final vowels are lost in polysyllables" and "the preposition *ko is suffixed to nouns", these two historical events can occur in either order, giving different outcomes: in one timeline, *tapa+*ko>*tapko, while in the other *tapa+*ko>*tapak. In the first timeline, vowel loss occured prior to affixation, while in the second timeline the opposite was true. In theory, then, if different suffixes develop at different times, they can have different morphophonetics. Is this likely? Depends how the suffixes relate to one another. If they're very similar in function - say, an allative and an ablative - then they'll probably (though not definitely) have been added at the same time; and even if they were added at slightly different times, their parallelism would likely level the differences through analogy. But if they're very different in function, it's not at all unreasonable that they could have been added at totally different times, and thus at different points in the sound change history. This might look as though the suffixes are dictating different sound changes, but in reality everything is the result of perfectly regular, semantics-agnostic sound changes.

- where some feature develops in some words, it can spread to others by analogy. The feature could develop, for example, due to the phonological oddities of a particular word*, due to borrowing from a different dialect. Analogical spread is more likely from common and important words, when it's from a larger class to a smaller class, and when the spreading form preserves information lost in the expected form. So, for example, if a language had suffixes -ko and -ka, and a rule that lead to final vowel loss in words of more than two syllables, these suffixes would merge as -k for all polysyllabic roots, but remain distinct for monosyllabic roots. Now, a language could put up with that. But alternatively, particularly if monosyllables are rare, the merged form might spread to monosyllabic roots by analogy. Or, if the distinction encoded is really important, the unmerged form could spread to polysyllables by analogy. The latter becomes more common if there are some really important and common monosyllabic roots. Either form of analogy would lead to outcomes that are not the regular result of soundchange for some words. And different suffixes could theoretically see analogies in different directions. So by analogy in this way you'd end up with different suffixes seeming to 'dictate' different sound changes. But really it's just analogical spread from well-justified exemplars.



*for example, Ringe thinks that the whole of the 4th and 5th conjugation strong verbs in Germanic are in part an anology from the oddity of "to eat" having an initial laryngeal.

[Davush]

In the words of Reverend Lovejoy: short answer yes with an if, long answer no with a but.

No, it's not wise to think of morphemes as dictating their own diachronic rules. However, it's possible to set up regular or irregular sound changes that superficially might look as though that's what's happening.

Thank you! I did suspect morphemes having their own rules wasn't quite right. I've just come across an interesting example, though, where a sound change does appear to be sensitive to morphology: dentalization of coronals in Belfast English seems to not occur with agentive or comparative -er, but does with "regular" -er, e.g. it happens in "dinner" but not in "runner". I'm not sure if this is actually an example of "morpheme dictating sound change" or maybe just "sound change being sensitive to morpheme boundaries"? I suppose it's possible the normal words in -er are more common, so it just hasn't spread to the other forms yet.

Another question would be: is it ever heard of for a language to re-apply an "old" affix after a sound change has stopped operating, even though the original sound-changed form also exists. For example, let's say that vowel syncope applies, so *xani+ta > xanta. After syncope stops applying, a new suffix -mi appears, which prefers to attach to some other suffix. In this case, could "old" -ta be re-suffixed before -mi, yielding *xanitami, even though *xanta is already available? I imagine this would be very unlikely, since the old form is still available, but perhaps plausible if xanta and xanita were competing forms due to -ta being "reapplied".

A further, somewhat unrelated question: are there strong tendencies (or universals?) regarding how vowel reduction/syncope operates (other than it usually targeting unstressed syllables)? For example, I had the following idea:

Unstressed short vowels reduce to /ə/, except in word-initial and final syllables. However, there is a constraint against consecutive /ə/. If this would appear, the first left-most reduce-able vowel is not reduced, but the following is, and this alternates. So, *pásurina > pásurəna, *pásurinama > pásurənama, *pásurinamata > pásurənaməta. (There's probably something to do with stress, secondary stress, morae, etc. here but it the amount of different theories can be overwhelming.)

Thanks!

[Creyeditor]

I have a question about morpho-phonological processes.

Essentially: can morphemes "dictate" certain phonological processes limited to themselves and/or "take precedence" over normal sound changes? This means that they don't operate as "ordinary" across-the-board sound changes.

