DERIVATION WITHOUT AFFIXATION IRREGULAR WORD-FORMATION with the emphasis on SHORTENINGS
The aim of the following essay is to expose the irregular character of derivation without affixation. To this end, the derivational processes of conversion and shortening will be surveyed first and different types of shortenings will be scrutinised as to their heterogeneity; second, the unpredictability of their output will be exposed, and, third, it will be shown that no definite rules can be sustained; fourth, a certain unifying approach to the consideration of irregularities will be outlined.
In its broadest sense, derivation refers to any process which results in the creation of a new word. The diversity of such processes in various language has given rise to a large number of theories which are often not compatible with each other. A common feature of these theories is a differentiation between regular and irregular processes. The most productive regular derivational process is affixation further differentiated into prefixation, suffixation, and infixation. Other regular processes like featural derivation, functional derivation, transposition, or expressive derivation also involve adding an affix to the stem.
On the contrary, irregular derivation describes far less coherent patterns; moreover, their irregularities are due to different factors, e.g. the process involved in conversion does not change the form of the word, whereas various shortenings usually imply contraction. However, a closer look at the latter shows a great diversity of the mechanisms and the material involved.
Controversial issue – there is no unifying theory, different theories and approaches :
2. no general terminology (conversion, zero derivation, functional change, functional transposition, unmarked change of word-class, etc.)
3. no agreement on the status (branch of derivation, a separate type of word-formation parallel to derivation and compounding, or a phenomenon of morpho-syntax and not word-formation at all)
4. consequently, no unifying definition (change of word-class/category without any change of form; derivation by means of a zero-affix; a word-formation process without phonological change, etc.)
Conversion generally occurs in English along the four following lines: N->V, A->V, V->N, A->N. Most productive among them is the N->V type (to bottle, to fool), lesser productive is the V->N type (a look, a guess); the A->V pattern is not productive in the contemporary English (to tidy, to mature) and the A->N pattern is rare and syntactically restricted (the rich – no plural form).
Apart from these major types, various other forms can undergo conversion: adverbs, compounds, acronyms, blends, prepositions, conjunctions, interjections, affixes; examples, to up (the process), the hereafter, they downed tools, a down on me, the down train. Verb-plus-adverb combinations of the following sort are very frequent: a go-between, hang-over, hand-out, frame-up, break-down, clean-over. Apart from these more systematised cases there is a host of hapax legomena depending on the actual situation, the speaker's and listener's mutual knowledge, and uniqueness: Margaret 747'd to London, He enfant terrible'd gracefully.
In order to avoid new coinages with the same roots and already existing meaning, blocking applies certain restrictions on conversion: derived nouns rarely undergo conversion and particularly not to verbs; example, arrival is not further converted into a verb since there already exists the verb arrive from which arrival is derived.
Many treatments of conversion raise diverse problems, the most frequently mentioned are the following three. The first concerns the question of how widespread conversion is, since most discussions focus on its use in English. The second problem is how to decide for a pair of words linked by conversion, which is the basic word and which the derived one. The third issue is the problem of the interpretation of newly coined words, which implies differences in meaning, such as to carpet a room vs. to weed a garden.
Another large branch of derivation without affixation is presented by shortenings which include several processes, their major shared characteristic being the lack of phonetic material in the derived word. Although there is no consistency in terminology and classification (even in English), the following types of shortenings are generally distinguished: abbreviation, clipping, blend, alphabetism, and acronym. The major features that help differentiate these types are the number of words constituting their input and the pronunciation of the output, both of which expose the peculiarities of coinage.
Abbreviation denotes a single word or a phrase reduced to one or a few letters, with the output pronounced alphabetically (as in p. 'penny') or retaining the pronunciation of the input word/phrase (as in Dr., ASAP, or WW2).
Clipping is used to describe a shortening of at least one word one or more syllables of which are deleted, with the pronunciation according to the English pronunciation rules (orthoepic). E.g. fax, lab, or fridge.
Blend is the process of merging parts of two input words into one which is pronounced orphoepically, as in motel, Oxbridge, or slanguage.
Alphabetism and acronym describe a process in which the initial letters of two or more words constituting a phrase combine into a new word pronounced either alphabetically (alphabetism, as BBC, OED, or VAT), or orphoepically (acronym, as UFO, NATO, or LAN).
Predictability of the output and definite rules.
Judging from the above classification, the output appears predictable and it seems instrumental to infer certain rules. However, by taking a closer look, especially at less common examples like 2nd, yuppie, or C of E, and by comparing the above classes, the processes behind these coinages turn out to be less transparent and more bewildering.
