Aims and principles of morphemic and word-formation analysis

1.2 Aims and principles of morphemic and word-formation analysis

A synchronic description of the English vocabulary deals with its present-day system and its patterns of word-formation by comparing words simultaneously existing in it.

If the analysis is limited to stating the number and type of morphemes that make up the word, it is referred to as morphemic. For: instance, the word girlishness may be analysed into three morphemes: the root – girl – and two suffixes – ish and – ness. The morphemic classification of words is as follows: one root morpheme– a root word (girl), one root morpheme plus one or more affixes – a derived word (girlish, girlishness), two or more stems– a compound word (girl-friend), two or more stems and a common affix – a compound derivative (old-maidish). The morphemic analysis establishes only the ultimate constituents that make up the word.

A structural word-formation analysis proceeds further; it studies the structural correlation with other words, the structural patterns or rules on which words are built.

This is done with the help of the principle of oppositions, i.e. by studying the partly similar elements, the differences between which are functionally relevant; in our case this difference is sufficient to create a new word. Girl and girlish are members of a morphemic opposition. They are similar as the root morpheme – girl – is the same. Their distinctive feature is the suffix – ish. Due to this suffix the second member of the opposition is a different word belonging to a different part of speech. This binary opposition comprises two elements.

A correlation is a set of binary oppositions. It is composed of two subsets formed by the first and the second elements of each couple, i.e. opposition. Each element of the first set is coupled with exactly one element of the second set and vice versa. Each second element may be derived from the corresponding first element by a general rule valid for all members of the relation.1

Observing the proportional opposition:

child = woman = monkey = spinster = book

childish womanish monkeyish spinsterish bookish

it is possible to conclude that there is in English a type of derived adjectives consisting of a noun stem and the suffix – ish. Observation also shows that the stems are mostly those of animate nouns, and permits us to define the relationship between the structural pattern of the word and its meaning. Any one word built according to this pattern contains a semantic component common to the whole group, namely: 'typical of, or having the bad qualities of.

In the above example the results of morphemic analysis and the structural word-formation analysis practically coincide. There are other cases, however, where they are of necessity separated. The morphemic analysis is, for instance, insufficient in showing the difference between the structure of inconvenience v and impatience n; it classifies both as derivatives. From the point of view of word-formation pattern, however, they are fundamentally different. It is only the second that is formed by derivation. Compare:

impatience n = patience n = corpulence n

impatient a patient a corpulent a

The correlation that can be established for the verb inconvenience is different, namely:

inconvenience v = pain v = disgust v = anger v = delight v

inconvenience n pain n disgust n anger n delight n

Here nouns denoting some feeling or state are correlated with verbs causing this feeling or state, there being no difference in stems between the members of each separate opposition. Whether different pairs in the correlation are structured similarly or differently is irrelevant. Some of them are simple root-words, others are derivatives; they might be compounds as well. In terms of word-formation we state that the verb inconvenience when compared with the noun inconvenience shows relationships characteristic of the process of conversion. Cf. to position where the suffix – tion does not classify this word as an abstract noun but shows it is derived from one. This approach also affords a possibility to distinguish between compound words formed by composition and those formed by other processes. The words honeymoon n and honeymoon v are both compounds, containing two free stems, yet the first is formed by composition: honey n+moon n=honeymoon n, and the second by conversion: honeymoon n>honeymoon v. The treatment remains synchronic because it is not the origin of the word that is established but its present correlations in the vocabulary and the patterns productive in present-day English.

The analysis into immediate constituents described below permits us to obtain the morphemic structure and provides the basis for further word-formation analysis.

Analysis into immediate constitute

A synchronic morphological analysis is most effectively accomplished by the procedure known as the analysis into immediate constituents1 (IC's). First suggested by L. Bloomfield2 it was later developed by many linguists.3 The main opposition dealt with is the opposition of stem and affix. It is a kind of segmentation revealing not the history of the word but its motivation, i.e. the data the listener has to go by in understanding it. It goes without saying that unmotivated words and words with faded motivation have to be remembered and understood as separate signs, not as combinations of other signs.

