Relations and Proportions in the Structure of English Blend Words
Abstract : Blending as a word-formational process has attracted a great deal of interest from linguists, and several attempts have been made to explain the structure of blend words of various languages. These attempts have mainly concerned the words’ semantic factors and the stress patterns of the source words and the final blend. One particular aspect, dealing with the proportions taken from the source words and retained in the blend, suggests a hypothesis according to which there would be a natural tendency to minimize the loss of information by favoring the originally shorter source word in the final blend. Applying the theory in a small-scale study to a pool of 102 English blend words shows that the majority of examined blends conforms with the idea, and even in those cases which do not, a large proportion from the shorter source form is retained in the blend.
1. Introduction
The usual opening comments in almost all articles on blending in the last 50 years have emphasized the growing interest to this word-formational process, typically explained by the fact that the use of blending in forming new words seems to be increasing (especially in English). At the same time, scholars are somewhat unanimously in agreement that a majority of blend words as such are characteristically rather short-lived (e.g. Bergström 1906: 42; Cannon 1986: 725, 738; Veisbergs 1997: 42). The more intricate ways of forming blend words, in which two words are combined so as to form a new word, joining together orthographic and phonemic items of both source words (the most often used example in English being the word brunch from breakfast and lunch), have also been a topic of discussion among linguists. The usual comment on blending, however, is that there are no clear-cut rules as to their formation: for example, Bauer (1983: 235) has claimed that “the coiner is apparently free to take as much or as little from either base as is felt to be necessary or desirable”. Ten years earlier, Adams (1973: 151) had likewise commented that “it seems difficult to make any generalizations about the shapes that splinters are likely to assume”. More recently the question of the structure of blend words has been looked into, e.g., by Štekauer (1997), Piñeros (1998; 1999), Gries (2004), and Jin (2005). Kelly (1998) examined issues such as the ordering of the elements in English blend words in relation to the frequency and the number of syllables of the source words. Encouraging “a hypothesis-testing approach to blends” (Kelly 1998: 588), he discovered, for example, that blends show certain perceivable patterns, being, in other words, more predictable as regards their form than had been thought. In this paper I will outline some of the previous work on the structure of blends, and examine a pool of blend words from the perspective of a theory proposed in Kaunisto (2000), based on the relations and proportions of the source words in the final blends.
2. What are blend words?
As a starting point, it is useful to examine some of the views on defining a blend word, as there are, in fact, different types of formations which are regarded as blends. The issue of what is a blend word is important to address as there seems to be a degree of variation among linguists as to which lexical item actually is a blend. In addition, the potentially problematic instances need to be pointed out considering the compilation of a pool of blend words for closer examination in the present study.
It has been established in the earlier literature on blend words that there are different types of blends. Some linguists have argued that in cases where one of the source words is represented in its entirety and a part of another word (or “splinter”, as they are also called) has been added to it, only those cases where there is overlapping should be regarded as blends. For example, in tangemon, the two words share the letter e in the middle. But in some words, such as keytainer (from key and container) or paratroop (parachute + troop), there is no overlap between the final part of the first source word and the initial part of the following splinter, and subsequently are considered, e.g., by Barber (1964: 89) to be compounds of one word plus a clipped item rather than blends. Other linguists, on the other hand, do regard formations of this type as bona-fide blends (e.g. Algeo 1977 and Pharies 1987). This is also the view adopted for the purpose of the analysis in the present paper.
In some languages, further difficulties are caused by those groups of blends which involve a recurring source word, usually the second: for example, in English, the word magazine appears in blends such as fanzine, videozine, and letterzine. As noted by Berman (1961: 280), Cannon (1986: 734) and Štekauer (1997: 29), the popularity of such formations has reached an extent where endings such as -zine, -athon, and -burger could well be considered suffixes in their own right – blending thus being an important source of new suffixes. It would appear reasonable to exclude all but the very first instance of such formations in the analysis. Similarly, inflectional endings in some cases may complicate the issue somewhat: for example, in the English blend skurfing (ski + surfing), the -ing ending could as well be ignored from the analysis, or considered to have been contributed by both source words.
