[back to list of essays]

Chapter 5: Possible answers to Study Questions

1. Cockroach is opaque in that its meaning is not related to the meanings of its parts because it comes from Spanish 'cacaracha'. Sparrowgrass is a reinterpretation of asparagus, and its meaning is also not related to the meaning of its parts. Blackbird is opaque in that the meaning, which relates to a bird species, is not seen in the parts. Similarly, greyhound refers to a breed of dog. The meanings of these words, though related to the meanings of the parts, extend beyond those meanings and are therefore opaque to a degree. The words nuthatch and winchat refer to types of birds. Their meanings are not even related to the meanings of their parts, therefore they are opaque.

2. Matthews mentions the "semiproductivity" of some compounds. Some forms can be predictable and understandable, but not "natural." These 'semi-productive' compounds are dependent on context, and are not necessarily what we might call 'creators of context'. 'Meat-eater' is a common, and 'context creating' compound, while 'egg-eater' is understandable only by analogy, and in a proper context. As Matthews explains, some words, such as 'cheese-eater' have "acquired clearer status as a unit." This is mysterious; no simple answer can be offered, and in fact, it is a 'fuzzy' status that a word such as 'cheese-eater' has acquired, compared to the fully acceptable status of 'meat-eater', and the unnatural status of 'egg-eater'. While 'egg-eater' (meaning a person who loves to eat many eggs) can be considered analogous to 'meat-eater' (meaning a person who loves to eat much meat), 'lady-killer' and 'nun-killer' have a less transparent semantic relationship. While 'lady-killer' refers to a person who metaphorically 'kills' women (in other words, arouses women), 'nun-killer', it would seem, can only take on a more literal meaning. 'Cheroot-smoker' is easy to analyze, (if fuzzy in its status as a compound) as meaning one who smokes cheroot in the same way that whiskey drinker means one who drinks whiskey. While 'meat-eater', 'lady-killer', and 'whiskey-drinker' are all fully independent word forms, ('context creators'), 'egg-eater', 'nun-killer', and 'cheroot-smoker' are dependent on the existing compounds, proper context, and a little imagination on the part of the speaker or interlocutor.

3. "Any compound may, in principle, be an immediate constituent of a further compound." Give examples. Essentially what this comes down to is viewing a compound as a single unit. Should this be the case, then this single unit may then be treated as a simple word, which may be further compounded, as in "fruit-juice carton" or "football player". Both are compounded in layers; the entire unit is a compound, as is one of its elements ("fruit-juice" in one and "football" in the other). The fact that such compounds exist seems to show that compound formation results in units which may be used in further compounds. (Matthews, p. 85).

4. As pointed out by Matthews, an important argument in favor of the "syntactic view" of compounding is that it "simplifies the design of a grammar." Given a transformational model of syntax, the domain of morphology is essentially subsumed by syntax. In the example given by Matthews, the compound 'bookkeeper' and the construction 'keeps books' are transformationally related in the syntactic component of the grammar. There is no need for a morphological level of analysis to describe the formation of the compound. The "syntactic view" is thus simpler. The argument against this view is that it is not clear exactly what the underlying deep structure will be. The Matthew's examples 'windmill' and 'flourmill' illustrate the point. The relationship seems not to be semantically transparent. A 'flourmill' is a mill that produces flour, but a 'windmill' is not a mill that produces wind. The transformational/syntactic view of 'compounding' is clearly too powerful and in the original sense of the word, not generative. A generative model originally was intended to mean 'specific'. The form of the transformation that can relate both 'windmill' and 'flourmill' is not specific.

5. Distinguish between endocentric and exocentric compounds. Give examples. Endocentric: compound in which the head noun makes sense by itself, book example: blackbird. Exocentric: function of the noun is different, and out of context, the noun usually does not make sense(p. 90). Book example: Whitethroat, bigmouth.

6. The minimalist approach to classifying English compounds only distinguishes between formations when there is a clear structural difference. With this approach, book-keeper (Noun + Agentive), pickpocket (Verb + Noun), and moonshine (Noun + plain Verb) are distinct. Daybreak and spacewalk, book-keeper and office-worker are not distinguished because the basic structure is the same. The maximalist approach makes the same distinctions as the minimalist approach, but also makes a distinction between other types which can be generalized. Here, washout, lie-in, and playback (all Verb + Preposition) are considered distinct.

