In Section 5.2 we said (or at least strongly implied) that roots are always free and affixes are always bound. The second part is true by definition, but the first part is not. Consider the following words, all of which contain the affix {\/-\u0259bl\/} \u2011able\/\u2011ible<\/em>:<\/p>\n The cases in (1a) all conform to our expectation: if we remove the affix, we are left with free roots \u2014 the verbs \/\u200b\u02c8ka\u028ant\/ count<\/em>, \/\u200bk\u0259n\u02c8v\u025d\u02d0t\/ convert<\/em>, \/\u02c8m\u025b.\u0292\u025a\/ measure<\/em>, \/\u200bp\u025a\u02c8f\u025bkt\/ perfect<\/em>, \/\u200b\u02c8p\u0279\u0251\u02d0.f\u026at\/ profit<\/em> and reverse<\/em>. These can be used, for example, to paraphrase the adjectives:<\/p>\n The cases in (1b) are different: if we remove the affix, we are left with the roots \/\u02c8\u0251\u02d0d\/ aud\u2011,<\/em> \/\u200b\u02c8k\u00e6l.kj\u0259l\/ calcul\u2011<\/em>, \/\u200b\u02c8\u025b.l\u0259d\u0292\/ elig-<\/em>, \/\u200b\u02c8m\u025b.m\u025a\/ memor\u2011,<\/em> \/\u200b\u02c8nu\u02d0.m\u025a\/ numer\u2011<\/em>, \/\u200b\u02c8t\u025b.\u0279] terr\u2011<\/em> and \/\u200b\u02c8t\u0251\u02d0.l\u025a\/ toler\u2011<\/em>. These are not free morphemes \u2014 we cannot use them by themselves:<\/p>\n But although aud\u2011, calcul\u2011<\/em>, memor\u2011, terr\u2011<\/em> and toler\u2011<\/em> are not free morphemes, they are clearly roots. For one thing, they also occur with other affixes: both aud\u2011<\/em> and toler\u2011<\/em> occur, for example, with {\u2011\/\u0259ns\/} \u2011ence\/\u2011ance<\/em> (in audience<\/em> and tolerance<\/em>); both toler\u2011<\/em> and calcul\u2011<\/em> occur with {\u2011\/e\u026at\/} \u2011ate<\/em> (in tolerate<\/em> and calculate<\/em>); calcul\u2011<\/em> occurs with {\u2011\/\u0259s\/} \u2011us<\/em> in calculus<\/em>; memor\u2011<\/em> with {\u2011\/a\u026az\/} \u2011ize<\/em> in memorize<\/em> and with {\u2011\/i\/} –y<\/em> in memory<\/em>; numer\u2011<\/em> occurs with {\u2011\/\u0259l\/} \u2011al<\/em> in numeral<\/em> and {\u2011\/\u0259s\/} \u2011ous<\/em> in numerous<\/em>, and terr\u2011<\/em> with {\u2011\/\u0259\u0279\/} \u2011or<\/em> in terror<\/em> and {\u2011\/\u0259.fa\u026a\/} \u2011ify<\/em> in terrify<\/em>.<\/p>\n In all cases, the root contributes a fairly consistent meaning that we can identify by subtracting the meaning of the suffix from the word: audible<\/em> means \u2018able to be heard\u2019, auditory<\/em> means \u2018by means of hearing\u2019, audience<\/em> means \u2018those who hear something\u2019 (at least that is its original meaning \u2014 it now also refers to a group of people perceiving something via other senses), so aud\u2011<\/em> means something like \u2018hear\u2019. Likewise, calculable<\/em> means \u2018able to be assessed in terms of numbers\u2019, calculate<\/em> means \u2018to assess in terms of numbers\u2019, calculus<\/em> refers to a branch of mathematics, so calcul\u2011<\/em> means something like \u2018count\u2019.<\/p>\n So, they are roots, but they are not free morphemes. In other words: they are bound roots<\/strong>.<\/p>\n There are even cases of bound complex bases<\/strong> \u2014 combinations of a root and an affix that cannot be used without an additional affix. Consider the following words, which, again, all contain the affix {\/-\u0259bl\/} -able\/-ible<\/em>:<\/p>\n At first glance, these words are all derived from adjectives containing the suffix \u2011able<\/em> by prefixing them with the negative preffix {\/\u026an\/\u2011} in\u2011.