Semantic Cuisine

Author(s): Adrienne Lehrer

Reviewed work(s):
Source: Journal of Linguistics, Vol. 5, No. 1 (Apr., 1969), pp. 39-55
Published by: Cambridge University Press
Stable URL: http://www.jstor.org/stable/4175017 .
Accessed: 03/11/2011 15:42
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Linguistics.

http://www.jstor.org

JL5 (I969)

I-I92

Printed in Great Britain

Semantic cuisine
ADRIENNE LEHRER
Departmentof Languageand Linguistics,Universityof Rochester,
RochesterN.Y. 1461 I
(Received30 NovemberI967)
Structuralsemanticsseeks to discover certain relationshipsamong the words in
the vocabulary of a language.' This area of linguistic analysis has received
relativelylittle systematicstudy by most linguists until recently(Gleason, I962:
86), and the vocabularyhas been considereda ratherunorganizedset of items.
One approach which has been successful in semantic analysis, however, is
the FIELD THEORY. According to this view the vocabulary of a language is
organized into lexical or conceptual fields, and the items within each field are
tightly structuredwith respect to each other. The field approachhas been successfully applied to some semantic areas, for example, kinship terms, colour
terms, plant taxonomies and other fields with clear denotationalreferents. In
some semantic analyses, the field theory is implicit, although the term FIELD
is not used, but a set of vocabularyitems which are relatedare analysedtogether
(Bierwisch, I967; Bendix, I966).
However, most semantic analyses of this sort deal only with paradigmatic
contrasts, e.g. a set of commutablenouns, verbs, or adjectives,but such a set
constitutes only part of a field. Syntagmaticpresuppositions(kick with a foot,
lambs bleat) and productive or partially productive word-building processes
also constitute an importantpartof the lexical field and should be studied along
with the paradigmaticsets (Lyons, I963: 78, I968: 428). The importanceof
syntagmaticpresuppositionhas long been advocatedby Porzig (I934, I950).
I have applied this approachto the lexical field of cooking terms in English,
which includes a basic set of verbs and formallyand semanticallyrelatednouns
and adjectives. This analysis shows, I believe, that the items in the field are
highly organized and that the semantic and syntactic analyses tend to support
each other.
My assumptionsand conclusionsagree in generalwith those of Lyons (I963).
In particular, I assume that a transformationalgrammar as developed by

Chomsky(I957,

I965,

etc.) andothersprovidesthe bestbasisfor the statement

of the semantic relations that hold between lexical items. In this respect I am

also in agreementwith Katz & Fodor(I963), Katz & Postal(I964) and Katz
[i]

This article is based upon my dissertation (Lehrer, I967). I wish to thank John Lyons
for many helpful suggestions and for his encouragement. I am also indebted to Angus
McIntosh and Donald Reiff for their thoughtful comments.

39

JOURNAL OF LINGUISTICS

(I966, I967). However, I invoke the notion of SITUATIONor UNIVERSEOF DISCOURSE in the establishmentof semantic fields, and I accept that 'the context of
situation ... must be held to comprehendall the conventionsand presuppositions
acceptedin the societyin which the participantslive, in so far as these are relevant
to the understandingof the utterance'(Lyons, I963: 63; cf. also Lyons, I968:
4I3). This does not imply that we must first representall the knowledgespeakers
have about the world before we begin a semantic analysis (cf. Katz & Fodor,
I963: I78), but merelythat it is legitimateand necessaryto use such information
in the course of semanticanalysis.The notionof 'universeof discourse'is relevant
to semantic analysis in that certain lexical items contrast paradigmaticallyin
some fields but not in others.
Reference is distinguished from meaning (or sense), but it is accepted that
referencemay be appealedto in establishingthe meaning of a lexical item. The
meaningof a lexicalitem is describedin termsof the relations(of incompatibility,
antonymy, hyponymy, synonymy, etc.; cf. Lyons, I968: 443-170) that hold
between it and other lexical items. The results of an investigationcarriedout in
terms of this view of meaning can then be converted, as I will demonstrate
below, into a componentialdescriptionof the kind proposed by Katz & Fodor

(I963)

and othersemanticists.

In what follows I use the terms 'word' and 'lexeme' interchangeablyto stand
for 'lexical item'.
The field of culinaryterms in English is most convenientlytreated by taking
verbs as basic in describing the paradigmaticcontrasts,although in a few cases
lexemes may be synchronically derived from nouns. The set of verbs is as
follows:
cook, boil, simmer,stew, poach, braise,parboil, steam, reduce,fry saute',pan-fry,
French-fry, deep-fry, broil, grill, barbecue,charcoal (or charcoal-broil),plank,
bake,roast, shirr,scallop,brown,rissoler,sear,parch, toast, burn(unhappily)and
flamber.
In addition, a number of semanticallycompound terms are found (French-fry
is grammatically,but not semantically,compound):
steam-bake,pot-roast,oven-poach,pan-broiland oven-fry.
Lexemes such as smokeand thaw are used in food preparation,but they do not
belong to this narrow field. The more general field of food preparationwould
include not only the subset of cookinglexemes,but subsetsfor mixing, chopping,
coating, adding ingredients,separatingsubstances,preserving,etc.
The word cookhas three levels of generality.In its most generalsense (cook,)
it means to 'preparea meal', belongingto the field of householdtasks along with
dust, wash, vacuum,etc., and belonging to the field of occupations,with repair
electricalappliances,sweepchimneys,etc. My use of SENSE, it should be noted, is
40