A few additional remarks have to be made on the general structure of some blend words. Although blend words, as Devereux (1984: 210) notes, “are ordinarily formed by taking the first few letters from one word and combining these with the last few letters from the other”, other types of blend words also exist. As observed by Soudek (1978: 465), in some cases the source words can be seen as sharing their initial or final elements, causing some potential difficulties in determining which elements represent the source words. Examples of such blends in English include stagflation (stagnation + inflation) and fantabulous (fantastic + fabulous). In the former example, the source words may be regarded as sharing the items ‑ation, and fa- in the latter. This may be a more attractive description of the structure of the blend words rather than claiming that stagnation in the blend stagflation is represented only and exclusively by the splinter stag-. In fact, there are other examples of discontinuous or non-sequential blends in English, such as chortle (coined by Lewis Carroll from chuckle and snort), although they are a small minority among existing blend words, as noted by Cannon (1987: 154).
As regards the semantic types of blend words, Plag (2003: 122) distinguishes between two groups of words: ones which may be considered to be abbreviated forms of previously existing compound words (e.g. breathalyzer from breath and analyzer, or an earlier compound breath analyzer), and ones which denote concrete or abstract things sharing the semantic characteristics of both source words (e.g. brunch from breakfast and lunch; with no previously existing compound breakfast lunch). In the present study, the two groups are not examined separately, although it is worth considering that the fine structure of the final blends may differ depending on whether there was a previously existing compound of the two source words or not.
3. The proportional representation of source words in blends
When considering the structural characteristics of blends, it is clear that a number of factors have contributed to the ultimate shape and form of the blend word. It is not impossible that different options or alternatives for blend words suggest themselves, or even that different rival forms have existed before a particular form has eventually prevailed in popular use. Spelling variants are also occasionally encountered (e.g. gues(s)timate from guess and estimate). In some instances, there may be clear semantic reasons explaining, for instance, the ordering of the two source words in the final blend: in brunch, the chronological ordering of breakfast and lunch makes it logical that the splinters from the two source words are given in the same order. Some studies have been conducted on other characteristics in blend structure. Kelly (1998) observes that the first source words in blends tend to be higher in frequency and they usually contain fewer syllables than the second source words. He furthermore notes that the boundaries between the two splinters occur primarily at major phonological joints (such as syllable boundaries) of the source words, and that the phonemes at the juncture are often phonologically similar (Kelly 1998: 585-587). Other observations have been made on the general prosodic aspects in blending; e.g., Plag (2003) and Jin (2005) have looked into the syllable structures of the source words and the final target words.
As early as the beginning of the 20th century, Bergström (1906: 46) pointed out in his thesis the possibility of investigating “the quantity of the contribution of each element in each different case”. Unfortunately he did not specify exactly what he had in mind. It is evident that he did not think much of such an approach, saying that he preferred not to consider the structuring elements of blend words “as a relation or proportion almost in a mathematical sense and way” (Bergström 1906: 16). Later on in the thesis he explained that based on his material he did not see enough reason to present any such rules. He did, however, present a table of blend words according to the stress patterns of the source words and the resulting blend word.
The question of mathematical relations and proportions, mentioned by Bergström but not commented on since by linguists, is nevertheless one that seems to deserve further investigation. As background and support for this view we could consider the observations made by, for instance, Pound (1914: 1), who wrote that blend words “retain, for a while at least, the suggestive power of their various elements”. Cannon (1987: 144) has also commented that when forming new words by blending, “the reduction usually requires retention of major parts of the two source words”. Attention could be drawn here to the word “major” and also to the importance of “the suggestive power”. How could we define what exactly constitutes a major part that has to be retained in order to produce new blend words which would be well-formed? How large does a part of one source word have to be so as to still have the suggestive power of the complete original form? And then, could this be an aspect from which the two source words could be compared?
It could be argued that the deletion of any items from the source words presents a certain amount of “danger” or “threat” to the understandability of the final blend word. From this viewpoint, ideal blends then would naturally be ones where the ending of the first source word and the beginning of the second one overlap, resulting in a way in no deletion at all. Examples of this kind of “ideal” blending – which Cannon (1987: 144) calls the “traditional” kind of blending – in English include shamateur, slanguage, netiquette, and palimony. But as observed earlier, this type of blending covers only a fraction of all blend words, as there are also blends where elements of either one or both of the source words has undergone deletion.