<\/em> For the cases in (6a), this is indeed the case:\u00a0 inaudible<\/em> is derived from audible<\/em>, intolerable<\/em> is derived from tolerable<\/em>.<\/p>\n For the words in (6b), it is not true. The words introuvable<\/em> \u2018impossible to find\u2019 and intransnatable<\/em> \u2018impossible to swim across\u2019 occur in Present-Day English (even though they are very rare). However, the hypothetical words *trouvable<\/em> and *transnateable<\/em>, from which they seem to be derived, do not occur and have never been documented in the entire history of English. In other words, intransnat\u2013<\/em> and\u00a0introuv<\/em>\u2013 are bound complex bases. The words introuvable<\/em> and intransnatable<\/em> were borrowed into English from French and Latin respectively, but their bases trouvable<\/em> and transnateable<\/em> were not.<\/p>\n A special case of bound roots are unique roots<\/strong>, sometimes referred to as \u201ccranberry morphemes\u201d, based on the first part of the word \/\u02c8kr\u00e6n.be\u0279.i\/ cranberry<\/em> \u2014 the second part is clearly recognizable as the word \/\u02c8b\u025b\u0279.i\/ berry<\/em>, which occurs by itself, but also in compounds like \/\u02c8blu\u02d0\u02ccb\u025b\u0279.i\/ blueberry<\/em>, \/\u02c8str\u0251\u02d0\u02ccb\u025b\u0279.i\/ strawberry<\/em> or \/\u02c8bl\u00e6k.b\u025b\u0279.i\/ blackberry<\/em>. In all those cases, the first part of the compound can also occur by itself: blue<\/em>, straw<\/em>, black<\/em>, but in the case of cranberry<\/em>, this is not the case: *cran<\/em>. The base \/\u02c8kr\u00e6n\/ cran<\/em> occurs only in the single word cranberry<\/em>\u00a0\u2014 it is unique to that word.<\/p>\n Unique roots are rare \u2014 a number of them are found in the names of berries, e.g., mulberry<\/em>, boysenberry<\/em>, raspberry<\/em>. Other examples are luke\u2011<\/em> in lukewarm<\/em>, hinter\u2011<\/em> in hinterland<\/em>, twi\u2011<\/em> in twilight<\/em>, and elig\u2011<\/em> in eligible<\/em>. Unique roots typically come about in two ways: first, borrowing, and second, language change. An example of borrowing is the unique root hinter\u2011<\/em>: when the word Hinterland<\/em> \u2018back country\u2019 was borrowed from German into English, speakers of English recognized the word land<\/em> and treated the entire word as as a compound based on words with a similar structure, such as homeland<\/em>, mainland<\/em>, highland<\/em>, farmland<\/em>, heartland<\/em> and woodland<\/em>, with hinter-<\/em> becoming a unique root carrying the meaning that distinguishes the word hinterland<\/em> from other types of land<\/em>. An example of language change is the unique root luke\u2011<\/em> in lukewarm<\/em>. Luke<\/em> was a free morpheme in Middle English, where it meant \u2018of average temperature\u2019 or \u2018without much feeling or enthusiasm\u2019. It has since disappeared from the language, except as part of the compound lukewarm<\/em>.<\/p>\n Occasionally, speakers start using a bound or even a unique root or complex base by itself, turning it into a free base. This is called back-formation, we will look at it in Section 5.6<\/a>.<\/p>\nBound complex bases<\/h2>\n
Unique roots<\/h2>\n