CUISINE

SEMANTIC

somewhat differentfrom what is often meant by this term, for I do not wish to
suggest that these three senses are necessarily distinct. In fact, there is great
overlap.The firstsense is the least markedand the third sense is the most marked.
A slightly less general sense of cook (cook2)contrastswith bake. (Bake has a
general and specific use, too, but the one here is general, bake1.)The semantic
distinction is that baking refers to the preparationof cakes, cookies, breads,
pastriesand other things which are sold in bakeriesand preparedby professional
bakers. Cooking refers to the preparationof most other kinds of foods. Cook2
and bake1are the only lexemes in the field which occur intransitivelywith an
is acceptable,but not *I saute'
animatesubject. I can cookor I saute'mushrooms
(except in those circumstancesin which the correspondingagentive -er nouns
might be used 'absolutely': see below. For example, a chef might say to his
subordinatesX will grill, Y will saute',etc.). Sentences like He is boilingor He
fried would not be interpretedwithin the culinaryfield.
The most markedsense of cook(cook3)involves the applicationof heat which
produces an irreversiblechange in the object (food) cooked. The most specific
use is incompatiblewith cool and chill, whereas cool and chill are included in
the more general sense of cook because the preparationof some foods, e.g.
puddings, requires both heating and cooling, and we would still wish to apply
the term cookto the preparationof such foods.
The lexical field covered by cook3can be divided into four main categories
headed by the lexemes boil,fry, broil and bake2(the specific sense). The terms
constitute an almost closed set and are largely incompatiblewith one another.
These four lexemes, then, are hyponyms of cook3.
Of this set boil has the most complex subset of terms included in it. Boil
has a general and specific meaning. Both senses involve cooking with liquid,
usually water or a water-basedliquid (stock, wine, milk), but oil (grease, fat)
is excluded. The phraseboilingin oil is semanticallyacceptablein many contexts
because oil, like other liquids, has a boiling point, can have things immersed
in it, etc. But in the culinaryfield this phrase is unusual. (Actually some modification is requiredsince we can speak of letting a kettle of oil or butter boil or
simmer; but instead of boilinga solidin oil orfat, the term deepfry is used.)
The specific sense of boil, boil2 adds the component of vigorous action (occasionallyreferredto as a full boil, though the modifierneed not be present for
this meaning), and this sense contrasts with simmer.Simmermeans 'to cook
just below the boiling point' without the rolling bubbles which characterize
boil2. Boil1, the general sense, is unmarkedwith respect to the vigour of the
process and includes both simmerand boil2.Simmerand boil2collocatewith both
liquids and solids. However, if what is boiled is a solid, it is presupposedthat
some liquid is present. The meat boiledin a dry pan is anomalous,although one
might boil tomatoes without additional liquid since the naturaljuices provide
enough liquid.
4'

JOURNAL OF LINGUISTICS

The lexemes poach and stew are hyponyms of simmer(and therefore of boil1
since the relationshipis transitive). To poach is 'to cook by surroundingwith
simmering(not boiling) water or other liquid using care to retain shapes' (Betty
Crocker,I956: 14). Foods which are typically poached are eggs, fish and fruit.
Stewing is 'a long slow method of cooking ... in a liquid which is kept at
simmering point' (Good HousekeepingCookingEncyclopedia,I964: 407), and
applies to meat, vegetables and fruit. Since fruit can be poached or stewed, and
since poaching and stewing are extensionallyidentical, the possibility arises of
treatingpoach and stew as synonyms, used in free variationwith fruit and in
complementarydistributionwith other foods. This proposalis not satisfactory,
however, since the purpose of these cooking processes is semanticallyrelevant,
and the purposes are different.The purposeof poaching is to retain the shape
of the food, while that of stewing is to make the food softer, and a long cooking
time achieves this purpose. In the case of fruit, where poaching and stewing
seem to be the same, the differencein meaning still exists. One would poach a
fruit to preserve its shape and stew it to make it soft, although these ends can
often be accomplishedsimultaneously.We speak of stewedtomatoesbut hardly
of poached tomatoesbecause tomatoes do not usually retain their shape when
cooked in water. Here is a case where we can separatereferenceand meaning.
(Poached eggs and boiled eggs differ in their reference in that the former are
cooked without shells while the latter are cooked with them. While it would be
possible to define poach and boil to account for this difference, the semantic
descriptionwould be complicatedto a point of diminishingreturns.It is therefore
preferableto treatpoachedegg and boiledegg as unitary lexemes. Another term,
coddle,seems to collocateonly with egg, so that coddledegg may be considereda
lexemic unit. Since cookbooks and dictionaries disagree as to exactly how to
coddle an egg, it is difficultto abstractany precise meaning for coddle).Poach
and stew collocate only with solid (or solidifiable)foods, which is logical, given
the purposeof the processes.One cannotpreservethe shape of a liquid by boiling
it in water or make a liquid softer.
Braise refersto two processescarriedout sequentially.Braisedfood (meat and
vegetables) is 'browned in a little fat, then cooked in a little liquid over a low
heat in a covered pan' (Gastrononmique,
I960: 303). After the first process
(browning,see below), the rest of the cooking is done by stewing. The semantic
overlap of stew and braiseis parallelledby identical collocationalpreferences.
All of the cookbooksconsulted as well as the ShorterOxfordEnglishDictionary
agree that braisingis done in a tightly coveredpot, and thereforethe use of a lid
is probably to be specified in a semantic description of braise. This further
component makes braise a partial hyponym of stew rather than a synonym:
Braise = stew and [+Lid]
Braise is also a partialhyponym of brown.
42

