[back to list of essays]
1. Matthews prefers the term "word-formation" instead of "derivation," because within a lexeme it is not necessarily obvious that one word is derived from another, and because there are also inflectional derivations. First, there are pairs of words such as Italian ZIO 'uncle' and ZIA 'aunt' and CUGINO 'male cousin' and CUGINA 'female cousin'. We cannot tell whether ZIO is derived from ZIA or ZIA from ZIO because the words are equally complex, as with CUGINO and CUGINA. Second, words can be derived within a lexeme. Generations is the plural derived from generation, just as generation (N) is derived from generate (V). To avoid ambiguity, "word-formation" is preferable to "derivation."
2. In English, a word-form such as attentiveness can be derived from another word, attentive, and from attentiveness, inattentiveness can be derived. In both of these examples, affixes are added to an existing word-form to derive another word-form. By the same token, we could say that all of the word-forms are derived from lexemes, and all of the lexemes are derived from other lexemes. For example, ATTENTIVENESS is derived from the lexeme ATTENTIVE, and so on. The process of word formation, then, can be said to involve the forming of 'words' from either word forms or lexemes. In Latin, noun, verb, and adjective forms are said to be paradigmatic, or sets of inflected word forms belonging to the same lexeme. Since Latin does have uninflected word-forms, it is necessary to say that lexemes (abstract lexical units) are derived from other lexemes. To make this more clear; as mentioned above, English produces different word-forms by inflecting existing word-forms. Latin cannot do this. For example, ve:rum 'truth' is a noun, while ve:ra 'true' is an adjective. It cannot be said that ve:ra is derived from ve:rum, because ve:rum is an inflected form belonging to a different paradigm, and therefore cannot derive a word-form in another paradigm. In Latin, it seems unlikely that word-forms in separate paradigms would be arbitrarily derived from one another so as to get ve:ra from ve:rum, or from any other word-form (such as luceo 'shine') to another (such as lucidus 'shining'). For this reason, we say that Latin derivation involves lexemes, rather than word forms.
3. "-trepid-" can be seen as the ROOT of the words "intrepid" and "trepidation". This 'root' is the smallest form that a word can be practically broken down into. similarly, "jubil-" and "butch-" can be seen as roots. Note that in English, what is a 'root' and what is a larger 'stem' (an element that forms the basis for another derived or compounded, etc. form , and is a lexeme on its own) is often the same. Take, for example, the word "joy", which is simultaneouely root and stem. In the case of "butch" (according to OED), the situation is similar. However, "trepid" and "jubil" are not independent words of their own; they might contribute form and meaning to another word when they are a part of that larger whole, but cannot exist separately. Diachronically the situation may change through the process of "backformation", in which the 'root' is reanalyzed as a 'stem' in terms of analogy with other words using the same affix of the previously existing larger word. The classic English example is "edit", from "editor". The larger word "editor" was reanalyzed as having a stem "edit" and an affix "-or" (phonetically identically to the productive affix "-er"). This stem was then BACKFORMED from the original single lexeme, single stem, single root "editor".
4. The meaning of devived words is in part synthetic or nocompositional in that the meaning is not predictible from the sum of their parts. The meaning of derived words is analytic or compositional when that meaning is predictible based on the meaning of the composing morphemes. The examples direction, generation, and action are nouns that have been derived from verb forms. The derived noun forms are partly analytic because they have the predictible meaning 'the act of ... '. They are also partly synthetic because each has an unpredictible sense: wrong direction, third generation, court action. The examples unsavory and unholy are partly analytic in that they have the predictible meanings 'not savory' or 'not pleasing to the taste or smell' and 'not holy'. The same examples are partly synthetic in that there is also an unpredictible sense for each such as in an unsavory character and an unholy mess. The examples jailor and prisoner are derived from jail and prison with the analytic sense of 'person related with ...'. But, unlike other examples such as baker and jailor, a prisoner is a person who is jailed, rather than a person who bakes or a person who jails. The exact meaning of the person reference in this example is unpredictible and synthetic.
5. The pairs seem to represent the competition between different formations with varying levels of productivity. Some forms are restricted by their origins, like in-, which being of Latin origin, applies more to Latin stems, while the more recent un- is not so restricted. In terms of semantic separation, un- readily negates a positive aspect, while in- can occur with a wider range of less direct meanings. Non- has a more neutral inflection, as in the difference between unrenewable and nonrenewable, with less of a negative and more of an informative meaning. Dis-, on the other hand, seems to have even more of a negative inflection than un-, as a totally negative reversal is implied. The variations in productivity seem to relate to the degree of freedom allowed to the affix by its meaning and its origin.