Taking clippings into account, there can be distinguished three major patterns: initial clipping which is predominant in English (as in ad<advert<ADvertisment, or vet<VETerinarian, lunch<LUNCHeon, intro<INTROduction, fan<FANatic), middle clipping (fridge<refrigerator, flu<inFLUenza), and final clipping (plane < airPLANE, phone<telePHONE, bus<omniBUS, van<caraVAN). Furthermore, there are combinations of initial and final clipping (bros<BROtherS, proctor<PROCuraTOR, vibes<VIBErationS), or even more complicated combinations with phonetic and orthographic changes (brolly<umBRELLA, bike<BIcyClE, tix<TICetS). Some clippings derived from phrases involve ellipsis of part of the phrase, as in pub derived from PUBlic house where the head is deleted and the modifier is clipped (other examples are Met<METropolitan opera house, fax<FACSimile transmission, or Fed<FEDeral government employee, pop<POPular music, zoo<ZOOlogical garden). The unpredictable character of clippings is also maintained by Bauer (1983).
The predictability is even more problematic if acronyms are taken into consideration. Compared to AIDS (Acquired Immune Deficiency Syndrome) where each word contributes an initial letter to form the word, in laser (Light Amplification by Stimulated Emission of Radiation) it is only content words that are used as input (and functional words are left out), and in radar (RAdio Detection And Ranging) the first input word gives two letters and one letter is taken from the conjunction. The acronym QWERTY has no initial phrase but is constituted from the first letters of the English standard computer keyboard.
The pronunciation of the output shows as little consistency due to which one and the same shortening can be placed in two or three different classes, as for example VAT, LAN, ASAP, or AKA pronounced either alphabetically (alphabetisms), orphoepically (acronyms), or retaining the pronunciation of their input phrases (abbreviations).
Notwithstanding the unpredictability and seemingly sporadic character of shortenings, these words when first coined are often accepted in the language community due to their similarity to already existing words and identification with their input. Most language users seem to know the underlying conventions of shortening and even use them to creatively play with language by coining such shortenings themselves. This implicates certain rules behind those derivational processes. However, a close scrutiny of the issue exposes patterns rather than rules, more or less unconscious processes that dominate at the certain period of time than definite prescriptions to obey.
The overwhelming majority of clippings, in all languages, are from the initial portion of the input. Theoretically one long lexeme may yield two or more different clippings, the initial portion of the word, the final part, and possibly the middle, but this does not really occur. The automobile has become bil in Danish, Norwegian and Swedish and auto in English, German and Spanish; but these are exceptional: in general, international words like kilogram, photograph, and telephone have essentially identical clipped forms in a number of languages. At any rate, the part that remains conforms to language-specific canonical preferences. Shortened forms typically resemble other words of the same language.
In French clippings are mostly of two syllables and the syllables end with vowels. The final vowel is predominantly [o], whether part of the input word or an addition to the truncated material. Examples: hebdo < hebdomadaire “weekly”, h é lico < h é licopt è re “helicopter”, labo < laboratoire “laboratory”. Spanish clippings also consist of open syllables, usually two: foto < fotografia “photograph”, poli < policia “police”, secre < secretaria “secretary”. German clippings may end in a vowel or consonant: Abi < Abitur “examination to qualify for university”, Foto < Fotografie “photograph”, Labor < Laboratorium “laboratory”. The addition of a vowel [i] or [o] is common: Ossie < Ostdeutscher “East German”, Ami < Amerikaner “American”, Normalo < normal “a person normal in some respect”. In English most clipped forms are one syllable in length and end in a consonant: fem < feminine, lab < laboratory, dorm < dormitory, vamp < vampire.
The semantic aspect of the shortening process displays considerable variety as well. In some cases the output is essentially identical with the input in its reference: gym < gymnasium, ref < referee, but differs from the input in the more subtle matter of connotation: clipping has the semantic element of “familiarity”. On the other hand, the output may become totally separated from its input. The separation may be due to the fact that the input is forgotten, as in the case of taxi < taximeter, which referred originally to the device that computes the distance travelled by means of pebbles and hence the fare. Otherwise, the output may come to have a meaning distinct from that of the input, as in the case of cute < acute. In numerous other instances the shortened form occurs in a more restricted range of applications. Vibes are not the vibrations of a machine in operation, but rather musical or mystical vibrations. On the other hand, the shortened form may show an expansion of use or meaning or both. The word psych is used as a verb (often with the particle out), “to deduce the motivations behind some action or result of action”, yet the input must be a noun such as psychology or psychologist. Thus, one single shortened form can stand for a number of input words – several different input words appear as a single output. German Krimi can be Kriminalfilm, Kriminalroman, or Kriminalstück, respectively “detective film”, “detective novel”, or “detective play”. In English it is easy to find homophonous clippings: sub for submarine and also substitute (noun and verb).
Case study: Blending
Blending can be defined as merging parts of words into one new word.