The method is based on the fact that a word characterized by morphological divisibility (analysable into morphemes) is involved in certain structural correlations. This means that, as Z. Harris puts it, «the morpheme boundaries in an utterance are determined not on the basis of considerations interior to the utterance, but on the basis of comparison with other utterances. The comparisons are controlled, i.e. we do not merely scan various random utterances but seek utterances which differ from our original one only in stated portions. The final test is in utterances which are only minimally different from ours. «1

A sample analysis which has become almost classical, being repeated many times by many authors, is Bloomfield's analysis of the word ungentlemanly. As the word is convenient we take the same example. Comparing this word with other utterances the listener recognizes the morpheme un-as a negative prefix because he has often come across words built on the pattern un-adjective stem: uncertain, unconscious, uneasy, unfortunate, unmistakable, unnatural. Some of the cases resembled the word even more closely; these were: unearthly, unsightly, untimely, unwomanly and the like. One can also come across the adjective gentlemanly. Thus, at the first cut we obtain the following immediate constituents: un + gentlemanly. If we continue our analysis we see that although gent occurs as a free form in low colloquial usage, no such word as lemanly may be found either as a free or as a bound constituent, so this time we have to separate the final morpheme. We are justified in so doing as there are many adjectives following the pattern noun stem+ly, such as womanly, masterly, scholarly, soldierly with the same semantic relationship of 'having the quality of the person denoted by the stem'; we also have come across the noun gentleman in other utterances, The two first stages of analysis resulted in separating a free and a bound form: 1) un + gentlemanly, 2) gentleman + ly. The third cut has its peculiarities. The division into gent-+-leman is obviously impossible as no such patterns exist in English, so the cut is gentle+man. A similar pattern is observed in nobleman, and so we state adjective stem + – man. Now, the element man may be differently classified as a semi – affixor as a variant of the free form man. The word gentle is open to discussion. It is obviously divisible from the etymological viewpoint: gentle<.0Fr gentil<Lat gentilis permits to discern the root or rather the radical element gens – and the suffix – il. But since we are only concerned with synchronic analysis this division is not relevant.

If, however, we compare the adjective gentle with such adjectives as brittle, fertile, fickle, juvenile, little, noble, subtle and some more containing the suffix – le-ile added to a bound stem, they form a pattern for our case. The bound stem that remains is present in the following group: gentle, gently, gentleness, genteel, gentile, gentry, etc.

One might observe that our procedure of looking for similar utterances has shown that the English vocabulary contains the vulgar word gent that has been mentioned above, meaning 'a person pretending to the status of a gentleman' or simply 'man', but then there is no such structure as noun stem + – le, so the word gent should be interpreted as a homonym of the bound stem in question.

To sum up: as we break the word we obtain at any level only two IC's, one of which is the stem of the given word. All the time the analysis is based on the patterns characteristic of the English vocabulary. As a pattern showing the interdependence of all the constituents segregated at various stages we obtain the following formula:

Un – + {[(gent -+ – le) + – man] + – ly-}

Breaking a word into its immediate constituents we observe in each cut the structural order of the constituents (which may differ from their actual sequence). Furthermore we shall obtain only two constituents at each cut, the ultimate constituents, however, can be arranged according to their sequence in the word: un-un – + gent-+ – le +-man +-ly.

We can repeat the analysis on the word-formation level showing not only the morphemic constituents of the word but also the structural pattern on which it is built, this may be carried out in terms of proportional oppositions. The main requirements are essentially the same: the analysis must reveal patterns observed in other words of the same language, the stems obtained after the affix is taken away should correspond to a separate word, the segregation of the derivational affix is based on proportional oppositions of words having the same affix with the same lexical and lexico-grammatical meaning. Ungentlemanly, then, is opposed not to ungentleman (such a word does not exist), but to gentlemanly, Other pairs similarly connected are correlated with this opposition. Examples are:

ungentlemanly = unfair = unkind = unselfish

gentlemanly fair kind selfish

This correlation reveals the pattern un-+adjective stem.

The word-formation type is defined as derivation. The sense of un-as used in this pattern is either simply 'not', or more commonly 'the reverse of, with the implication of blame or praise.

The next step is similar, only this time it is the suffix that is taken away:

gentlemanly = womanly = scholarly

gentleman woman scholar

The series shows that these adjectives are derived according to the pattern noun stem-Mi/. The common meaning of the numerator term is 'characteristic of (a gentleman, a woman, a scholar).