In Kaunisto (2000), a theory was proposed concerning the aspect of preservation of the elements of the source words, attempting to at least partly explain where and how deletion of elements takes place. In blending two words together, one might assume that there would be a natural tendency to preserve as much from the shorter source word as possible, and thus to minimise the loss of information on the source form that would be under a greater “threat”. It could be proposed that as a result of such a tendency, the relation between the part of the shorter form represented in the final word and the entire source word is greater than (or equals) that between the part of the longer word in the final word and the entire longer source word. In other words, a greater percentage of the originally shorter source word is retained in the blend. To illustrate the idea, we could consider the English blend word brunch, from the source words breakfast and lunch. The longer word, breakfast, is represented in the final word, brunch, by only two letters (br), while the originally longer word, lunch, is represented in the final word by four letters (unch), and the relation 2:9 is less than 4:5, so here the theory would seem to work.
One question arising from this theorizing, of course, is what is exactly meant by length here: orthographical or phonological length? Although it could be argued that letters themselves do not have a significant weight in word-formation in general, the argument could be at least tested with the numbers of orthographical units. One could also postulate whether blend words often originate in written rather than spoken language – a matter which itself deserves further examination.
If we consider then that two words, X and Y, are blended to form a third word, Z, and that X is represented in Z by A (a part of X, that is) and Y is represented in Z by B, the hypothesis can be presented in the following simple axiom:
if x > y, then a:x ≤ b:y
where x = the number of letters/phonemes in X
y = the number of letters/phonemes in Y
a = the number of letters/phonemes in A
b = the number of letters/phonemes in B
Thus for the word brunch we could present the following:
X = breakfast, Y = lunch;
A = br, B = unch;
x (breakfast) = 9, y (lunch) = 5,
a = 2, b = 4;
in which case the final blend does indeed conform with the suggested axiom, as a:x = 0.22, which is smaller than b:y = 0.8.
For the word tangemon, where X is tangerine and Y is lemon, the corresponding breakdown would be x = 9, y = 5, a = 5, b = 4; and again the axiom holds, as a:x = 0.56, which is smaller than b:y = 0.8. A notable thing here is the fact that the e in the middle of the blend is counted twice, due to the overlap of the source words in that element. Cannon (1987: 144) writes that there are blends “where both source-words share one or more letters/sounds that often prevent us from determining which word has provided the shared element”. As was observed earlier in connection with the word stagflation, there seem to be no reasons why such letters or sounds could not be regarded to have been provided jointly by both source words. Instead, as observed by Kaunisto (2000), this is very probably one significant aspect why the formation of such blend words was attractive in the first place, therefore including the shared elements in the calculations of both source words would appear to be perfectly justifiable. Following this line of analysis, in the blend word fantabulous (from fantastic and fabulous), the second – and originally shorter – source word, fabulous, is represented in the final blend in its entirety, albeit in a discontinuous fashion.
4. Examining of a pool of English blend words
To further investigate the strength of the axiom on English blends, a small-scale study was conducted on a pool of blend words compiled from a variety of sources. When compiling the blends for closer analysis, it was considered best to focus on items which could be regarded as having shown a reasonable degree of viability in the lexicon. Although there are numerous instances of blend words that could be found altogether, many of them may not have gained much currency in general use, and while there could be several reasons for the obsolescence of a blend, one reason could be the formal or structural deficiency of the word itself.
The blend words were compiled from a number of linguistic articles on blend words (Algeo 1977, Cannon 1987, Štekauer 1997, and Kelly 1998), and the CD-ROM editions of The Oxford English Dictionary (1992, 2nd edition), Collins Electronic English Dictionary and Thesaurus (1995, 3rd edition), The Random House Compact Unabridged Dictionary (1996, 2nd edition), and The Merriam-Webster’s Collegiate Dictionary and Thesaurus (1999, 10th edition). Occurrences of 20th-century blend words from these sources were then checked in the 100-million-word British National Corpus as well as the Internet (using the Google search engine), and any words not found in these sources (together with ones of which only metalinguistic instances were found) were excluded from further analysis. This resulted in a pool of 102 blend words (listed fully in the Appendix in groups according to their relation to the examined axiom). Although one could argue that the selection of the words as described here provides a markedly conservative selection of blend words or a set of words which is does not reflect the most recently coined items, it is worth reiterating that the intention was to find examples of blends that have in some discernible fashion stood the test of time. As a point of comparison, Cannon (1986) and Kelly (1998) examined collections of 132 and 320 blend words, respectively, but it is worth noting that their studies also included cases which may have been rather ephemeral.