For example, let's say there's an "ordinary" sound change in which unstressed high vowels are deleted in open syllables: *túkuma > tókma (with lowering due to new closed syllable). This is straightforward enough, but let's also say it fails to apply in some environments caused specifically by morphemes:

The suffix -ta "dictates" that all final vowels of certain word-shapes are removed before it is applied, so *túkuma+ta yields > túkumta, not tókmta~tókmata (which is the expected outcome according to the "ordinary" sound change above). This means -ta has blocked, or occurs before, the previous rule of vowel deletion. In this language, indiscriminate final-vowel deletion by suffixes is not universal: it only applies to -ta, so this also cannot be an "ordinary" sound change.

Also, the suffix -mi behaves differently again: vowel deletion does not occur when -mi is responsible for producing the environment for it: *xánimi > xánimi, NOT ?xánmi with ordinary vowel deletion. On the other hand, *túkuma-mi does yield tókmami, as the environment for deletion is not caused by -mi itself.

This looks like the morphemes -ta and -mi stipulate special rules only for themselves, which "override" the ordinary, purely phonetic-based sound change (high-vowel deletion).

I know it is often said that sound-changes are purely phonetic and are blind to things like this, but I think similar processes where morphemes stipulate their "own" rules do occur?

Any advice appreciated! Thank you!

As for synchronic phonological processes, this is an open question in theoretical phonology. Try googling "morpheme-specific phonology" to get some ideas about it. [Try looking at the data in the purely theoretical papers if they look intimidating.] The two positions are basically: "Yes, there are morpheme-specific phonological processes and they are all over the place" or "No, there are no morpheme-specific phonological processes, but certain morphemes might be underlyingly different [e.g. underspecified for a feature, equipped with a floating tone, etc] which crucially interacts exceptionally with regular phonological processes."

[Davush]

As for synchronic phonological processes, this is an open question in theoretical phonology. Try googling "morpheme-specific phonology" to get some ideas about it. [Try looking at the data in the purely theoretical papers if they look intimidating.] The two positions are basically: "Yes, there are morpheme-specific phonological processes and they are all over the place" or "No, there are no morpheme-specific phonological processes, but certain morphemes might be underlyingly different [e.g. underspecified for a feature, equipped with a floating tone, etc] which crucially interacts exceptionally with regular phonological processes."

Thanks - I think I have been conflating a synchronic analysis with a diachronic one, which has given me a bit of a headache. Am I correct in understanding that (synchronic) morphology–specific phonology would ultimately be the result of "ordinary" diachronic sound change (plus analogy, levelling, etc.) that has basically become opaque, so in the synchronic analysis they do appear for all intents and purposes as "morphology-specific phonological processes"? This is (I think) what Salmoneus mentioned:

This might look as though the suffixes are dictating different sound changes, but in reality everything is the result of perfectly regular, semantics-agnostic sound changes.

Or are there some theories hold that (synchronic) morphology-specific phonological processes cannot be explained by the diachronics?, i.e., "regular" diachronic sound change cannot be used to explain these outcomes?

Your suggestion gave several interesting papers, an example from one (The Morphology-Phonology Connection by Inkelas):

In Mam, some suffixes cause vowel-shortening, but others don't. There's no predictable (synchronic) rule behind this, each suffix has to be learnt individually as retaining or shortening the long vowel:

juus+b'een > jus-been (shortening)
waa+b'an > waa-b'an (no shortening)

However, diachronically, I assume this would be expected to have come about by ordinary "semantics-agnostic" sound change – except by this stage, the sound changes producing this are no longer transparent?

If this is the case, then I think the Belfast dentalization should just be a "coincidence" of its (current) distribution not having effected agentive -er words (yet), i.e., it's a regular sound-change currently ongoing, but for some reasons hasn't reached those words yet, so it looks kind of morphology-specific, but ultimately might not be (or maybe it is...)?

[Salmoneus]

Thank you! I did suspect morphemes having their own rules wasn't quite right. I've just come across an interesting example, though, where a sound change does appear to be sensitive to morphology: dentalization of coronals in Belfast English seems to not occur with agentive or comparative -er, but does with "regular" -er, e.g. it happens in "dinner" but not in "runner". I'm not sure if this is actually an example of "morpheme dictating sound change" or maybe just "sound change being sensitive to morpheme boundaries"? I suppose it's possible the normal words in -er are more common, so it just hasn't spread to the other forms yet.