SEMANTIC

CUISINE

To parboilis 'to partiallycook food in boiling water'. The term is used only
with solids, and the simmer-boildistinctiondoes not apply. Parboilcontrastswith
stew with respect to the length of time involved (short v. long). Parboil brings
up a presupposition that should perhaps be incorporatedinto the semantic
description- that at least some of the other cooking lexemes assume that food
will be thoroughly cooked. The sentence These cookedpotatoes are half raw
is ratheranomalousunless stress and intonationpatternssuggest an ironicaluse
of the word cooked.
The two remaining lexemes in the boil set are steam and reduce,which are
incompatible with simmer.In order to steam a food, 'the water must not be
allowedto go off the boil' (GoodHousekeeping,
I964: 406) and the food is not submergedin the liquid as it is in the other boiling processes.To reduceis 'the process of boiling a mixture ... in an uncoveredpan . . . to evaporatesurplusliquid
and give a more concentratedresult' (PocketGuideto GoodCooking,1955: 234).
The purpose of the process- to reduce the bulk - is a relevantcomponent, and
a vigorous boil and an uncovered pan accomplish this more efficiently than
simmering a liquid in a covered pan. Although there may be some doubt as
whether to include [- Lid] and [+ Vigorousboil] as componentsof the meaning
of reduce,I will do so since these components occur in the analysis of other
lexemes subsumed under boil1. Reducecollocates with liquids and steanmwith
solids.
Figure i summarizes the terms which are hyponyms of boil. This chart is
arrangedwith respect to the vigour of the boiling action, while one arranged
according to collocations preferences or restrictions would look somewhat
different.
boil1 (unmarked)
simmer
poach

stew

boil2 (marked)
parboil

steam

reduce

braise
Figure I
(A lexeme is a hyponym of a term above it; lexemes on the same line are incompatible if separated by a vertical bar.)

The set of cooking words headed by fry includesfry, saute,pan-fry, Frenchfry and deep-fry.Frying is characterizedby the use of fat (oil, grease)in cooking,
although the fat may be present in the food being fried. The invention of nonstick frying pans requiresa modificationof this analysis.For some speakers,fry
is characterizedby cooking a food in a frying pan or similar utensil above the
43

JOURNAL

OF LINGUISTICS

heat source. Water, however, must not be used. Such an analysis requires
disjunct components.
[+ Fat] v. [+ Cookingin frying pan above heat]
A further complication is that for some speakers,fry contrasts with deep-fry,
althoughthey will admit that deep-fryingis a kind of frying. This case suggests
that fry may have a general and specific use. Pan-fry is sometimesused for the
specific sense. Pan-fry is partiallysynonymouswith saute',but sauteing requires
fat while pan-fryingneed not. French-fryand deep-fryaresynonymous,involving
a relativelylarge amountof fat, that is, enough to cover what is being fried, and
both are incompatiblewith saute, which is characterizedby a small amount of
fat. Fry collocateswith solids.
The term broil and its hyponyms-grill, barbecue,charcoal and plankbring up the problems of dialect differencesand range of reference. To broil
is 'to cook directly under a heating unit or directlyover an open fire'. Although
broilis a common Americanterm, grill is used in Great Britain instead. ('Broil
is the older English word and was currentin British cookerybooks up to about
I900': New YorkTimesCookBook,685.) Broil andgrill arepartiallysynonymous,
but grill has a slightly wider range of applicationthan broil. To grill is 'to broil
on an open grill or cook on a griddle'. Cooking a food (e.g. pancakesor hamburgers)on a griddle is referentiallymore like frying than broiling,so that grill
may have to be considered a partial hyponym of fry as well as of broil. (We
could establishgrill as the more general term, with broil a hyponym of it, but
this is not satisfactoryfor AmericanEnglish becausebroilis the commonerword,
used at a rankequal to fry, bakeand boil. Grill is less common.)
The lexeme barbecuehas two senses. The one that is relevant to cooking is
'broiling over a bed of glowing coals'. The component added to those of broil
is that the sourceof heat is hot coals. The othersense of barbecueinvolvescooking
a food with a special sauce, usually containing tomato, vinegar and seasoning.
By using such a sauce one can barbecue meat by baking it as well as by
broilingit.
Charcoal(or charcoal-broil)is synonymous with barbecuewith respect to the
method of cooking, but the use of a sauce is not implied.
Toplank is 'to cook (usually meat or fish) on a wooden boardin a hot oven or
under a broiler'. There is some semantic overlap with bake, but plank can be
considered a hyponym of broilcontrastingwith barbecueand charcoal.
The set of broillexemes is restrictedto collocationswith solids, mainly meat,
poultry, fish and occasionallyvegetables.
To bakeis 'to cook by dry heat in an oven' such that the heat acts 'by conduction and not by radiation'.That is, the source of heat is indirect rather than
direct as for broiling.
The generalsense of bake,as suggestedabove, bake1,refersto the preparation
44