Considering the proportions in which the source words were represented in the resultant 102 blend words examined, it can be observed that in the majority of the cases the axiom tested holds; in other words, one can find higher proportions of the originally shorter words retained in the final blends. The results representing different groups of blend words according to how the words relate to the proposed axiom can be summarized as follows:
Table 1. The results of the examined blend words
|
relation of the blends with the proposed axiom |
# of words |
examples |
|
conforming with the axiom |
71 |
fantabulous (fantastic + fabulous) gasohol (gasoline + alcohol) smaze (smoke + haze) stagflation (stagnation + inflation) |
|
blends with source words of equal length (the axiom has no prediction on them) |
13 |
citrange (citrus + orange) meld (melt + weld) motel (motor + hotel) Spanglish (Spanish + English) |
|
not conforming with the axiom |
18 |
infotainment (information + entertainment) lemandarin (lemon + mandarin) tangelo (tangerine + pomelo) zebrule (zebra + mule) |
|
total |
102 (100.0%) |
|
As can be seen in Table 1, there are some blend words where the source words are of equal length (e.g. meld from melt and weld), and the axiom therefore makes no prediction as to the proportional representation of the source words in the final blend. In a good majority of the instances, however, the source words were of different length, and of these 89 cases, in as many as 71 blends the proportions of the source words in the blends were in agreement with the proposed axiom. Of all the blends examined, 15 words were so-called “ideal” blends where both source words were preserved in their entirety with the two source words overlapping at the boundary (e.g. shamateur from sham and amateur, and palimony from pal and alimony). As many as 14 of these blends are also included in the words conforming with the axiom, as the source words were of different length (one instance, glas(s)teel, had equally long source words).
As in only 18 instances of the words in the entire pool of 102 blend words the proportional representation did not agree with the axiom, it can be said that the tendency to preserve more elements from the originally shorter forms appears to be rather common. This is further supported by the observation that even in the blends which did not conform with the axiom (i.e., the proportional representations of the longer forms were greater than those of the shorter ones), the shorter forms were represented in the final blends on average by 50.0% of the lengths of the original forms. In other words, half the length of the shorter forms on average was retained even if the longer source word was “proportionally dominant” in the final blend.
It is worth noting that the results on the words examined would not look drastically different if the analysis had focused on the numbers of phonemes instead of letters. The assessment of whether the blend conformed with the axiom or not would have changed in only a few instances. Some of these examples also illustrate the deliberately playful character of the words themselves: from a phonemic point of view, in blaxploitation (black(s) + exploitation) both forms are preserved in their entirety, although a strict orthographical analysis would indicate that the shorter form is represented by fewer elements in the blend (cf. Cannon 1986: 747).
Some additional comments can be made as regards the structures of the blend words examined. There did not seem to be any striking difference as to whether it was the first or second source word that had greater proportional representation in the final blend: in 33 instances, the first source word had greater representation, whereas in 50 cases it was the second source word which had a proportionally greater number of letters retained in the final blend. Interestingly enough, this also corresponds with the observation made by Pharies (1987) on Spanish blend words. Furthermore, among the English blends examined, in 17 blends the first source word was retained in its entirety and followed by a splinter from the second word, while 10 blends were comprised of a splinter from the first word and the whole of the second source word.
5. Conclusion
Based on the examination of the pool of blend words in the present study, it appears that the tendency suggested earlier for English blends by Kaunisto (2000) might be one possible factor explaining why some blend words have been formed as they have. The extent to which the blend words seem to conform with the idea seems worth noting, and adds to our knowledge on their formation. One can share the notion expressed, for example, by Kelly (1998: 588) of the promising future prospects in the study of blends, for the observations made here also have other possible implications and possibilities to expand the investigation. For example, as regards the question of the well-formedness of blend words, the conformity with the proposed theory might in future be looked into from the viewpoint of the words’ lifespan, i.e. it could be argued that the structural unattractiveness of blend words might be one reason why they have become short-lived, a characteristic which is rather common among blends. In a similar vein, it might be interesting to examine the structural aspects of blend words in different languages in a contrastive or comparative fashion: for example, a brief look at the Spanish blend words analysed by Pharies (1987) and Piñeros (1999) would appear to show a slightly stronger tendency for the Spanish words to preserve a greater proportion of the shorter source word, with almost 80 per cent of the words agreeing with the proposed axiom.