I don't know about this, but I suspect that what's happening there is that the sound change conceptually does (or did) occur throughout the lexicon, but that it's subsequently analogised away by analogical levelling of the root wherever it occurs. That is: if you have a recognisable root like "run", there's a strong tendency to want to have that root be as similar as possible in all the inflected forms it appears in, and also to some extent across derived forms. A sort of symmetry exists between different forms built on the same root, which is how people recognise that it's the same root in each case: we understand a word like "runner" because it looks immediately like "run" + "er". If a sound change applies to "runner" but not to "run", that breaks that symmetry. There's a tendency to want to restore that symmetry by either 'undoing' the change, or spreading it to all forms of the root.

[Of course, this analogical tendency is very frequently overridden by sufficiently powerful or complicated sound changes, which is why many languages have highly irregular paradigms and opaque derivatives. But sometimes analogy wins out, particularly when we're only talking about relatively few 'iregular' forms being regularised. This is why part of why most languages do NOT gradually develop indecipherably irregular and opaque paradigms! (the other reason is that if parts of the paradigm get too weird they can be dropped completely and replaced with other ways of saying the same thing).]

[and of course when I say X happened and then subsequently Y happened, that's a simplification - that's the "logical order" of events, and may well be identifiably the actual order, but in some cases Y (analogical levelling) could actually be occuring simultaneously with X (regular soundchange). Languages after all arent' really simple entities, but broad tendencies at any one time...]

Oh, and if this weren't English, an alternative solution would simply be to say that the sound change occured before the agentive suffix -er developed. Obviously we know that wasn't actually the case here, though, since the agentive -er is West Germanic if not Proto-Germanic (borrowed from Latin!), and the comparative -er is even older than that.

Another question would be: is it ever heard of for a language to re-apply an "old" affix after a sound change has stopped operating, even though the original sound-changed form also exists. For example, let's say that vowel syncope applies, so *xani+ta > xanta. After syncope stops applying, a new suffix -mi appears, which prefers to attach to some other suffix. In this case, could "old" -ta be re-suffixed before -mi, yielding *xanitami, even though *xanta is already available? I imagine this would be very unlikely, since the old form is still available, but perhaps plausible if xanta and xanita were competing forms due to -ta being "reapplied".

Again, this shouldn't happen in theory - the language can't remember previous sound changes and old affixes. However, there's two ways this can happen in practice:

- there can be analogy from words in which the sound change didn't operate. So, if "xani" appears as a root elsewhere, and if lots of words also show that ta+mi=tami, then yes, xani+ta+mi can =xanitami even if sound changes say that it shouldn't. This also works with suffixes themselves: if *ta usually becomes just *t, it could still be restored as *ta across the lexicon by analogy with a few words where for some reason it didn't become *t.

- sound changes don't actually always occur neatly in order, but can overlap, which can sometimes result in it seeming as though one change occurs before the other in one circumstance, but after it in a different circumstance. This doesn't happen all that often, but often enough that it's definitely A Thing. This can become more of a thing when there's also dialectical differences to be considered, with a change occuring in one dialect faster than in another. A Germanic example: although not everyone agrees, it seems as though North and West Germanic both had shortening of final *e: to *e, and backing of *e: to *a:... except that they do it in a different order. WG backs all *e: to *a:, then shortens it when final to *a; whereas NG shortens final *e: to *e (which then regularly raises to *i), and THEN backs all remaining *e: to *a:. An easy way to explain these suspiciously similar changes is to imagine that both changes were spreading through a dialect continuum in opposite directions...

A further, somewhat unrelated question: are there strong tendencies (or universals?) regarding how vowel reduction/syncope operates (other than it usually targeting unstressed syllables)? For example, I had the following idea:

Unstressed short vowels reduce to /ə/, except in word-initial and final syllables. However, there is a constraint against consecutive /ə/. If this would appear, the first left-most reduce-able vowel is not reduced, but the following is, and this alternates. So, *pásurina > pásurəna, *pásurinama > pásurənama, *pásurinamata > pásurənaməta. (There's probably something to do with stress, secondary stress, morae, etc. here but it the amount of different theories can be overwhelming.)