6. Competition tends to inhibit productivity. A word will not be created if it is not needed or if the sense it conveys already exists in another word. For example, bishopess is not necessary, because there aren't female bishops. Horse-ess is not created because the word mare exists to specify a female horse. Semantic implications also play a role, as demonstrated by the many prefixes used to form negative adjectives: in-, un-, non-, dis-, and a-. For instance, immeasurable and unmeasurable do not mean the same thing: in- is a Latinate prefix; non- is neutral; a- is used scientifically. Analogy can stimulate productivity of potential lexemes.
7. Looking at English syncronically, we can say that some years ago there was a rule in the language that said roughly "Add -ed to verbs to inflect them for PAST," and DIVE was among the large number of verbs to which this rule applied. Today, for many speakers of English, DIVE is on a special list of exceptions to this rule, and has its own special "rule" that says it should be changed internally to dove in the PAST. Thus these two synchronic snapshots made at different points in time reveal a difference in "formation by rule." What brought about this difference? This would seem to be a diachronic question for which synchronic linguistics has no answer. The usual diachronic answer is one that involves "creation by analogy." Presumably there were sufficient people who started lumping DIVE with verbs like DRIVE [drive:drove=dive: x ] so that a new end-result was achieved. But how/why did they shift to this sort of analogous thinking and away from the "V + -ed" rule approach. Or are they really two different approaches? Perhaps they had been analogizing all along using verbs like thrive and writhe and the vast majority of other verbs [bribe:bribed=dive: x ], etc., until one day they happened to zero in on verbs like drive and strive, to get a new and different result. Did dove come about by rule or by analogy. "The answers are Yes and Yes; the choice implied is unreal," says Matthews in response to similar questions he poses involving derivation; our example with dive involves inflection, where Matthews says "such distinctions seem quite easy."
(3) repay (possible affix: re- (V > V))
Although both "pay" and "repay" are verbs, the expected meaning, should "repay" be a derivation, would be that of 'payment twice' or 'to pay again', or something to that effect. "Repay" is a verb meaning to 'pay back', or 'return in kind'. It may have originally been a derived form (a related word, "requite", with the same meaning, originally came from the affix "re-" and the stem "quite", an obsolete form that used to mean 'to pay'), but if so, the meaning has shifted over the years to the point that it can no longer be said to be a purely derived form, despite its surface structure.
(5) rewrite V [other words on same pattern, incl. semantics?]
(8) justice (possible affix: -ice)
The word "justice" may appear to be a derived form of the word 'just'; and it probably was, in olden days of yore. "Justice" comes from the Latin derived form "justitia" (from the Latin root "justus"), and is the result of further shift through French and English over the intervening years. The regularized spelling which applies to this word and similar ones ("malice", for example) seems to have applied at one time to French loan words that were originally Latin derivations of simpler roots. It is just historical coincidence that "just" (from Latin "justus") was also borrowed, and that the orthography should have resulted in so similar a form in Modern English. There is otherwise no regular application of an affix "-ice" in English.
(10) delight - not affixed
(13) optionality (possible affixes: -al (N > A), -ity (A > N))
The effect of these two affixes on a noun should be to cancel the effects each have on the other, but instead, their use (ie: "nationality") is to create a noun indicating the manifestation, physical or abstract, of a quality that is essential in the root word. The predicted meaning of "optionality" should thus be something along the lines of "an object with a non-compulsory nature". That the word has been borrowed by linguists to refer to something a great deal more specific is indication that the word can no longer be seen as a purely derived form; its meaning is not predictable from its elements.
(15) comfortable A [other words on same pattern, incl. semantics?]
(18) uncouth (possible affix: un- (A > A))
This is a fascinating word, as it is historically a derived form in Old English from the prefix "un" (not) and the root "cuo" (known). There is still an archaic meaning associated with "uncouth" along these lines (referring to something foreign or unfamiliar, ie: 'unknown'), but by and large it is seen today as a regular derivation of the word "couth" ('suave, sophisticated'). "Couth" is, of course, a back-formation from the reanalyzed word "uncouth", which now carries the meaning 'crude, unrefined'. Therefore, although historically the story is somewhat more complex, the existence of a well-used and recognized root, however it came to be, and the regularity of the meaning after derivation, would lead me to call the word "uncouth" a purely derived English form.
(20) disease - not affixed, now one lexeme
[back to list of essays]