CUISINE

SEMANTIC

of bread, pastries, etc., and English has a number of pairs of terms for these
products before and after they are baked: dough-bread,batter-cake.The specific
use of bake, bake2, is another method of cooking, contrastingwith fry, broil
and boil, and most bakery products are preparedby this method. (Doughnuts
may be fried, however.)Bake collocatesprimarilywith solids.
The principallexeme relatedto bake2is roast,but there is a semanticdifficulty
in this classification.'In its true sense, roasting means cooking by direct heat
in front of an open fire ... but the modern method of cooking in a closed oven,
though normally called by this name, is really baking' (Good Housekeeping
CookingEncyclopedia,364). We see that roast overlapswith broil semantically
in that some roasted foods (meat, marshmallows)are cooked over (or under)
an open fire. One alternativeis to establishroast, as an equal member of the set
of lexemes bake, broil, boil and fry, but roast does not contrast with bake and
broiland is not completelysynonymouswith either term. Nor is there adequate
motivation for establishingtwo senses of roast, one synonymouswith bake and
the otherSYNONYMOUS with broil.It is moreaccurateto attributeto roast a range
of meaning which overlaps with bake and broil, making roast only a partial
hyponym of bake.
Shirr and scallopare the remaininghyponyms of bake. Shirr, used primarily
for eggs, means 'to bake in a small shallow container'. To scallop is 'to cook
and serve in a scallop shell or a dish like one' (English Cooking,I960: 232).
But the scalloped foods are usually cooked in a cream sauce, and this sauce is
a semantic component,perhapsmore importantthan the baking dish used.
The lexemes included under cook which have been discussed so far can be
summarizedas in Figure 2. The range of roast extends over bakeand broiland
that of grill over broilandfry.

Cook
boil
simmer

(full) boil

saute
pan-fry

bake

broil

fry
French-fry
deep-fry

grill

barbecue
charcoal

plank

roast.

shirr

scallop

See Figure i

Figure 2

Brown heads another subset of lexemes subsumed under cook. This set is
relatedto thefry-broil-bake-boil
set but cannotbe placedin a hierarchicalrelationship to it. To brownis 'to give a dish ... an appetizinggolden-browncolour by
472) or by frying.
placing it under the grill or in a hot oven' (GoodHousekeeping,
Boiling is excluded because boiling a food does not brown it (unless the water
cooks out, in which case one is no longer boiling the food). Therefore, brown
implies not boil.
45

JOURNAL OF LINGUISTICS

Burn in the culinaryfield is relatedto brownas a gradablemember of a pair.

To burnis 'to brown too much'. One can burn a food by overfrying,overbaking
or overbroiling.Boiling, for the reasonmentioned above, is excluded.
Brownhas four hyponyms: sear, rissoler,toast andparch. To toast is 'to brown
by direct heat', that is, by broiling; to parch is 'to brown with dry heat', that is
by baking, and to sear and rissolermean 'to brown by frying'. Searing is done
quickly, rissolering,slowly. The four hyponyms of browncollocate with solids,
but browncan be applied to things like butter (which would be in a liquid state)
in additionto solids.
The lexeme flamber, meaning 'to flame with brandy or fortified wine', is
another hyponym of cook, but it is a comparativelyrare word. Its rarity, of
course, reflectsAmericanand Britishculinaryhabits. Althoughflambercontrasts
with fry, broil, bake and boil, I shall tentatively classify it with the brownset,
since flamingfood in this way browns and crisps the surfaceof the food. Moreover, the thing to be flambeedis often previouslycookedby some other method
(e.g. duck, steak).

brown- -*burn
parch

toast

sear

rissoler

flamber

Figure3

There are a number of compound lexemes in the culinary field: steam-bake,

pot-roast,oven-poach,
pan-fry; and presumablymany new ones can be generated:
charcoal-bake,simmer-grill,oven-brown,etc. How these compounds should be
classifiedsemanticallydepends on whether priorityis to be given to referenceor
formalsignals.Pot-roasting,for example,is almostthe same as braising,although
pot-roast has a narrowerrange of collocations than braise. This would make
pot-roasta hyponym of boil rather than of bake. Even if reference is selected
as the determining factor, it is difficult to classify a term like oven-poach,
which is referentiallylike baking in some respects and like boiling in other
respects.
The compound lexemes clearly show the limitation of hierarchicalanalysis.
The rangeof lexemes like roastandgrill partiallyextendingover two other terms
also revealsthis limitation.A componentialanalysismight handlethese semantic
relationshipsmore conveniently. In a componentialanalysis one looks for the
smallest number of components which will provide all of the relevant information, and so the most generalelements are sought. However, some definitions
call for quite specific and unique components, as 'in a small shallow container'
in the case of shirr.
46

