Are all morphemes meaningful?

Question

According to the notes I kept during a lecture on Morphology, morphemes are meaningful themselves and they can also differentiate meaning. Are all morphemes considered to be meaningful? For example {-ize} is a morpheme and it can differentiate the meaning of the morpheme {author}. (author and authorize are two meaningful distinct words). But is the morpheme {-ize} meaningful?

Answer 1

No, there are a small class of morphemes called interfixes which are needed for phonological reasons, but are not considered to carry any semantic content. One example is the i in humaniform.

Answer 2

Well it depends on how you define "morpheme". Usually, it's defined as a sign, i.e. a form-meaning correspondence. In this case, the answer is "yes" by definition (and the -i- in humaniform is not a morpheme but sandhi)

Answer 3

Traditionally, a morpheme is defined as the minimal meaningful unit of language. Under this assumption, every morpheme is meaningful by defnition.

However, this is not always that simple.
The definition works well for most of both free and bound morphemes - definitely, free morphemes such as dog, run, red are meaningful, affxies like -ize (verbalisation), -ed (past tense marker), un- (opposite marker) can claimed to have a clearly identifiable meaning and even morphemes that have a status somewhere in between free and bound, such as confixes (like fanat- in fanat-ic or polit- in polit-ics, polit-ical, polit-ician, which never occur freely but can combine to standalone-words together with affixes, which affixes among each other can not) do.

But then there are special cases like the so-called cranberry morphemes. The cran- in cranberry never occurs alone but is not a classical affix either, yet it does create a unique meaning in combination with berry.
Is the morpheme cran- now meaningful itself, when it only ever creates a meaning with one particular morpheme in combination?
I would say yes, because it contributes to the menaning of the newly formed word in a systematic way, but this is not as clear as for the usual free and bound morphemes.

Another problematic kind of morpheme are circumfixes. German computes the past participle form of most verbs with the circumfix ge-STEM-t (e.g. tanzen -> tanz- -> ge-tanz-t).
In combination, ge-X-t certainly is meaningful - but what about the single moprhemes ge- and -t, are they really meaningful individually? Or are they not morphemes at all?
I would bluff it out by saying that only ge- and -t taken together are a morpheme and ge- in isolation simply isn't one, but that is probably not the most elegant way to put it.

The case I find one the most interesting in this respect are the so-called interfixes (not to be confused with infixes) already mentioned in curiousdannii's answer.
An English example would be -o- in words such as speed-o-meter, or the already mentioned -i- in human-i-form. In German (where they are usually called "Fugenmorpheme"), such interfixes frequently occur between two morphemes in compounds and derivates such as Jäger-s-mann (huntsman) or Arbeit-s-platz (workplace) and it is very difficult to tell what it actually is.
It is definitely NOT an inflectional morpheme (like a genitive or plural marker), as is often assumed, because Arbeits- nowhere occurs in the inflectional paradigm of Arbeit.
But it does not seem to have any identifiable meaning either, you could as well leave it away and would not change the meaning of the word (just that it would sound weird because the word is simply in the lexicon like this); it is assumed that it is mostly there for phonotactic reasons (like easier pronounciation or better contrast between similarly sounding phonemes).
Due to the latter, one might even claim that it is not a morpheme at all, but what is it then? Morphological analysis should be able to segmentize a word into all of its immediate constituents, Arbeit and Platz clearly are the smallest meaningful units and it is very unpleasent to have something in between that is merely a phoneme but morphologically not identifiable at all.

So one might want to give up the above definition and go with a more carful, but also less explanatory one like the minimal unit relevant to morphological and syntactic analysis. This does not make any claims about meaningfulness, but then again it is kind of circular, because morphological analysis needs to rely on what a morpheme is (which we now defined as what is treated by morphological analysis), so there is no much gain in choosing such a rather loose definiton.

All in all, I would claim that the vast majority morphemes are meaningful, and not only free morphemes, but also bound ones (like -ize) or rarer cases like confixes or cranberry morphemes, because they do contribute to the meaning of a word in a systematic, i.e. compositional and predictable way.
However, at least for the presumably merely phonologically originated Fugenmorpheme and possibly also some other special cases, it seems like some morphemes indeed do not have any inherent meaning at all, neither in isolation nor in combination with other morphemes.
My answer to the question would thus be no, not all morphemes are meaningful, knowing that this impairs the classical and actually intuitive definition of a morpheme as "the minimal meaningful unit of language" heavily.

Answer 4

According to the traditional definition, morphemes are by definition meaningful.

Edit: Based on the disussion, I had to retract my statement about uncontroversiality. Yet, so far I haven't seen any non-circular definition of morpheme that doesn't relate it to meaning.

As you noted, -ize can be used to differentiate meanings. This makes me think of Saussure, whose ideas have been very influential on modern linguistics. The cornerstone of his structural approach to language is the concept of value. To put it simply, linguistic signs (such as words or morphemes) don't exist in isolation and their values are determined by the oppositions to other signs in the language. From this perspective, if -ize can be used to differentiate meanings, then it's meaningful by definition.

You may want to take a look at the chapters about morphology in Cognitive Grammar by John R. Taylor. The author analyses a few interesting edge cases of morphemes, such as n- in never and neither. He also argues why h- can't be a morpheme, even though there are semantically related words hear and ear.

Answer 5

I think it is possible that every morpheme sometimes has a meaning. In "personalize", the "-ize" means "cause/make". But I can't find a meaning for "-ize" in your example "authorize", though etymologically it could have once had a meaning. Yet in both these examples, "-ize" is a morpheme, since removing it gives a word, and it has a secondary stress.

"Of" sometimes has the meaning of possession, but in the nominalization of a transitive verb, it is appended just to satisfy the grammatical requirement that a noun may not take a direct object. "The eating of horse liver is not allowed."

"A" expresses indefiniteness, except before a predicate noun, it seems to be there just to help locate the noun phrase. "She is a mailman."