It's possible to have constraints against consecutive /@/, yes, as a form of vowel dissimilation. However, I wouldn't think it was, as a diachronic sound change, something you'd find that often: big, specific assimilations and dissimilations of unreduced vowels at a distance aren't that common. HOWEVER: this could easily be explained in terms of stress placement rules instead. You could simply say that the first syllable after the primary stress receives secondary stress, as does every second syllable from then on, and that vowels don't reduce when they have secondary stress. [the weird thing, though, is that you'd usually expect the opposite: pas@rin@, etc. Your rule both has secondary stress next to primary AND has a switch between allocation of the first secondary (adjacent) and the remaining ones (alternating). I don't know enough about stress to know how odd it is, but it's odd. I assume there's some way to explain it, though.]

Thanks!
[/quote]

[Creyeditor]

Thanks - I think I have been conflating a synchronic analysis with a diachronic one, which has given me a bit of a headache. Am I correct in understanding that (synchronic) morphology–specific phonology would ultimately be the result of "ordinary" diachronic sound change (plus analogy, levelling, etc.) that has basically become opaque, so in the synchronic analysis they do appear for all intents and purposes as "morphology-specific phonological processes"?

Many theoretical analyses of synchronic processes remain agnostic with regard to diachronic explanation. The line of argumentation is roughly the following. A possible phonological pattern must be learnable. The learner (aka the child) does not have access to diachronic information. Therefore it must be learnable (and describable) without involving diachronic information. [But see Blevin's Evolutionary Phonology for an alternative view. She claims a pattern is possible if diachronically derivable.] Of course some synchronic analyses based on ordered phonological rules basically recapitulate diachronic sound changes, so there's that.

It is definitely possible for regular diachronic sound changes to yield a synchronically opaque pattern that appears morpheme-specific.

And Belfast dentalization seems to be a favourite of some UK phonologists. I think I recall an analysis where the rule of dentalization was ordered before er-affixation.

[Salmoneus]

And Belfast dentalization seems to be a favourite of some UK phonologists. I think I recall an analysis where the rule of dentalization was ordered before er-affixation.

A good example of why theoretical linguistics is so useless for studying languages, since "the rule of dentalization is ordered before er-affixation" is just a different way to say "dentalization occurs before Vr sequences except before -er occuring as an affix". It sounds like we're saying something more meaningful, but it's literally just rephrasing the question and pretending it's an answer. It's playing with words.

One reason it sounds like an answer is that it sounds like a real-world claim: that the process took place literally "before" affixation in the order of linguistic developments. But of course we know that in reality it did NOT take place before affixation. Comparative -er affixation took place (albeit with a less specific meaning) in pre-Proto-Indo-European, and there's no reason to think that dentalization is some pan-European change that has coincidentally survived only in southwestern dialects of Scots (and English dialects affected by them).

So the "ordered before" explanation doesn't just fail to answer the question, it fails to answer the question in a way that requires us to use "before" and "order" in novel ways that cannot be cashed out in real terms, and that require us to postulate an entire additional metaphysical dimension of space-time. And that then require us to postulate an entire universe of superfluous metaphysical entities, these "rules", that can be distributed relative to one another throughout that metaphysical spacetime dimension.

By contrast, the rival explanation - that speakers have analogised the form of the root from tokens in which the triggering context of dentalisation was not present - has several advantages that the 'linguistic' answer does not offer:
- it explains what actually happened, answering the question in a meaningful way
- it's falsifiable (it is only compelling if root-analogisation is a an actual, observable process, and only if tokens do exist to provide the basis for the analogy*)
- it gives us useful information about linguistic processes in general from which we can extrapolate hypotheses about other languages
- it remains ontologically responsible, respecting the principle of parsimony
- it's compatible with the physicalist models of existence that underpin other sciences

On the other hand, though, rephrasing questions and answers and then obfuscating them with gratuitous neologisms and random symbols has provided a lot of people with jobs, without the need for them to actually study languages or anything tedious like that, so I guess it's swings and roundabouts.


[sorry, not getting at you personally, I know you're just giving the party line, but it's just that that is such a perfect example of sophistry that it's hard not to pick up on it...]


*admittedly it's not AS falsifiable as an answer like "because it occured literally before affixation", but unfortunately that one's been falsified, so...