Some ~roblems of Word-Formation ~'ithin the Framework of 
a Generative Grammar 
Szab6 Zolt&n 
Word-formation has not yet received due attention in 
~enerative grammars, probably becal~se it is an interim 
problem between that of the more-or-less clearly establi- 
shed ~orpho-phonological possibilities and the problem of 
the lexicon, which h~s not yet been worked out (re~ardin~ 
wor@formation see the productive attempts of Chomsky, 7,o 
Worth, ~,otsch, Volotskaia, Zimmer). 
~;y intention is to examine word-formation from a 6e- 
nerative approach, i.e. to trace the possibilities of ~e- 
neratia c derivatives.l shall base my attempLs on e~amples 
drawn from word-form~tion in liuncarian , a lanouaoe excep- 
tionally rich ia for,n~tive devices. 
I. Une first step is the separation of ~rars,~atical and 
lexicol%ical derivatives.\['his is necessary in order to 
show wnafi belonbs to the lexicon and wJ.at does not.To ma - 
ke this de\]i~,itation we can use a relatively simple brans i 
fomrration: derivation (D)~ syntactic construction 
(SC).ilere the meanirl 6 of the stem ($tM) and of the suf- 
fix (Su~) is carried by dif'ferent elements of the sen- 
tactic construction, e. G . STY': predicate, word qualified 
Author's address: Cluj str, Horea 31, ROMANIA 
-2- 
(by an adjective} ~n8 Su~: adverb, adjective.lts formulm : 
if t;tis t~:~nsforrm~tion c~ be c~r~-ied re, t, t:~e deriva- 
t:ive is j'~,:~ati~l; it aces not belon~ to t~.e lexicon. De- 
riv~tives of great semantic corv, ple×ity ~,~ve to be made ~% 
subject to ~ complementary and ~t tAe s~me ti~,e also con- 
tL'ollin c t~'ansforr~ation,, (see my paper: La d61i~t, itation 
6e i,~ d6~'iv~tion lexicologi~ue et 6r~m~r~tiu,~le do hoc~6rois 
l'ai0e 0e l'analyse transform~tionnelle. Caniers de lin- 
Luistique thSorique et ap\[}liqu6e, Buc~z'est, ii, 19.~5). 
2. ?he ~ener~ti~&~ of Cr~mmatical or even of prodtctive 
lexicolo~ical @e~'iva~ives can be carried out ~;,'ith relati- 
v~ ~a,~e in th~ m~nner in which d.ghomsky demo~stra~es ~e 
f'or~.g~tion of, e.g., the p~st tense. Accordin61y , the ~ene- 
r~i~6~ of the ~erJvatives in .~testioi~: do__~b 'tnro'~' 4- ire- 
qtentative----~ c~ob~il 'theow one after the other'. 
/Lnd even if some difficu\]t c~ses ~z'e to be found, they 
c~n be resolved by applyin ~ a no~im~lization (o~" nem~tion 
etc.) transformation ~t the appropriate sta~e of ~eriv~t- 
tion tu the cleneralized khrase-m~rker. 
The productive ~ra~m~tical or even lexicolotic~l deri- 
vatives will not be entered in the lexi6on (\[~.Ghoesky: As- 
sects of the~E~ ~leory of 8271tax 184). 
3- /)eriv~tional processes discussed before raise no se- 
rious c\]ifficulties, for generative 6rammars.But there are 
lot of derivatives that create much more problems. C~onsi- 
-3- 
si@er first of' all the quasi-prodl,ctive p1"ocesses or the 
croat va~'iety of sporadic, sinLle cases, l~ is pretty dif- 
ficult to find rules of any ~enerality tiLat produce deri- 
vatives of such types.'.\[here are, ho~ever, so~e wide clas- 
ses of cases "~it~ v~ryin L 8ecrees of proSuc~iviby which 
~equire non-ac~\]~oc sol~ tJoas. L'~I~ "~ solL, tions ~i~at can be achi 
eyed ~re char~cte,'\]stic of' derqvation ~s a tjpical morpho- 
semantical process. 
i'~e most difficu\]t tusk is t~e ~ener~tin L of' de~'ivati- 
yes uf mreat \]e×ico\]otic,~l ,n~2 morL~J,,J\]ocJual comp\]e×ity, 
e.(. ~szt~los 'cabillet ~ker' (asztal 'table'+-os: aszLa- 
los). 
In the c~se of s\[\]ch ,'ords as ~sztalos one millet seek 
s//ut;actio j L stific~tion for a tl'~:~sflol'mation~ an~ijsis 
from ~ an ~nderlyini~ constriction so that "he is a table :~ 
maker" or "he ~akes table" woul~ @erive~ from the more 6e- 
:~eral and more abstract structtre "he is the ,r, aker of' F" 
or "i~e ~,~kes kM% F"oThis ~ene\['ation f'~ises many se~,,~itic~l 
objections becat~se the as"t;~los is not a 'table makez", 
but a 'ft~'niture maker'oThe solution must be t~ez'efore in 
~ny case se~,,~ntic~l. ~e ~ene~'ation of derivatives oi" this 
so~'t proceeds by ex\[)ansion of tAe meaning, of 5~e stem: 
(i) ~'0 i ~ F 
table~ chai~ bed+ o.. furniture 
If and only if f~\[Oi\] ; ,~h~re Oi is an option~l object, 
f is the semantic distinctive feature 'furiiture' and \[0~ 
-4- 
is the set of semantic distinctive, featuces of the object 
bi.~'herefore sem~ntically \[Ci\]----~ IF\], i.e. the set of 
the semantic distinctive features of any individual 0 i 
can be L.eplaced by the set of semantic distinctive fed, 
tures of F.Aceordin~ly \[~ is the expanded meanin 6 of the 
St~ (formulated with the help of l et6fi S. J~nos). 
(2) St~ ÷ Su~ ------V D 
' f~rni tt re' ' maker' ' ft, rnituz'e-make r' 
• "~nere are still further possibilities. 
C ne of the most important conclusions that can be drawn 
points out ~iso the remote task: the i~eneratin~ of deriva- 
tives reiDires a semantical solt:tion, such as can De wor- 
ked o~t only b~ ~ mathematical means. 