SEMANTIC CUISINE

An examinationof even the very specific components reveals that they can
be grouped into a small number of classes. For example, a number of cooking
terms involve, as part of their meaning, a certain kind of utensil. We can state
that a special utensil is relevantwithout specifyingwhat that utensil is. Although
this would not resultin a completesemanticanalysis,it shows that the seemingly
large number of unique components can be determined by a few relevant
parameters.
The following components and parameters are involved in the cooking
vocabularyof English:
Use of water (wine, milk, etc.)
Use of fat
Direct or radiatedheat v. conducted heat
Vigorous v. gentle cooking action
Long v. short cooking time
Large amount v. small amount of some
substance

[+ Liquid] v. [-Liquid]
[+ Fat] v. [-Fat]2
[+ Direct] v. [- Direct]
[+ Vigorous] v. [- Vigorous]
[+ Long time] v. [-Long]
[+ Large] v. [-Large]3

The following specificationsinvolve unique components, but a binary notation

can show whether a given parameteris relevantor not to the semantic analysis:
Kind of cooking utensil
Special ingredientadded (sauce)
Special purpose (e.g. preserveshape)

[+ Relevant]
[+ Relevant]
[+ Relevant]

In Figure 4 where certain distinctions do not apply or where parametersare

irrelevant,4there is a blank, although o could be marked.As in phonology, '-'
implies a feature,not the irrelevanceof one.
This chart handles compound lexemes, but does not deal with grill and fry
adequately. A statement allowing disjunct components would be more satisfactory. For example,
Grill: [-Liquid], [+ Direct heat v. on griddle]
or
Fry: [-Liquid], [+ Fat v. special utensil (frying pan)]
[2] This is somewhat redundant because usually [+ Liquid] implies [-Fat]

and [+ Fat]
implies [-Liquid].
[3] The use of binary notation is purely a convenience, and there is no implication of a
binary structure of semantics. One could easily establish several degrees of vigorous
cooking action, length of cooking time, etc., but it is not necessary to do so for this field.
[4] This analysis lists only the components necessary to distinguish lexemes from one
another within the field. An analysis based on the theory outlined by Katz, Fodor,
and Postal would assign additional semantic markers to most terms: e.g. (Process
(Irreversible)) (Application of Heat) and perhaps others. Components such as [Liquid]
would be given the status of distinguishers, as would the highly specific components,
such as [Utensil: small dish].
[5] Hot coals.

47

JOURNAL OF LINGUISTICS

The main advantageof a componentialanalysisis that it providesinsight into

how and why partially anomalous sentences are easily understood. Consider
an instruction to poach an egg in oil. Although poach calls for [+Liquid] and
[- Fat] and the instruction blocks these components, there are still some
components left to provide some understandingof the sentence: [+Preserve
shape], [+ Gentle cooking action] plus those in all cooking words (Heat),
(Process), etc. Communication is likely to break down, however, when all
componentsare blocked,as Fry the eggin waterin a coldoven.
Certain asymmetriescan be noticed in the analysis. For example, a special
utensil is specified for terms like braise,roast and fry, but not for boil, broil,
etc., even though certainutensils (or appliances)are characteristicallyused when
braising and roasting, and this informationis common knowledge. The asymmetry is intentional, however, and components (many embracinginformation
about the world) were introduced as needed in order to distinguish lexemes
which we wish to distinguish. The dilemma in semantics is this: on one hand,
'thereis no seriouspossibilityof systematizingall the knowledgein the worldthat
speakersshare'(Katz & Fodor, I963: 178) and on the otherhand such knowledge
is often used to understand(disambiguate,etc.) utterances.But the dilemmais in
fact not real. We draw on as much informationaboutthe worldas we need to for
semantic purposes. There is no need to draw any a priori distinction about a
speaker'sknowledge about the world and his knowledge about his language.
In considering the grammaticalproperties of the set of lexemes described
above, it becomes apparent that not all of the forms are basically verbs.
Some verbs, which are derived from other parts of speech, have certain
syntactic limitations, e.g. plank and scallop. Whereas The meat is stewing or
The salmonis poachingare acceptablesentences, *The meat is plankingor *The
potatoesare scallopinggo beyond the fringe of grammaticality.The verb plank
is derivedfrom the noun, and scatllopprobablyfrom scallopshell (with additional
semanticextension). Interestinglyenough, it is the least generalwords which do
not fit into the patternwith the rest of the set. A potentialproductivedevice for
English is to derive new verbs from utensils or vessels. The meaning of such
new forms would be 'to cook [a food] in (on) an X by (method most suitable)'.
To casserolecelerywould mean 'to cook celery in a casserole dish', and if this
were done by baking, casseroleas a verb would be a hyponym of bake in the
lexical field of cookery.It seems to me that only a semantictheorythat dealswith
the universeof discourseor socioculturalsettings can make semanticprojections
with much accuracy.When a verb is derived from a noun, there are a large
number of possible meanings, but a context or semanticfield together with the
presuppositionswhich speakersshare about that field greatly limit the number
of meanings.
Cookand bakeare the only lexemes in the set which can freely occur intransitively with an animatesubject (or alternatively,with a deleted object). I cook,
48