However, it must be noted that the theory on the proportional representation of the source words in the final blends should be regarded as only one of the factors that would affect their formation, and not as a conclusive one that would override others (including syllabic structure, stress patterns, and semantic factors). Support for this view can be seen when considering those instances of blend words where the source words are of equal length, for if the tendency suggested was strong enough to rule out other possibilities of forming the blend, the splinters from the source words should always be of equal length as well – and this was by no means the case. Considering the theory from conceptual viewpoint, one can also question why the interest as regards the understandability and interpretability of the blend word should be focused solely on the preservation of the originally shorter source word. For example, how is the notion of interpretability eventually served in the case of brunch when we consider the fact that the longer source word, breakfast, is represented in the final blend by only two out of nine letters (or two out of eight phonemes)? Nevertheless, it can be concluded that based on the examination of relations and proportions of blend words, the issue clearly deserves attention as one of the potentially contributing factors in the formation of the words in question.
References
Adams, V. 1973. An Introduction to Modern English Word-Formation. London: Longman.
Algeo, J. 1977. “Blends, a structural and systemic view.” American Speech 52, 47-64.
Barber, C. 1964. Linguistic Change in Present-Day English. Edinburgh: Oliver & Boyd.
Bauer, L. 1983. English Word-Formation. Cambridge: Cambridge University Press.
Bergström, G. A. 1906.On Blendings of Synonymous or Cognate Expressions in English. Lund: Håkan Ohlsson.
Berman, J. M. 1961. “Contribution on Blending.” Zeitschrift für Anglistik und Amerikanistik 3, 278-281.
Cannon, G. 1986. “Blends in English word formation.” Linguistics 24, 725-753.
Cannon, G. 1987. Historical Change and English Word-Formation. New York: Peter Lang.
Devereux, R. 1984. “Shortenings, blends and acronyms.” Word Ways 17, 210-215.
Gries, S. T. 2004. “Shouldn’t it be Breakfunch? A Quantitative Analysis of Blend Structure in English.” Linguistics 42-3, 639-667.
Jin, M-J. 2005. “English Blends: A Descriptive Study of Their Distributional Patterns and Prosodic Features.” The Modern Linguistic Society of Korea, 20-3, 195-231.
Kaunisto, M. 2000. “Relations and proportions in English blend words.” Paper presented at QUALICO 2000, Prague, August 26, 2000.
Kelly, M. 1998. “To ‘brunch’ or to ‘brench’: some aspects of blend structure.” Linguistics 36, 579-590.
Pharies, D. 1987. “Blending in Spanish Word-Formation.” Romanistische Jahrbuch 38, 271-289.
Piñeros, C.-E. 1998. Prosodic Morphology in Spanish: constraint interaction in word-formation. Ph.D. dissertation, Ohio State University, Columbus. ROA-272-0798, Rutgers Optimality Archive, http://ruccs.rutgers.edu/roa.html.
Piñeros, C.-E. 1999. Word-blending as a case of non-concatenative morphology in Spanish. ROA-343-0999, Rutgers Optimality Archive, http://ruccs.rutgers.edu/roa.html.
Plag, I. 2003. Word-Formation in English. Cambridge: Cambridge University Press.
Pound, L. 1914. BLENDS. Their Relation to English Word Formation. Heidelberg: Carl Winter.
Soudek, L. I. 1978. “The Relation of Blending to English Word-Formation: Theory, Structure, and Typological Attempts.” In Proceedings of the Twelfth International Congress of Linguists, Vienna, August 28-September 2, 1977, Wolfgang Dressler (ed.), 462-466. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft.
Štekauer, P. 1997. “On Some Issues of Blending in English Word-Formation.” Linguistica Pragensia 1, 26-35.
Veisbergs, A. 1997. English and Latvian Word-Formation: Contrastive Analysis. Riga: University of Latvia.
Appendix: The list of blend words examined.