Kubozono (1990) specifies this definition in the following way: blending "involves two input words in a paradigmatic relation, i.e., words that might substitute for one another, as opposed to words that occur side by side". This is the point where blending differs primarily from clipping and clipped compound (compound shortening):
blending: smog < smoke / fog
clipping: exam < examination
clipped compound: breathalyser < breath analyser, sitcom < situation comedy
Blends are subject to diverse constraints, both on the input and the output of the process.
Constraints on the input of blending:
1. both input words belong to the same syntactic category (< paradigmatic relation); according to statistic data, N-N blending is the most productive pattern in English, followed by V-V forms
a. two words belong to the similar, if not identical semantic content
b. the word whose final component constitutes the final component of the blend serves as the head of the blend: motel < MOTor(ist) + hoTEL is a kind of hotel, not a vehicle.
Having sketched the constraints on the input, a number of constraints on the output will be considered now.
Constraints on the output of blending:
1. the most productive formation process underlying blending appears to be the following:
AB / XY -> AY, where either B or X can be null, especially when the input words have some phonemic overlapping, e.g., boatel < BOAT + hoTEL (borrowed into Czech as botel to name the famous resort hotel floating in the Vltava River in the middle of Prague) , sexpert < SEX + exPERT,
2. Phonological constraints seems to be a crucial factor in coining new blends since they have to ensure conformity of new words to the phonological rules of the language.
a. phonotactic constraint - prohibits forms which do not conform to the phonotactic structure of the language, e.g., SMOke + driNK -> [smouk]* (violates restrictions between the peak [ou] and the coda [ŋk])
b. phonemic constraint - prohibits phonemic identity to either of the input words, e.g., BEst + moST -> best *
c. syllable structure constraint – states that switches in the input words take place between the same syllable-constituents (onset-peak-coda), it prohibits switches within a syllable constituent resulting in a split of an onset or of a coda, at the same time it implies that the input words must switch in the same syllable position, i.e. either between the onset and the peak or between the peak and the coda the majority of English blends show the preference for the onset-peak boundary, e.g., SMoke + drINK -> smink, Fist + hOIST -> foist, GLAnce + gliMPSE -> glampse
d. length constraint – ensures correspondence between input and output words with respect to their phonological length the righthand input word and the output blend form consist, predominantly, of the same number of syllables, i.e.:
In AB / XY -> AY, XY and AY are equal in phonological length, e.g., AR-GU-ment + sig-ni-FI-CA-TION -> ar-gu-fi-ca-tion (foolosophy, futilitarian)
This fact can imply that the righthand input word is the more important element phonologically. Moreover, in monosyllabic words the RHR holds because the nucleus (rhyme) of the split syllable comes from the righthand input word, since the switch takes place mostly at onset-peak boundaries; both evidences testify that the head of the blend form is its righthand element.
This case study is based on statistic data and delineates the prevailing lines along which blending functions in contemporary English, but it fails to propose rules since it does not make clear which pattern will be chosen the next time a blend is coined. The investigation made by Kubozono merely describes the state of things but does not impose any rules on this type of derivation. The process of coining blends is rather arbitrary, the choice of patterns is rather random and is restricted semantically, syntactically, and phonologically.
It can be argued that the category of blends is much larger than Kubozono stipulates it: e.g. the blend 'par excellence' coined by Lewis Carroll chortle<chuckle+snort injures the morphological constraint whereas galump<gallop+triumph does not comply with the morphological constraint. Other renowned blends would fall short as well: electrocute<electricity+execute, sexational<sex+sensational. Blends of the recent period (1980's and 90') such as aquarobics, slimnastics, or Reaganomics are also marginalised by the above constraints. However, due to the increasing popularity of blending in public domains (mass media, politics, show business, sports) parts of blends become productive as pseudo-affixes: info- in infoglut, infobahn, infodump; -tainment in transportainment, eatertainment; - gate in Monicagate, Camillagate. One can argue that whereas the latter coinages could be an alloy of blending with affixational derivation, the former cases display blending which major criterion is, different from Kubozono's too differentiating approach, simply the merging of the initial part of one word and the final part of another. A solution to this controversy can be a less discrete categorisation of shortenings and more flexible category boundaries.
Prototype theory for categorisation of shortenings
Prototype theory is often employed for categorisation in semantics. Within this approach, categories exhibit an internal centre-periphery structure and fuzzy boundaries. Prototypes function as the organising entities and as the points of reference with which all members of the category share at least one feature. The extended version of the prototype theory allows any item to become a 'secondary' prototype which associates further members to the category due to shared properties. This development was a necessary adjustment to deal with certain linguistic phenomena, such as different senses, which defy the monocentric structure.
