ARGUMENTS FOR BIDIRECTIONAL MORPHOLOGICAL CHANGE
Camiel Hamans
Anne Vondeling Stichting/European Parliament
Brussels
Abstract
This paper discusses two different directions in morphological change: from full form to suffix and from unspecified segment of a simple word to suffix or even free form. Together with the well known notion of lateral change these two opposite directions show that there is not a preferred direction in language change. In language change opposing forces are at work.
1. Universality of directionality
In their introduction to a collection of papers about language change Fischer and Rosenbach (2000: 19) summarize: “Unidirectionality is said to apply on all linguistic levels: the semantic (fully referential > bleached/grammatical meaning; less subjective > more subjective), the syntactic (lexical>grammatical; less bound > more bound) and the phonological (full phonological form> reduced phonological form)”.
In this paper it will be shown that morphological changes may go in both directions: from less bound > more bound or from more substance > less substance but also the other way round from less bound > more bound or less substance > more substance.
There will not be claimed that changes are reversible. So far no examples have been found of changes which first went in the direction from A>B and later on took the reverse direction from B>A completely.
As is well know, directionality of changes is a major issue in the theory of grammaticalization (for an overview of the discussion see for instance Fischer and Rosenbach 2000:19-23), where the unidirectionality of the grammaticalization process has been taken as almost axiomatic in a majority of the relevant literature. Whether the examples adduced here should be considered as a counterargument against the unidirectionality of grammaticalization, thus as examples of degrammaticalization, or as instances of anti-grammaticalization or of lexicalization will not be discussed here. See for this discussion Brinton and Traugott (2005), Norde (2009) or Hamans, Fisiak and Jahr (2009).What counts here is the possible direction of changes.
2. Lateral change
Before turning to examples of changes in a certain direction we first will address the question whether changes have to follow a direction at all. This question, reformulated as whether changes should imply a decrease or an increase of grammaticality necessarily, has been discussed by Brian Joseph (2005) eloquently. In his comparison of Ancient to Modern Greek he found numerous examples of affixes, which remain affixes but received a new grammatical function. So in terms of grammaticality the forms stay as grammatical as before. They do not climb or descend the cline, but move ‘laterally’ on the cline. Just as conversions from noun to verb or adjective to noun.
Norde (2009: 62) exemplifies the notion of lateral shifts with an example from Blake (2001:180). It concerns a case of transferring a system of case marking to a system of tense/aspect marking in Kala Lagau Ya, an Australian pidgin language spoken on the Western Torres Strait islands.
Table 1. Kala Lagau Ya case and tense/aspect marking
(Blake 2001:180)
______
Marker Case Tense/Aspect
-n ergative (nouns)/ completive
accusative (pronouns)
-pa dative/allative incompletive
-pu comitative habitual
-ngu ablative yesterday past
-nu locative immediate past
The shift may be illustrated by example (1), where –pa functions as a marker of a dative case in (1a) and as a marker of an incompletive aspect in (1b):
(1a) Nuy ay-pa amal-pa
He food-Dat mother-Dat
‘He [went] for food for mother’
(1b) Ngoeba uzar-am-pa
1Dual.Inclusive go-Dual-Incompletive
‘We two will go (are endeavouring to go)’
This shift, which clearly is a change in function, does not imply any movement upwards or downwards the cline of grammaticality. It simply is a lateral movement. So it is evident changes do not necessarily have to move to a greater or lesser extent of grammaticality.
3. Cline of lexicality
Traditionally the origin of some derivational affixes has been described as an instance of directional change: from a construction of full words or a full word only, possibly via compound, and an intermediate stage of semi-suffix/suffixoid to derivational affix.
Hopper and Traugott (1993:7) call this cline ‘a cline of “lexicality”, which would include derivational affixes such as –ness in ugliness, etc., and presumably also such positions as syntactic phrase, compound, and affix, e.g.:
(2) a basket full (of eggs…) > a cupful (of water) > hopeful’
This change is not an incident. The English adjective full followed the same path as its counterparts voll in German and vol in Dutch:
(3) ein Korb voll / voller frischer Eier, ‘a basket full of fresh eggs’ > ein Hand voll (Handvoll) Geld, ‘a handful of money’ > eine Hand voll (Handvoll) Leute, ‘a handful of people’> hoffnungsvoll ‘hopeful’
(4) een mand vol verse eieren, a basket full of fresh eggs > een handvol geld, ‘a handful of money’> a handvol mensen, ‘a handful of people’> hoopvol ‘hopeful’.