A. Blend words which conform with the axiom:
1. advertique (advertisement + antique)
2. advertorial (advertisement + editorial)
3. airmada (air + Armada)
4. alphametic (alphabet + arithmetic)
5. Amerindian (American + Indian)
6. autopia (auto + utopia)
7. bagonize (bag + agonize)
8. ballute (balloon + parachute)
9. beautility (beauty + utility)
10. beefalo (beef + buffalo)
11. biathlete (biathlon + athlete)
12. bimboy (bimbo + boy)
13. boatel (boat + hotel)
14. booboisie (boob + bourgeoisie)
15. breathalyzer (breath + analyzer)
16. californicate (California + fornicate)
17. caplet (capsule + tablet)
18. cassingle (cassette + single)
19. cinnamint (cinnamon + mint)
20. compander (compressor + expander)
21. cosmeceutical (cosmetic(s) + pharmaceutical)
22. dancercise (dance + exercise)
23. decathlete (decathlon + athlete)
24. diesohol (diesel + alcohol)
25. dramedy (drama + comedy)
26. emoticon (emotion + icon)
27. editated (edit(ed) + annotated)
28. fandemonium (fan + pandemonium)
29. fantabulous (fantastic + fabulous)
30. floatel (float + hotel)
31. frauditor (fraud + auditor)
32. gasohol (gas + alcohol)
33. gaydar (gay + radar)
34. glas(s)phalt (glass + asphalt)
35. glitterati (glitter + literati)
36. gues(s)timate (guess + estimate)
37. happenstantial (happen + circumstantial)
38. heliport (helicopter + airport)
39. icecapade (ice + escapade)
40. irregardless (irrespective + regardless)
41. Japlish (Japanese + English)
42. medicare (medical + care)
43. melatonin (melanocyte + serotonin)
44. mockumentary (mock + documentary)
45. musicassette (music + cassette)
46. orature (oral + literature)
47. palimony (pal + alimony)
48. parafoil (parachute + airfoil)
49. pornotopia (pornography + utopia)
50. privilegentsia (privilege + intelligentsia)
51. psytocracy (psychological + autocracy)
52. pulsar (pulsating + star)
53. scanties (scant + panties)
54. scrapnel (scrap + shrapnel)
55. sellebration (sell + celebration)
56. sexploitation (sex + exploitation)
57. shamateur (sham + amateur)
58. shoat (sheep + goat)
59. slanguage (slang + language)
60. slimnastics (slim + gymnastics)
61. slumpflation (slump + inflation)
62. smaze (smoke + haze)
63. smog (smoke + fog)
64. spork (spoon + fork)
65. stagflation (stagnation + inflation)
66. swel(l)egant (swell + elegant)
67. tangemon (tangerine + lemon)
68. telecast (television + broadcast)
69. transceiver (transmitter + receiver)
70. wavicle (wave + particle)
71. wintertainment (winter + entertainment)
B. Blends where the two source words are of equal length (the axiom/hypothesis does not comment on such instances):
1. celebutante (celebrity + debutante)
2. citrange (citrus + orange)
3. ecdysone (ecdysis + hormone)
4. glas(s)teel (glass + steel)
5. meld (melt + weld)
6. motel (motor + hotel)
7. organule (organism + molecule)
8. pomato (potato + tomato)
9. positron (positive + electron)
10. rurban (rural + urban)
11. Spanglish (Spanish + English)
12. squadrilla (squadron + guerilla)
13. zorse (zebra + horse)
C. Blends which do not conform with the axiom:
1. blaxploitation (black + exploitation)
2. burbulence (burp + turbulence)
3. cafetorium (cafeteria + auditorium)
4. docutainment (documentary + entertainment)
5. etorphine (ether + morphine)
6. floptical (floppy + optical
7. geep (goat + sheep)
8. infotainment (information + entertainment)
9. Klavern (Klan + cavern)
10. lemandarin (lemon + mandarin)
11. skorts (skirt + shorts)
12. skurfing (ski(ing) + surfing)
13. snurfing (snow + surfing)
14. tangelo (tangerine + pomelo)
15. traxcavator (tractor + excavator)
16. vibronic (vibration + electronic)
17. zebrule (zebra + mule)
18. zonkey (zebra + donkey)
Kieliskooppi
11/2013