SEMANTIC

r4

CUISINE

Other relevant parameters

Collocates

~~~~~~~~~

.0

+-

C:
-

I4-)I

~ +4~

6tew

4)

poach

braise

+
+

+
+
+

ote

+ Preserve shape
i

.
+ Rack, sieve, etc.

+
+

+Reduce bulk
+Frying pan

charcoal
plank

+
+

+Frying pan
+

_
+

+
_+Shell+

?(Griddle)

+
-

?Vood board
.

+
_

roast
shirr
scallop
brown o

burn

_
_

+ BarBQsauce
,

dish

-+Small

+ Shell

+ Cream sauce
+Brown surface

+
+

-+,

flamber

steam-bake
pot-roast
oven-poach
pan-broil
oven-fry

+
+
+

--

--

--

+
+

+
+
+
+

-+,

+ Alcohol

-(?)

+Browni
+,

+
+
+
+

9+

rissoler
sear
parch

+
+

toast

+
-

garil
barbecue

Lid
+Frying pan

+
+
+

Figure4
D

+
+

|+

bake2

+ To soften

-+

parboil
steam
reduce
fry,

4)

deep-fry

II

broil

~~s

boil2.
simmer
stew

sautr
pan-fry
French-fry

+
+
+

boil,

4)

cook3

49

JOURNAL OF LINGUISTICS

but not * I fry. Sentences like She can cook or Henry bakesimply the general
senses of the terms. The semantic generality of these words is parallelledby
greatersyntactic freedom.
Most of the other verbs in the cooking field (and the marked sense of cook
and bake)seem to be basicallytransitiveverbs, with the intransitiveform derived
from them.
X saute'sonions(adverbof manner)-+Onionssaute'(adverb).
An adjunct is necessaryfor an acceptablesentence.
Onionssaute'quickly,but not *Onionssaute'.
Simmerseems to be an exception, however. One would say J1ohnlets the water
simmer,not *7ohnsimmersthe water.
Boil is an odd memberof the cookingset, and it patternssomewhatdifferently
frtm the other verbs. It is a more generalterm than other words in the field in
that it patterns in several other fields as well. Perhaps it is not primarily a
cooking word at all, as arefry, broil,toast, etc., but has been 'borrowed'into it.
Boil and some of its hyponyms (reduce,simmer)collocate with liquids whereas
the other lexemes collocate only with solids or solidifiable foods. Moreover,
boilis the only lexeme of the set which takes the adjectivalsuffix -able. *Fryable,
*simmerable,*broilabledo not occur as words in English, while boilabledoes,
but interestingly enough - not in its applicationto cooking. A typical kind of
boilableitem is waterproofbaby pants, and the meaningin this context is 'canbe
boiled without becomingruined'.
In dealing with derivationalitems, many linguists would include productive
processes in the grammar. 'Grammaticalrules are intended to generate all
grammaticalutterances, whether they happen commonly to be used or not'
(Lees, I960: I52). Although there is considerablemerit in Lees's proposal, I
believe that it is also relevantfor the linguist to show in his descriptionwhich
forms in fact occur and which do not, at least in studies of the lexicon. 'Accidental'gaps may now and then revealnew systematicinsights.To say thatfryable
does not occur because words in the culinary lexical field do not take -able
is, to be sure, not a very profoundexplanation,but it is a step above no explanation at all. And by studying other lexical fields or by looking at the nonlinguistic
situation, perhapsa more satisfactoryexplanationwill be found.
Nominals formed from past or presentparticiplesare productivefor all verbs,
as in steamedrice, barbecuedribs, stewing beef, broilingrack, though planking
steak is marginal.
Among the other nominalizations,the -er suffix is the most common one in
the lexical field. The Shorter OxfordEnglishDictionaryconsistently defines -er
terms in this semanticfield as 'one who or that which- '. This is quite accurate
in looking at the process from a productive point of view. However, of the
lexemes actuallyin use with -er, all except one are inanimate.(Broiler,fryer, and
50

CUISINE

SEMANTIC

roaster,referringto chickens, are neither agentive nor instrumental:see below.)

A toasteris a THING for toasting; a roasteris a THING for roasting. The animate
sense is readilyavailable,and we can easily imaginea situationin which it might
be used. For example, a chef might delegatejobs to his assistants,saying X will
and Z the egg-poacher.
be the bread-toaster,Y the mushroom-broiler,
The only -er animate derivativein common use in the culinaryfield is baker,
and it is relatedto the generalsense of bake,referringto the preparationof bread,
pastry, cakes, etc. The correspondinglexemes for the general sense of cook
are cook6 and chef, with cook having a wider range of application than chef
and with some differencein professionalstatus. The occurrenceof these currently used, animate agent nominalizations reinforces the semantic analysis
establishinga generalsense of bakeand cook.
The inanimatelexemes taking -er include broiler,roaster,toasterand cooker,
although the last term is primarilyBritish. (Stove would be used in America.)
A number of -er lexemes are compounds, for example, pressure-cooker,eggpoacher, French-fryerand double-boiler.The rules for generating compounds
are not the same in each case. For example,we can generate
X poaches eggs-*poacherof eggs-+egg-poacher(Lees, I960:

I52)

but hardly
X boils double-+*boilerof double-+double-boiler.
A few lexemes with -er are neither agentive nor instrumental,for example,
fryer and boiler, referring to chickens. These words have a different deep
structure from the agentive or instrumental -er lexemes (McIntosh, personal
