A Response to Anonymous’s Collected Graphic Evidence Against the Existence of the Morphome—Knauv Shauling, Ph.D., B.C.E SpecGram Vol CLX, No 1 Contents Linguomogy—Tiberius Bertrom, Ph.D.

More foam Please: Getting morphology from polysemy without it having to have ever been there

Horst Q. Wurmmacher and Optatia Tarnhelm
Institut pro Lankstuudinnet
Vaknek, Republic of Latveria

Aronoff’s (1994) introduction of the term morphome was a marked advance in the field of morphology. In the 60s, generative phonology had managed to purge biuniqueness from polite discussion1 but morphophonology, left free to wander where it willed, then got up to far more than it should. The hoary morpheme had long been cohabiting with bits of semantics; one almost didn’t dare let one’s grammar interface with it, for fear it would suffer a rash of lexicalism, or explode into a frenzy of uncontrolled downward feature percolation. The morpheme was, to put it politely, behaving in a very distributed way.

The morphome, by contrast, was, and is, much more safely constrained; when it mediates anything at all, it is simply morphophonology on the one hand and morphosyntax on the other. It lets you play matchmaker for nice, well-behaved forms like the Latin perfect and future participles, which then can enjoy a perfect future together. And while it is constrained, it is nevertheless sociable; you have doubtless heard of it, after all, which is most likely not true of the morphon even though the latter appeared at least twenty-five years before the former.

Nevertheless, the modern morphologist still faces a problem, which is that at first glance there are not so many morphomes wandering the landscape as one might like. We would like to take this opportunity to argue that they are, in fact, quite numerous; one merely needs to know how to look. There are two key points to keep in mind. First, in reference to morphosyntax, one needs to be able to argue that two (or more) categories manifest via one postulated form, without that form being taken as a sign of either one separatelye.g., as with the aforementioned Latin participle. This is particularly important since one doesn’t want to encourage the young and impressionable to think of any of these as signs in the first place; that path leads to semiotics. Second, the form must clearly involve morphophonemes, not phonemes.

Both of these conditions can be met quite easily, as long as one keeps in mind that linguistic categories are friendly things; there is an inexhaustible supply of them, and they are always on the verge of always having been there waiting for the analyst to discover them by putting them in a model. They are a conceptual equivalent of what, in biological terms, would result from a cross between the schmoo and the tribble.2

We are now ready to find our morphomes.
A.  //   
B. {}
C. //
D. {}

Take any word for which you can easily claim multiple meaningsthis should not limit you overmuch. We shall take foam as an example; it can be a noun or a verb, it can be an edible substance or a form of rigid plastic (which contradicts itself in a way that the edible stuff does not). Nothing is needed but the right attitude toward one’s model, and categories will subsequently pre-exist themselves; let us for purposes of this case posit “lattetude” and “squeevle” as purely grammatical categories.

You, dear reader, may be objecting at this point, saying that these are obviously semantic categories. The senses have different statistical relations to other words, but we distinguish foam1, representing the lattetude, from foam2, representing the squeevle, only by context, and these labels are based on the terms’ meanings. A grammar that tries to take wider context into account wouldn’t be context-free, and thus is not to be mentioned in polite society. But none of this has to be the case! A rather simple set of arguments will resolve the difficulty:

  1. View “XP” as a general term standing for “XPy”, where y can be an arbitrarily large number. Rules making reference to XP are generalizations; [S → NP VP] is, for example, shorthand for millions of rules (but they are constrained; their actual number is irrelevant, of course).3

  2. Posit that the elements we have been viewing as “heads” are, in fact, a different category entirelycall it a “finagle”and each XPy is actually the expansion of an element that determines the y. In “I’d like more foam on my latte,” for example, “more foam” is an expansion of {Ø1}, whereas in “We need more foam around the alembic,” “more foam” is an expansion of {Ø2}. There is ample precedent for this kind of thing, as anyone who applied the DP hypothesis to languages without determiners can tell you.

These two moves will suffice to turn semantics into morphosyntax quite handily. A lattetude element, foam1, happens to function as a finagle of only one kind of head, whereas a squeevle element, foam2, is a finagle of a different one; the difference in “meaning” is actually a difference solely in grammatical category. They are related only in the same way as the future and the perfect in Latinthey are grammatical categories that appear to map onto the same form.4 For the phonological side of the equation, one step is sufficient:

  1. Allow underlying phonemes that are never realized.

Since one cannot, after all, view a base as actually rich if all of its elements have to work, this should be uncontroversial. Given an array of apparently-similar forms, foam1...n, we can decide that these clump into however many arbitrarily distinct forms we want. In this case, we shall put foam1 and foam2 into one set, and any of the forms foam3...n in another. Move (3) allows one to posit that the phonological representation of the set comprising [[foam1, foam2]] can be something like /fo:m&/; the remainder are /fo:m%/. We now have the same kind of mismatch as supports the morphome-based analysis of the Latin participles/fo:m&/ does not directly represent any one morphosyntactic category, but rather a distinctive combination not motivated by any semantic considerations, and it is phonologically distinct from /fo:m%/; it simply isn’t phonetically distinct.5

This simple set of procedures allow the enterprising morphome-hunter to find prey wherever native speakers think there is a difference in meaning. That it does so by eliminating semantics is not a problem, we would argue, but rather a feature. Semanticists are prone to vigorously experimenting on categories and killing them off at an alarming rate; morphologists are content with form, chasing after no incontinent content, and thus create much less strain on our theoretic ecosystem.

1 If you have not been exposed to this concept, we do not wish to burden you with it; suffice it to say that it does not refer to something that allows one to stand out from a crowd in an unusually broad range of nightclubs.

2 Schmoos, being schmoos, will manage to cross with anything if you want them too, so this is not as far-fetched as it sounds. If you object to the analogy on the grounds that schmoos and tribbles are fictitious, we gently suggest you not think too closely about linguistic categories; you’re probably happier just as you are.

3 The right sort of constraints are timeless, fortunately, and therefore not subject to the depredations of psycholinguists quite as much as are rules, the discussion of which tends to lead students toward thinking about processing.

4 You might be objecting at this point, saying that these steps fly directly in opposition to what Chomsky (1970) argued. Perhaps you think that in our efforts to avoid lexicalism, we have waxed overly dogmatic. To that, we will reply that (a) Chomksy himself said that those were just “remarks”it’s in the titleand (b) your arguments are possible, but lead to a conclusion that we find uninteresting, and therefore not part of the theory of grammar.

5 Most elements such as the & and the % (and there are millions of them) are null in all extant human languages, primarily because they all have one or more features that are not realizable using current anatomy; the “You have to have the articulator” constraint, and the “Don’t choke yourself to death” constraint, universally outrank the “Say all the parts” constraint. The #4512 family of elements, of course, is a special case, violating temporal rather than anatomical limits (it involves sounds being produced simultaneously at various points outside the light cone of the speaker) and thus potentially standing as evidence of a universal universal constraint.

A Response to Anonymous’s Collected Graphic Evidence Against the Existence of the MorphomeKnauv Shauling, Ph.D., B.C.E
LinguomogyTiberius Bertrom, Ph.D.
SpecGram Vol CLX, No 1 Contents