Rúa (2002) applies this approach to the linguistic categories of shortenings. All categories (acronyms, alphabetisms, abbreviations, clippings, and blends) are described in relation to initialisms (i.e. acronyms and alphabetisms). By selecting a number of parameters such as number and type of input entities, pronunciation and orthography of the output, degree of shortening and others, she assembles the shortenings in groups, each group representing a major category (acronyms, alphabetisms, abbreviations, clippings, and blends) according to their varying degrees of representativity, i.e. the items are classified in terms of central, peripheral, and borderline members. The items belonging to the same category are further arranged according to the centrality of their members. In this way it is possible to obtain a feasible explanation for the nature and location of those cases whose irregular features cause problematic classification, e.g. C of E (Church of England). The process of shortening is thus regarded as a unifying device whose application results in the creation of the main categories, namely initialisms, blends, and clippings.
Taking into account the dynamism and multifariousness exhibited by these categories, the prototype approach allows for an accurate explanation of certain phenomena. In the first place, ambivalences (i.e. hybrids combining features of contrasting categories). Secondly, category shifts, for instance, the change from abbreviation to alphabetism (LA). Finally, there are also changes within the same category. This is especially manifested by the category of acronyms where the relocation of the centre has taken place, from capitalised acronyms towards those only in small letters, which means that the word-like forms have been ousting capitalised forms.
The prototype approach can be demonstrated by the evolutionary development of shortenings in five stages which is reflected in the modern use of acronyms. The use of written abbreviations leads to letter-naming alphabetisms, which lead to letter-sounding acronyms. These orphoepic acronyms lead to apposite acronyms, forms which are identical with existing words considered semantically appropriate to the designata. The final stage is the prefabricated acronym, in which the designatum is given a complex name devised for the very purpose of creating an acronym which coincides with another lexeme.
Prototype approach accounts for trends in derivation of shortenings when the prototype cases favour their use as patterns for further creations, giving rise to whole 'families' of items. For example, there exist acronyms modelled on motel: boatel, airtel, aquatel, rotel: 'boat/air/aquatic/rolling + hotel'. There is a set of humorous acronyms based on yuppie, e.g. fuppie ('female yuppie'), huppie ('hippie yuppie'), guppie ('green yuppie'), juppie ('Japanese yuppie'), puppie ('previously yuppie'), etc.
Another increasingly popular pattern prompted by the prototype cases is a prefabricated acronym which coincides in orthographic and/or phonetic form with an existing word. The process is reversed: instead of starting with a phrase to reduce it to initials, a word is first found which provides desirable connotations and then its letters are used as initials to build the phrase. For example, PUSH < 'People United to Save Humanity'.
This tendency reflects wide-spread cases of folk etymology when shortenings (mostly alphabetisms) are given a jocular or ironic interpretation. The British civil service provides good examples for that. The lowest honour for long service with egg-free face is Commander of the Order of the British Empire (CBE). One has to work very hard to receive this honour, and the alphabetism CBE is sarcastically interpreted in the lobbies as “Can't Be Everywhere”. Higher honours are mocked like that: CMG (Companion of the Order of St. Michael and St. George) “Call Me God”, KCMG (Knight Commander of the Order of St. Michael and St. George, implying knighthood and a lifelong peerage) “Kindly Call Me God”, and GCMG (Knight Grand Cross of the Order of St. Michael and St. George) “God Calls Me God”.
Peripheral cases contain combinations of initials, clipped or full constituents, numbers, symbols, apostrophes, hyphens, etc., such as COLT < Carbon Dioxide – CO – Laser Technology, Algol < ALGOrithmic Language, jastop < Jet-Assisted STOP, B'ham < BirmingHAM, C-3PO (no existing input, one of the robots in the film Star Wars). There is a tendency to include numbers in shortenings, which designate syllables with the same or similar sound form, such as B2E < Business to Economy, 4M < ForuM. Such shortenings often appear as rebuses. Other peripheral forms owe their marginal quality to the peculiarities of the input form (QWERTY < a type of keyboard whose name derives from the first six of the upper row of keys), their degree of shortening, different spelling (Fax < FACSimile transmission), degree of fusion of their constituents (phonic integration).
Apart from distinguishing peripheral case, Rúa (2002) also scrutinises hybrids – coinages that combine or merely display features of two or more categories. Such hybrids are rather frequent in English, for example, Mrs < MistResS considered a hybrid between abbreviation and clipping, or MPhil < Master of PHILosophy which Rúa regards a hybrid between abbreviation and blend, but which can be argued to present a hybrid between abbreviation and clipping. The forms aka/AKA and asap/ASAP can be seen as potential hybrids displaying features of acronym, alphabetism, and abbreviation since they have two spellings (capital letters and lower case) and three possible pronunciations (orthoepic, alphabetic, and expanded pronunciation of the initial phrase).
Thus, by avoiding discrete, sharply bounded typology, Rúa succeeds to present the categories of shortenings in their complexity and unpredictability. The polycentric prototype model allows for a coherent and methodical account of their irregularities exposing them not as a negative quality but as a manifestation of linguistic flexibility.