The analysis into immediate constituents as suggested in American linguistics has been further developed in the above treatment by combining a purely formal procedure with semantic analysis. A semantic check means, for instance, that we can distinguish the type gentlemanly from the type monthly, although both follow the structural pattern noun stem+-ly. The semantic relationship is different, as – ly is qualitative in the first case and frequentative in the second, i.e. monthly means 'occurring every month'.

This point is confirmed by the following correlations: any adjective built on the pattern personal noun stem+-ly is equivalent to 'characteristic of or 'having the quality of the person denoted by the stem'.

gentlemanly→ having the qualities of a gentleman

masterly→ having the qualities of a master

soldierly → having the qualities of a soldier

womanly→ having the qualities of a woman

Monthly does not fit into this series, so we write:

monthly ↔ having the qualities of a month

On the other hand, adjectives of this group, i.e. words built on the pattern stem of a noun denoting a period of time +-ly are all equivalent to the formula 'occurring every period of time denoted by the stem':

monthly→ occurring every month

hourly → occurring every hour

yearly → occurring every year

Gentlemanly does not show this sort of equivalence, the transform is obviously impossible, so we write:

gentlemanly ↔*occurring every gentleman

The above procedure of showing the process of word-formation is an elementary case of the transformational analysis, in which the semantic similarity or difference of words is revealed by the possibility or impossibility of transforming them according to a prescribed model and following certain rules into a different form, called their 'transform. The conditions of equivalence between the original form and the transform are prefixed. In our case the conditions to be fulfilled are the sameness of meaning and of the kernel morpheme. Transformational analysis will be discussed in the chapter on Methods of Linguistic Study. E.O. Nida1 discusses another complicated case: untruly might, it seems, be divided both ways, the IC's being either un-+truly or untrue-^–ly. – Yet observing other utterances we notice that the prefix un~ is but rarely combined with adverb stems and very freely with adjective stems; examples have already been given above. So we are justified in thinking that the IC's are untrue+-ly. Other examples of the same pattern are: uncommonly, unlikely.

There are, of course, cases, especially among borrowed words, that defy analysis altogether; such are, for instance, calendar, nasturtium or chrysanthemum. The analysis of other words may remain open or unresolved. Some linguists, for example, hold the view that words like pocket cannot be subjected to morphological analysis. Their argument is that though we are justified in singling out the element – et, because the correlation may be considered regular (hog: hogget, lock: locket), the meaning of the suffix being in both cases distinctly diminutive, the remaining part pock – cannot be regarded as a stem as it does not occur anywhere else. Others, like Prof. A, I. Smirnitsky, think that the stem is morphologically divisible if at least one of its elements can be shown to belong to a regular correlation.2

Controversial issues of this nature do not invalidate the principle of analysis into immediate constituents. The second point of view seems more convincing. To illustrate it, let us take the word hamlet 'a small village'. No words with this stem occur in present-day English, but it is clearly divisible diachronically, as it is derived from OFr hamelet of Germanic origin, a diminutive of hamel, and a cognate of the English noun home. We must not forget that hundreds of English place names end in – ham, like Shoreham, Wyndham, etc. Nevertheless, making a mixture of historical and structural approach will never do. If we keep to the second, and look for recurring identities according to structural procedures, we shall find the words booklet, cloudlet, flatlet, leaflet, ringlet, tоwnlet etc. In all these – let is a clearly diminutive suffix which does not contradict the meaning of hamlet. Smirnitsky's approach is, therefore, supported by the evidence afforded by the language material.and also permits us to keep within strictly synchronic limits.

Another example of the same nature discussed by a number of authors is the word ceiling. Does it contain one morpheme or two? It may, be argued that ceiling should at present be considered a root word, because the root ceil – is no longer current, and the speaker no longer understands it as a covering or lining of the roof, although the existence of the words covering and lining is, as we have seen, sufficient in itself to consider the word divisible. There are, however, other words in which the same suffix performs a similar function. Thus, in flooring, decking, piping, paving, etc, – ing is equivalent to the semi-affix – work, so that framing is synonymous with frame-work.

There is also one more procedure that helps differentiation and can be offered as a test. This is substitution within similar or identical contexts. This testifies in favour of taking ceiling as consisting of two morphemes, since one may contrast ceiling and flooring. S. Potter quotes the following example: Every apartment is floored with sandal and ceiled with nacre.

This permits us to make one more conclusion, namely, that in lexicological analysis words may be grouped not only according to their root morphemes but according to affixes as well.