7. Matthews is arguing that there are some words that fall into the cracks and cannot be classified in either of these ways. For example, Matthews mentions 'bluebottle' (a kind of fly) and 'yellow-belly' as words which neither a minimalist nor a maximalist approach can specify the meanings of despite their classifications. Another example of this type might include 'paperback', where the meaning is unpredictable from the parts. Matthews also argues that there may be indeterminacy's in applying classifications, as in 'sightseeing' or 'sunbathing', where the first elements of the compounds are different grammatical entities (the first is an object, the second and adverbial modifier). Compare these compounds to 'body-surfing' where 'body can be an adverbial ('on' or 'with' the body) while SURF can be an object (where one can possibly say, "I use my body to ride the surf."). In addition to the first two problems mentioned above, Matthews suggests that both approaches are unable to deal with opaque compounds. He offers examples such as 'nuthatch' (mod. + N?) and 'cuptie' (comp. + V?). Another example might include 'snapdragon' where the elements that make up the compound might be definable, but they do not help us understand the compound itself. In the end, it would appear that any attempt to classify compounds of any kind is forcing the issue.

8. Should words like policeman and postman be considered compounds or derived words? Give evidence for and against. Phonetic evidence (the unstressed, reduced vowel in the final syllable) seems to indicate that words like "postman" (ie: "boogeyman", "seaman", "gunman", "salesman", "fireman", etc) have been reanalyzed as a single, compound unit, especially when compared to obviously derived forms that have NOT reduced this final vowel in "man", such as "pizza man". However, the fact that the plural of these forms is irregular and parallel to that of 'man'/'men' (thus: plural "seamen" instead of "seamans") seems evidence that the words may be derived -- evidence that is further supported by the existence of forms like "saleswoman" and "firewoman". (Matthews, p. 94).

9. Bridegroom is problematic because although the word groom is a free morpheme, it probably is a back-formation from bridegroom. Historically bridegroom is derived from OE bryd-guma 'man of the bride', where guma had the meaning 'man'. By the time of Modern English guma had lost its status as an independent morpheme and was reanalyzed as groom. (folk etymology) We can now speak of a bride and groom, but the existence of groom as an independent morpheme is the result the back-formation so the current status of bridegroom remains problematic. Cranberry remains a prototypical example of a problematic compound. In fact, a whole class of exceptional examples is often referred to as cranberry morphemes. cran- like huckle- does not occur as either a free morpheme or in combination with other roots. Cupboard presents a problem because it does not demonstrate the typical first member stress pattern. It is also semantically opaque.

Hiccough in its alternate, and much more common spelling, hiccup, would probably not even be considered as a possible compound. The notion that it might in some way be related to cough, suggests that might be another example of a form derived via folk-etymology. It possible to analyze magpie as two phonological words, but the problem is that the pie in magpie is in no way related to a pie that can be eaten. Magpie is not semantically transparent. There are problems with both the stress pattern and the meaning of pullover. Lord and lady may be historically derived from two member compounds, but to any modern speaker, there is no knowledge of what those compounds could be.

10a. Morphological criteria rely on the ability to distinguish the parts of a phrase/compound with regards to such inflections as pluralization, changing of tense and so on. Semantic criteria set forth conditions regarding the degree of relatedness between the meaning of the parts and the meaning of the whole. The greater the direct and clear relation between the meanings, the less likely they are to form a real compounded lexeme. Phonolgical conditioning moves the stress in a compound away from the head to the first syllable, which is therefore a condition or criteria for qualification as a compound. Finally, syntactic tests consist of trying to separate the parts and uses them in their separate roles. The degree to which a phrase is not a compound is the degree to which this is possible. None of these criteria are either clearly necessary or sufficient, but seem to provide indications of the diachronic shifts over time.

Alternative explanation: In morphological test, "socio-economic" is a compound because its first member is a stem or stem-variant plus suffix which can't form a word on its own. The elements of "heir apparent" are syntactically inseparable,but it's not a compound in morphological test because it is still inflected as separate unit as in "heirs apparent" in the plural.

In semantic test, "mental hospital" and "topless bar" might be compounds because their meanings are predictable. But "topless bar" is not a compound in phonological test because it has its main stress on "bar".

In phonological Test, "mental hospital", "madman", "Snow White", and "Main street" are compounds because they have a stress on the first element. But "socio-economic" and "Seventh Avenue" don't have a primary stress on the first element, though they are also compounds in morphological test because it is inflected as a whole.

In syntactic test, "put out" and "Ausmachen" might not be compound because they can be split, but "ausmachen" is a compound because it is one word and has one accent.

Accordingly, In deciding compound, no single test is decisive.


(1) lady Simple lexeme.