Although full, voll and vol show the same pattern and direction of change the extent of lexicalization may differ:
(5) two handfuls of eggs
(6) zwei Handvoll / Hand voll / Hände voll Eier
(7) twee handenvol eieren
In English the plural ending comes at the end of the complete form handful, German Handvoll (Hand voll) does without plural marker, although Hände (pl.) voll has been attested as well, whereas in Dutch the plural suffix still comes after the noun hand.
For the change from phrase to affix this does not play any role. It only shows that the change does not go at the same speed in all languages. Also the question that Marchand (1969: 292) raises and is not able to answer, whether Middle English two spoon ful ‘was felt to be a compound or a syntactic group’, is not important for the direction of the change either.
This development from less bound form to more bound form is quite normal. Noun forming suffixes such as –dom, –hood or ‘semi-suffixes’[1](Marchand 1969: 356-358) such as –like and –wise go back on full nouns originally. As to the origin of suffixes Marchand (1969: 210) distinguishes between two types of suffixes: ‘a. the suffix was once an independent word but is no longer one (in the case of semi-suffixes there still is a corresponding lexeme CH) b. the suffix has originated as such, usually as a result of secretion’. Here we are only interested in the first type.
In Old English, just as in Old High German and Old Dutch, the forms corresponding to dom and hood, or to their German and Dutch counterparts –tum and –dom and –heit and –heid, were still independent nouns (Marchand 1969: 210). The same with like, respectively German –lich and Dutch –lijk (cf. Köbler 1995:160) and wise, respectively German –weise and Dutch -wijze (cf. Koziol 1972:275). From these few examples which may be supplemented by numerous others, one may conclude that a change from more to less grammaticality or substance used to be quite a normal process.
4. Recent suffixations
A change from a full noun to a suffix or suffix-like form normally via compounds is still a productive process, though it is more complicated to find clear examples, since in synchronic language use the time span mostly is too short to conceive the whole development. One has to find the different stages of the development next to each other in modern usage.
The standard example comes from Booij (2002:), who borrowed it from Van Marle(1978).
Dutch
(8) boer ‘farmer’ groenteboer ‘greengrocer’
melkboer ‘milkman’
eierboer ‘eggman’
(8a) kolenboer ‘coal trader’
sigarenboer ‘cigar seller’
patatboer ‘chip shop owner’
(8b) lesboer ‘teacher’ (neg.)
In the compounds groenteboer, melkboer en eierboer the semantic aspect ‘farming’ originally still played a role, since it were the farmers themselves who hawked vegetables and milk around. Later the aspect ‘seller’ became more prominent, see the examples under (8a). Finally in lesboer not only the original meaning ‘farmer’ became irrelevant, but also the aspect of selling, although one could claim that teachers here are described as people who are not very enthusiastic about or motivated by the educational goals of their profession and only hawk their lessons around. Maybe this interpretation offers an explanation for the negative meaning of the word.
As the examples in (8, 8a & b) make clear the process starts with compounding. At a certain moment in the development the meaning of the second noun becomes bleached and less specific (8a). One stage further the segment –boer lost its independency completely and became a bound element, or in the words of Booij (2002:153):
‘It will be clear that such a semantic development will be furthered by paradigmatic compound formation: very probably, kolenboer has been coined on the basis of the relation: groente – groenteboer / kolen – kolenboer. The development of boer into a suffix is also an example of grammaticalization: a lexical morpheme becomes a derivational suffix through reinterpretation.
As we will see further, reinterpretation or reanalysis followed by analogical extension to a paradigm or series is crucial for the rise of new elements.
Although the groenteboer paradigm is the classical example for the rise of new suffixes, it is by far not the only one. In Dutch and German one may find ‘suffixoid’ such as –werk, –kracht / –kraft etc. which comes from the nouns werk / Werk ‘work’ and kracht / Kraft ‘power’.
(9)Du handwerk ‘manual work’ but also ‘handiwork, hand made, handicrafts’
strafwerk ‘lines, imposition’
(9a) dakwerk ‘roof construction’
vakwerk ‘craftmanship’
(9b) vuurwerk ‘fireworks
uurwerk ‘clockwork’
‘clock’
(10) mankracht ‘manpower’
paardekracht ‘horsepower’
(10a) arbeidskracht ‘working power’
‘worker’
uitzendkracht ‘temporary worker’
(10b) leerkracht ‘teacher’
vakkracht ‘professional’
In German one finds a similar development, although the word for ‘work’ became Arbeit rather early:
(11) Autowerk ‘car factory’
‘garage’
(11b) Handwerk ‘(handi)craft, trade’
Feuerwerk Fireworks
Uhrwerk ‘clockwork’
‘clock’
(11b) Laubwerk ‘leaves’
Blätterwerk ‘leaves’
(12) Menschenkraft ‘manpower’
(12a) Arbeitskraft ‘working power’
‘workerer’
(12b) Schreibkraft ‘clerk’
In all these examples one may conclude to bleaching of the specific meaning of the second noun to a more general or collective meaning. In the last examples (b) the second part shows no direct semantic relation with the original form any longer.