communication).This sense of fryer is 'a chickento be fried'.
The derivation of nouns from verbs (and vice versa) without any formal
change is a productivederivationalprocess in English, occurringin the culinary
field as well. A noun so derived from a cooking verb is used either to refer to a
utensil (grill, barbecue),a food (roast, toast), or an event (barbecue,bake).
Figure 5 summarizes some of these nominalizations.Others can be added,
of course. The diagram(on p. 52) also shows that the derived terms fall into a
smallnumberof classes
Not only can this chartbe extendedwith new terms, but new derivationscan
[6] One might suggest that there is a low-level rule (2), operatingupon the output of the
moregeneral(i), as follows:
(I) cook-er-+ cooker
(2) cooker[+Animate] -. cook
Childrenhave been observed to produce such utterancesas Mummyis a good cooker,
presumablyby transformationfrom the structureunderlyingMummycookswell. Their
later 'internalization' of rule (2) is comparable to their 'internalization' of such low-level

rulesasfly + ed -+ flew, etc.

5'

JOURNAL OF LINGUISTICS

be generated.It seems likely, however,that new nominalizationswould fall into

one of these categories.
One furthernominalizingsuffix occurs in this culinaryfield, -ery in the words
cookeryand bakery.But the suffix has different meanings, at least in the commonest senses of the words.
Cook:cookery#bake: bakery
Cookeryis the art of cooking (cf. flattery), whereas bakeryis a place for baking
(cf. scullery).The ShorterOxfordEnglishDictionarydefines cookeryin addition
as 'a place for cooking', but the ordinaryterm is kitchen.Lexemes like *fryery,
*broileryand *grillerydo not occur. This fact furtherreinforcesthe establishment
of generalsense of cookand bake.

-er
cook

Food

Appliance or utensil

Process

Different .form
stove

cooker
(pressure-)

boil
poach
stew
deep-fry
French-fry
fry
.broil
grill

Event

(double)
boiler
(egg)
poacher

cooker
(an apple)

pot, pan

stew
deep-fryer
French-fryer
frying pan
skillet
broiler
grill

barbecue
bake

fryer
broiler
(mixed)
grill

barbecue

barbecue
oven

roast

roaster

toast

toaster

(fish) fry

baker
(an apple)
roaster
roast
toast

(clam) bake
(wiener)
roast

Figure 5

The compound verb lexemes (pan-broil,pot-roast, oven-poach)employ a

productive device, that is, productivewithin the semantic field. A new lexeme
is formed by combining a noun referringto a utensil (vessel, appliance)with a
verb referring to a process. In addition to those lexemes already listed, one
52

SEMANTIC CUISINE

might generategriddle-bake,charcoal-poach,
oven-fryor skillet-boil.The following
rule sketchesthis derivation:
NP1+V+(NP2)+prep.+D+NP3-+NP3-V
where NP3 is a utensil and V is a cooking process.
For example,
X bakes(pancakes)on a griddle->griddle-bake.
There does not seem to be any certain way to predict which components of
the two processes are to be selected for the meaning of the new word. Oven-fry
in fact refersto cookingwith fat in a very hot oven, but it could referto enclosing
a frying pan in such a way as to make some kind of oven on top of the stove.
One cannot form new verbs with two utensils (vessels, appliances) such as
*to pot-stove or with two processes *to boij-roast,unless something sequential
is implied, that is, 'to boil and then to roast'.
A large number of cooking lexemes have additionalmeanings, that is, they
pattern in other lexical fields as well as the cooking field. Although there is not
space to present a complete analysisof these other senses, one interestingresult
of the analysis emerges. Some of these other uses fall into a small number of
classes, forming subpatternswithin the major pattern.For example, a number
of cooking words are used in the field of emotionalstates, particularlyreferring
to statesof angeror agitation:boil,simmer,steam,stewand burn.Certainsemantic
relationshipshold in the emotionalfield as in the culinaryone. For example, a
person who is boilingis more agitated than one who is simmering.(Terms like
heat, warm,cool, chill,freeze, thaw, etc., patternin the field of food preparation
in a semi-technical way and in the emotional field as well, so that the interrelationshipof terms in fields is even greaterthan I have shown here.)
Another subfield is that of torture, with fry, grill and burn.The explanation
for these senses can be sought in the gloomy pages of humanhistory.
A number of terms are used to refer to states of physical discomfort caused
by heat, as I'm roasting,I'm cookingin here or It's roastingin this room, It's
steamingin here. Even in these extensions certain cooking components are
retained. Roasting conveys the notion of dry heat, while steamingsuggests
dampnessand humidity.
These subpatternsemerge most clearly in the study of semantic fields. Such
investigationsmay help us understandbetter the dynamicsof the metaphorical
extension of words and will perhaps bring systematicobservationsto the traditional studies of figuresof speech (cf. Bierwisch(I967), who suggests a similar
point in extending certain visual adjectivesin German to auditory uses). In a
given semanticfield, e.g. cooking, if one or more lexemes have senses in another
field, e.g. emotional states, the other words in the first field are availablefor
53

JOURNAL OF LINGUISTICS