(2) cupboard

(3) heir-apparent The word is only partially predictable from its parts (ie: it indicates somesort of inheritant), the pattern itself seems less than fully productive ("lord-apparent", "teacher-apparent"), the stress seems to fall on the first syllable of the word, and the make-up of the word (N-Adj) does not follow that of English syntax. The weight of evidence seems to indicate that this is a compound, despite the fact that the word does not pluralize as a unit (ie: "heirs-apparent", and not "heir-apparents").

(4) mother-in-law

(5) cockroach Opaque compound; borrowed from spanish-cucaracha (cf. sparrow grass [p. 83]).

(6) winchat Compound.

(7) linch pin

(8) offside The word has a very idiosyncratic meaning, and cannot undergo tmesis without completely destroying its meaning. Furthermore, it seems to inflect as a unit ("offsides"). On the other hand, the pattern is fairly productive ("onside", "upside", "farside", "nearside", etc), stress seems to fall equally on both syllables of the word, and the make-up of the word (P-N) reflects that of English syntax. This is a bit more difficult a case, but due to the fact that the word's meaning is so unpredictable (and certainly not from its parts), I would guess that this is a compound.

(9) downsize (v.)

(10)crybaby Compound; spoiled brat (?).

(11) space walk Compound (N+V-->N); daybreak.

(12) falling out

(13) cranberry This word has been partially reanalyzed in its first syllable (what is a "cran"?), and thus has a very idiosyncratic meaning. It inflects as a unit ("cranberries"), and has stress on the first syllable (with weakening to schwa in the second vowel for many speakers). Despite the fact that the pattern is fairly productive ("blueberry", "raspberry", "blackberry", "strawberry", etc.) and that the syntax (Adj?-N) reflects that of English, this would seem to be a compound.

(14) madman

(15) postman

(16) overbook Compound (P+V-->V); overeat.

(17) ongoing

(18) inbred "Inbred" is a word that has a predictable meaning (to have something "inbred" is to have it "bred in(to)"), and can undergo tmesis without affecting that meaning. It is difficult to say whether or not the word inflects as a whole, as the portion that reflects that inflection is the second of the two ("inbreed", etc). The issue is complicated by the fact that this seems to be a fairly unproductive pattern ("inbred" does not seem to be related to "inbibe", "injest", etc), with a marked internal (P-V) syntax, and the fact that stress falls upon the first syllable of the word. If this is a compound, I would guess that it was formed through a process of analogy with words of a Latin origin.

(19) offload

(20) outdistance

(21) through-put Compound (P+V-->N/V); in put, out put.

(22) accident prone

(23) lackluster Although it has a partially predictable meaning, "lackluster" has an idiomatic use -- further, it is a very unproductive pattern (disregarding examples like "lackadaisical"). Although its elements are recognizable lexemes, the process of tmesis would eliminate any special meaning imparted by use of the word "lackluster". Although it is difficult to say whether or not the word inflects as a unit (I cannot think of an example where it would inflect), and the pattern is similar to that of English syntax (V-Adj), there is definate stress shift to the first syllable (compare "LACKluster" with "lack LUSter"), and I would venture to posit that this is a compound.

(24) plea bargain

(25) failsafe

(26) passer-by Compound (N+P-->N); runner-up, hanger-on.

(27) come to

(28) turn off (v.)The verb "turn off" has both predictable and unpredictable meanings (to shut off an electrical device, to disgust and/or repulse), and seems fairly unproductive, apart from its counterpart "turn on". However, it seems to behave less like a simple unit than a complex structure (ie: "that X turned me off") in both use and inflection, has no notable stress shift to the first syllable, and its ements retain separate meanings even when split apart. Although the noun form of this verb (ie: a "turn-off") is definately a compound, I will not venture an opinion on the compound-ness of the verb.

(29) add up

(30) lay off

Tests for compound status (from Group 2):

a. Do they behave like simple words?

b. Do they have idiosyncratic meanings? Is the meaning ambiguous from the elements? The SEMANTIC test....

c. Is the pattern only semiproductive? The MORPHOLOGICAL test...

d. Does the word inflect as a unit? Pluralization -- third SYNTACTIC test...

e. Does the stress differ from that of a phrase or derived structure? This is PHONOLOGY: If the stress is on the first syllable, it is a true compound. [an English compound noun, that is--BWB]

f. Do the elements retain any meaning? Has the word been reanalyzed into a single lexeme? IF tmesis is possible (i.e., splitting the word into individual units, generally at a SYNTACTIC level in the sentence), does that indicate it is still a compound?

g. Does it follow syntax in make-up? If, for example, the internal SYNTAX is not the same as that of the sentence-level syntax (ie: N followed by Adj) then it may be a compound.

[back to list of essays]