This will become even more striking when one takes a look at the following German examples, in which the second part originally was (the diminutive of) a first name (Elsen 2011:70)
(13) Trödelheini Trödel ‘junk’
Meckerheini meckern ‘nag’
Trödelfritze
Meckerfritze
Trödelliese
Meckerliese
Schnatterliese schnattern ‘cackle’
Kleckerliese kleckern ‘mess’
Heulsuse heulen ‘cry’
Nölsuse nölen ‘linger’
Transuse Träne ‘tear’
In Dutch one finds a few similar examples with first names as a second part:
(14) smeerkees ‘dirty fellow’ from smerig ‘dirty’ and Kees from ‘Cornelis’
lulhannes ‘wanker’ from lullen ‘bullshit’and Hannes from ‘Johannes’
Also here one may conclude to a less specific or more generalized meaning and as these examples show the process from compounding to suffixation is still a common process in modern languages.
5. Interpretation as a composite
The development in the other direction, from inseparable segment via reinterpretation as quasi-complex word to combining form or suffix-like form and sometimes to full noun has been documented as well, usually in the context of degrammaticalization (Ramat 1992 and 2001, Hamans, Fisiak & Jahr 2009 and Norde 2006 and 2009).
The example Ramat produces of –ade is a clear one. It starts with orangeade or lemonade and finally leads to an independent general noun ade. However what Ramat suggests about the abruptness of the change seems to be unlikely. As Brinton and Traugott correctly claim (2005 80/81 note18) the process starts first with a series of –ade(s): ‘however it seems more likely that orangeade itself did not give rise to –ade; rather, this derivational morpheme (…) came to be relatively productive (as in orangeade, lemonade, etc.) and then came to be used independently as a noun.’ So the process must be gradual.
The same applies to the example tig (Hamans 1993 and Norde 2006) in Dutch. –tig is the unspecified final segment in Dutch numerals for tens:
(15) twintig ‘twenty’
dertig ‘thirty’
veertig ‘fourty’
etc.
Although the forms look like if –tig is a regular suffix it is not. Twin– , der– and veer– are not lexical items of Dutch, these three numerals are fully lexicalized. Nevertheless the form –tig has been isolated and seen as part of composite or syntagma (Marchand: 1969: 211), just as happened with the French ending –ade that was borrowed into English via French words. Later tig became an independent form, an unspecified numeral meaning ‘quite a lot’, and as Norde (2006) showed it even became an adverb subsequently.
The serial nature of productive use of the element seems to be essential. In the jargon of the Dutch navy one finds the very unusual noun taris (Hamans 1993), referring to an officer of the administration. This form only can have originated from a series of words ending in
–taris as for instance:
(16) secretaris ‘secretary’
notaris ‘notary’
mandataris ‘mandatary, representative’
Such a process of reanalysis of simple words is not unusual. Lehmann (2002:13) calls it folk etymology, just as Szymanek (2005). Lehmann defines this phenomenon as the ‘bestowing [of] structure onto a hitherto opaque expression’. Since folk etymology may suggest a primitive way of reanalyzing linguistic material, the term appears not to be very appropriate, especially since the process turns out to be so common and frequent. Therefore here Marchand’s terminology interpretation as a ‘composite’ will be used.
Usually this type of change does not yet result in free forms[2], and that is why this kind of change may have been hardly noticed in this discussion, but reanalysis of opaque structures into more transparent structures containing quasi-prefixes or quasi-suffixes is a common pattern. Marchand (1969), Koziol (1972) and Hamans, Fisiak and Jahr (2009) list numerous examples mostly of English pre- and suffixes usually of foreign origin but not always. That foreign forms will become reanalyzed by speakers who do not recognize these lexemes is of course what one may expect. We give a few examples of this change in (17), (18) and (19):
(17) cavalcade landscape bootlegger panorama lemonade
aerocade seascape footlegger cinerama orangeade