application to the second field, and they carry with them all or any relevant
components.
A similarprocessis at workin derivationalextensionsin a field. Since barbecue,
clambakeand wiener-roastexist as social events, one can easily and meaningfully
extendthe restof the culinaryvocabularyto uses like we'regoingto a boiltomorrow,
or to a potato boil. New lexical formations,like to casserolecelery,as suggested
above, are modelled on other items in the same field. The same sort of dynamics
operates with collocations and semantic restrictions. Cook normally collocates
with words subsumed underfood, for example, chicken,and many food terms
belong to other lexical fields as well, in this case bird. It is a short extension for
any bird to be classifiedasfood, whetherpeople normallycook it and eat it or not.
The study of the meaningof words becomes much more orderlywhen placed
in a lexicalfield, especiallya field which dealswith a specificuniverseof discourse.
A lexical field and a universeof discourseare abstractions,too, but they provide
a useful, and I believe valid, intermediatelevel between the meaning(s) of an
item in isolation and its meaning in a specific utterance in a specific speech
situation.
An analysis of the field of cooking words reveals patterns which should be
exploredin other fields as well. In analysingthe verbs and their derivations,the
syntactic and semantic analyses tend to support each other. Cook and bake,
which are semanticallymore general than the other lexemes, also have greater
syntacticfreedom. They are the only verbs in the field which permit the object
to be deleted when the subject is agentive, the only ones with commonly
used animatenominalizations,and the only wordswhich takethe -ery suffix.
The study of lexemes derived from the cooking verbs shows that these derivations fall into a small number of sets. That is, nouns derived from cooking
verbs can be classified as appliances(utensils, vessels), foods, or events. Since
there are a number of empty slots in the matrix,one might predict the meaning
that a speakerwould attach to the new item. The same word in another field
might be given an entirely differentmeaning. By studying phenomena of this
sort within a given universeof discourse,I would expect the responsesof speakers
to be more uniformthan responsesto words in isolation. In short, this approach
shows better the productive aspects of the lexicon - not just what words mean
now, and not just how old words combine to form new sentences, but how old
words can extend their meanings.This is a synchronicprocess, not a diachronic
one.
The study of words in fields may lead to revealing generalizationsabout
semi-productive processes. For example, no cooking lexeme takes the -able
suffix (except for boil, but not in that field). Investigationsof other fields are
necessaryto verify this prediction.
Finally, in looking at the other meanings of the words, for example, nonculinary senses of words in the cooking field, one discovers minor patterns
54

SEMANTIC

CUISINE

operatingwithin a majorone. However,these minor patternsare best discovered

by studying the major field. This phenomenon offers other insights into the
productiveprocesses of the lexicon. A number of cooking words are potentially
availablefor use in the fields of emotion and torture since a few words already
patternin these two fields.
REFERENCES
Allen, I. B. (I960). Gastronomique.London: W. H. Allen.
Bendix, E. (I966). Componential Analysis of General Vocabulary: The Semantic Structure
of a Set of Verbs in English, Hindi, and Japanese. Bloomington: Research Center in
Anthropology, Folklore & Linguistics. (= IJAL 32, Part 2).
Betty Crocker'sPicture Cook Book (1956). New York: McGraw-Hill.
Bierwisch, M. (I967). Some semantic universals of German adjectivals. Foundations of
Language, 3. 1-36.
Chomsky, N. (1957). Syntactic Structures. The Hague: Mouton.
Chomsky, N. (I965). Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press.
Claiborne, C. (I96I). New York Times Cook Book. London: Paul Hamlyn.
Croft-Cooke, R. (I960). English Cooking. London: W. H. Allen.
Gleason, H. A. (I962). The relation of lexicon and grammar. In Householder, F. W. &
Saporta, S. (eds.) Problems in Lexicography, pp. 85-I02.
Bloomington: Research Center
in Anthropology, Folklore & Linguistics.
Good Housekeeping Cooking Encyclopedia (I964). London: Ebury Press.
Katz, J. J. (I966). The Philosophy of Language. New York: Harper & Row.
Katz, J. J. (I967). Recent issues in semantic theory. FL 3. I24-174.
Katz, J. & Fodor, J. (I963). The structure of a semantic theory, Language 39. 170-2I0.
Katz, J. & Postal, P. (I964). An Integrated Theory of Linguistic Description. Cambridge,
Mass.: MIT Press.
The Grammar of English Nominalizations. Bloomington: Research
Lees, R. B. (I960).
Center in Anthropology, Folklore & Linguistics.
Lehrer, Adrienne (I967). Fields and Categories: A Study in Structural Semantics. (Unpublished Ph.D. dissertation, University of Rochester.)
Lyons, J. (I963). Structural Semantics. Oxford: Blackwell.
Lyons, J. (I968). Introduction to Theoretical Linguistics. Cambridge: Cambridge University
Press.
McDouall, R. (I955). Pocket Guide to Good Cooking. London, Collins.
Porzig, W. (I934). Wesenhafte Bedeutungsbeziehungen. PBB 58. 70-97.
Porzig, W. (I950). Das Wunder der Sprache. Bern.

55